How to Understand Smart Shopping

Isaac and Patrick sat down with Kirk Williams, Owner of ZATO PPC, and Mike Ryan, Product Management Lead at Smarter Ecommerce, to discuss the benefits and drawbacks of Smart Shopping and manual vs automated bidding, data rights and past, present and future concerns with advertising.

Episode Transcript: 

Isaac Rudansky: Welcome, everybody to the How to Hide a Dead Body podcast. We have some very exciting guests with us today. It is going to be an episode packed with insights and perspectives — a future look into the world of digital advertising and how agency owners,  freelancers, and digital marketing managers can leverage some of these technologies and these strategies for their own campaigns, for their own businesses.

I'm Isaac Rudansky; founder of AdVenture Media. I'm joined here, as I usually am, by Patrick Gilbert, AdVenture Media's Executive Director. I like to tell clients that the buck stops with him, and it always does. 

Patrick is recently the author of Join or Die: Digital Advertising in the Age of Automation. It's a digital advertising masterclass, packed with under the hood technologies, insights into Google's automation, Facebook's automation. But it's also a very heartfelt story of how Patrick helped our agency grow over the last five years. So you have the technical side of things and you have the inspirational side of things. It's forwarded by myself and it's also forwarded by David Sable, the former Global CEO of Young & Rubicam. 

It's available on Amazon in hardcover, paperback and soon-to-be audiobook. Stock up, get a few copies for yourself, for your loved ones; it makes a great gift.

Speaking about gifts, it is a great gift to have Kirk Williams with us today on the call. Kirk Williams is the founder and CEO of ZATO PPC Advertising; a boutique digital advertising agency that works with some very large name clients. And if you look up their case studies, their reputation in the field, you'll see how impressive Kirk's work has been. It's a real honor to have him on the podcast with us today. 

Kirk also has a book out that's available on Amazon in paperback, Kindle, and audiobook. It's called Ponderings of a PPC Professional. Check it out. It talks a lot about theory and interesting insights into the world of digital advertising, into the world of Google. And reading through the book, my eyes were open to a bunch of different interesting perspectives that although I've been embedded in this industry for a lot of years, it was eye-opening for me to see an intelligent, different attitude about things. And certainly, helpful in a practical sense as well. 

We also have Mike Ryan, who is a product manager at Smec — a 150-person firm that works on digital advertising, strategy services, as well as software to help large advertisers manage their campaigns. He's also the product owner of a tool called Orbiter, which helps with data visualization. And Mike brings a very interesting data-centric perspective to today's conversation.

So we're going to cover a lot of interesting things today. But let me first kick it over to Kirk and then to Mike, just to introduce yourselves briefly, and to maybe outline what it is that brought you here today and what we're going to try to accomplish.

Kirk Williams: Yeah, I can start. I think a lot of this started with a conversation between myself, Patrick, Mike and this was about a year ago as we began discussing Smart Shopping Campaigns. I won't do too much of the discussion of exactly what Smart Shopping is and all that, but I'll at least look real briefly if you know you don't know Google Shopping at all if that's totally new to you. 

Basically, Google Shopping is what's called a comparison shopping engine; a CSE. It's a way that Google — but others have done as well — there used to be back in the day, Nexttag, Pricegrabber, and all that, and the idea was visually comparing key elements of a product. Especially for Google, that was a big deal because now it's not just simply text on a search engine result page. It's a shopping ad, too. Now there's image, price, things like that. So you can compare that. 

So that's been Google Shopping. The name has changed throughout the years. It's been a few different things. And Google, I think it was about the summer of 2018 if I remember, they introduced a new type called Smart Shopping Campaigns, which is highly automated. So there's been a lot of discussions since then of what Smart Shopping is and the three of us have engaged on some different things over the last year or so And finally getting the others to chat about it. So I'm super excited about it. 

Isaac Rudansky: And Kirk, you actually just wrote a really interesting 9,000-word shopping guide on Shopify's website. Tell our listeners how they could find that.

Kirk Williams: Yeah, the Shopify retail blog. Basically, my objective was to put out something especially for retailers that are maybe first going from being brick and mortar to just getting a few products online, or even startups, maybe D2C that don't really have the ability to hire someone in house or agency yet, and they just need a way to get their Shopify Store set on Google Shopping? Like, what all does that entail?

I wanted to have an actual walkthrough that someone could if they follow it, it really could actually go from not doing Google Shopping to going live. So, it's on the Shopify retail blog. I have no idea what the URL is, but probably if you Google it, you can find it.

Isaac Rudansky: I love those guides, similar to the one that you've written that is a step-by-step, well-organized, table of contents; it's substantial. And there's not a lot of substantial content on the internet. So well done with that. 

So we have shopping, we know that Google Shopping Campaigns is the advertising format of choice if you're an e-commerce business. You could set up your data feeds, you could set your prices, you send traffic, you buy traffic to your website. Google introduced Smart Shopping. Patrick, what is smart shopping? And how does smart shopping differ from what people are used to as far as traditional shopping campaigns go and what exactly is smart about it?

Patrick Gilbert: When we think of the shopping campaigns that Kirk was describing, I think that we now refer to that as standard shopping. And standard shopping was in the formats that Kirk described and would serve ad placements on the Google Search Network. So when you perform a search, you're searching for an LED TV and you have those product images at the top, you could also toggle over to shoppinggoogle.com and see a larger list of them. And for a long time, that was really where your shopping ads were displayed. 

Now, in 2018, when Google launched the Smart Shopping feature, it was a different perspective on this where a lot of people have viewed it as a fully automated solution, which it sort of is, but it's much more than that. The main differentiating factor between Smart Shopping and standard shopping is that you now unlock all of these placements that were not accessible with your product feed, historically. 

So you unlock display placements, YouTube placements, and Gmail placements. And again, these were always available. You can run a Gmail campaign, you can run a display campaign on the Google Display Network. What makes this unique is that it's all under one hood. So when you set a budget and you set your target return, ad spend goal, or whatever your goal is, Google can make these decisions about where those placements are going to be served in real-time. 

And what that does is it unlocks this concept of liquidity, which is a big topic in the Join or Die book, which basically encompasses an environment where very few restrictions, meaning budgetary restrictions would allow a machine-learning algorithm to make decisions in real-time about how to serve your Google Shopping ads. And so, the TV example, if you have a person that's shopping, they're in the market for a TV, they're searching on google.com for an LED TV, they're clearly in the market for it. But that single person and many others like them that are also in the market, they're not always looking at that search results page. They're also in front of display ads and YouTube ads, and Gmail ads. So this allows you to meet those people in all these different areas. 

And the real benefit of an automated solution like Smart Shopping is that machine learning, when it works well, can not only expand your reach in an absolutely exponential way but also look for marginal efficiencies to be able to say, "Okay, well, this display click or placement would be pennies on the dollar. And we know that this person is in the market for this so that it's an ad well served."

So you, basically, lean on all these different signals to be able to optimize your budget. Now, in theory, that's an incredible product: an automated solution that expands your reach and is able to more effectively sell your products. It has been met with a lot of pushback and criticism from our industry. And I think rightfully so because it's a very robust and complex platform, the solution, Smart Shopping. And I think it can only really be described as unfathomable because it is so complex. 

And in bringing that to market, Google made the decision to say they're just not going to report on a lot of the data that a lot of advertisers had been used to for many years running their Smart Shopping campaigns. Notably, you don't get access to your search term data, which is something that fueled a lot of optimization and strategy over the last, say, 10 years. We no longer get to see what users are searching for that is triggering our shopping ads. That's a big breaking point for a lot of people, which is one of the few things that led us to this conversation where Kirk wrote an article, I wrote an article, Mike wrote an article, We kind of went back and forth and this was over a year ago. 

So Mike, can you bring us up to speed on what that whole conversation looked like?

Mike Ryan: Just like you said, this is like full-funnel coverage from a single campaign, which sounds great, but there are different concerns that come up about this. Kirk was already well into a multi-part series on Smart Shopping Campaigns. And then I think this catalyzing article that he wrote there was about the loss of data that people are used to from standard campaigns. And when they're using smart bidding in the past or manual bidding third-party tools, however, they would have access to this kind of stuff. And for Kirk, this was concerning.

I think his argument is that, fundamentally, advertisers have a right to that information, and, Kirk, we can talk later if your views have changed or not, and also that this kind of data is valuable to advertisers, to their agency partners, and even often to Google itself because it gives us a chance to kind of put up smart guardrails to help the tool and guide it along. And this kind of data can be valuable also in other marketing channels and so on. Patrick came back and said that, from his perspective, we're not really owed this data, and that it's kind of a wrong assumption for many people that will be able to benefit from this data better than the way Google might use it. And there's a narrative that often comes out of Google, and Patrick, I don't want to misrepresent you, but I think you agree with this, that if we would see this information and not be able to act on it in a smart way, then it's just kind of agitating. So it's not really that big of a deal. 

Patrick challenged, I think, the way that traditionally, funnels are defined and pushing for a more progressive kind of marketing framework. And that Google also has so many more data points beyond just a search query, for example, which might look bad to us or too generic or whatever, but they see it in the context of all these other audiences or user data points. 

So, I jumped into the conversation as well and I bring a perspective here, basically, trying to understand the business logic about why Google is pursuing this. I would argue that this product could have been built differently and could have been built better. It is what it is. This product has a relentless focus on revenue and ROAS targets, which is fine and of itself. If we think in terms of Maslow's hierarchy of needs, the users of this tool are performance marketers, and raw performance is fundamental to them. 

But then, the next question for me is, okay, when your performance has been fulfilled, what can you do to advance the majority of your strategies and operate in a more differentiated way? And I think that the way Smart Shopping Campaigns are set up, it's not as possible to really differentiate yourself or pursue a particularly sophisticated marketing strategy. 

Patrick Gilbert: Mike, can you explain how the previous setup of standard shopping could have helped with that in the way that you're expanding upon a smaller subset, you're growing a business beyond bottom-funnel search, or whatever you're looking at? How did standard shopping meet that need that Smart Shopping now doesn't give you the ability?

Mike Ryan: If we look at a topic like a search term report that's missing or that data that's missing, Google Shopping is not based on keywords the way that text ads are. But there are these very clever hacks or workarounds to that are, I think, pioneered by Martin Roettgerding. And this allows you to funnel traffic. That way, let's say that you would be bidding on a product, but you'd always be bidding on it the same for all different kinds of queries. 

So there's an argument to be made that people can optimize performance this way. There's also an argument to be made that you could pursue other strategic goals. For example, a client comes to mind where they use a query sculpting approach. They have their own proprietary brands, and then they have vendor brands. And it's really important to them to treat these things differently because they want to pursue market share with their own brands, and they want to pursue other kinds of goals with the vendor brands. And they use a quarter sculpting approach to achieve that kind of an effect. There are other ways of doing it. 

Another example that comes to mind, and I know, Patrick, you wrote in that post that you really favor account consolidation and I think there are arguments to be made for that. I tend to like a more granular account structure and these possibilities are pretty limited in Smart Shopping Campaigns. The fundamental setup would be a single campaign, single ad group, only one priority level, and so on. So it limits the flexibility that you might have compared to standard shopping setups. 

Patrick Gilbert: And I'll say on that, the query level bidding approach that you're mentioning was one of the greatest workarounds that the PPC community had come up with. And I'll give you the best example that I have experienced. We used to work with a company that sold books, and they had four million-plus books in their library, and books cover everything. 

And for the search shopping campaign, it was the wild west where, all of a sudden, the Royal Wedding would be happening and people would be searching how to stream the royal wedding. And it would be triggering ads for things like the history of the royal family. It's nuts because there was no way to proactively see that happening. So this query-level bidding strategy basically said, "Hey, if somebody uses the word book, or if somebody uses an ISDN number, we're going to filter them to a different campaign. We're bidding more aggressively." And honestly, for like, probably three to five years, that strategy was run in every single e-commerce account that we ran. 

So when Smart Shopping came into the fold, and now you don't have the ability to do that, and then eventually, Google removed the ability with standard shopping campaigns, they removed the settings that allowed you to actually get that workaround, we panicked. And I'm sure everyone else did as well because it completely turned a lot of our accounts on the test.

Stressful as that may be, I still believe that and I've said this many times, our early adoption of Smart Shopping, despite all of these challenges and fears and threats, was, by and large, the most profitable decision that we have made on behalf of any of our clients since we've been in business. And I'm going to sound like I'm pushing propaganda from Google, but it's the idea that, look, if it works, who cares? Who cares if we're not getting the data? I don't want to go too far into this. But, Kirk, do you have anything to add as far as this background conversation?

Kirk Williams:  Yeah, I have so many thoughts that I'm not even sure how to start thinking about and voicing some of those. Some of this is my own journey, we've changed a lot in my agency, as I'm sure a lot of PPCers right now. Sometimes you look at things like automation in a changing environment. Sometimes you say, "Hey, I don't really like that, But also, this is the way it's going. So we need to evolve. We need to join or die." Sometimes you look at it and say, "Wow, there's a lot that makes sense here. This is good. I'm all in. This is why we need to change, right?"  So probably, we, like many others are adapting and evolving with Google, sometimes out of requirement, and sometimes because that's also what we're buying into. 

So I find myself with that mixed viewpoint, still, of Smart Shopping of seeing things that still concern me, but of having a dramatically different look at it than I did a year ago. I just feel like I have a ton of questions that I'm trying to answer. I'm trying to get answers from Google, maybe from you guys, things like that, too. Some of that is I look at this, and I say, Smart Shopping, to me, I have questions about the commoditization thing. Mike brought that up in his article. 

So, some of that is, is all of the early-adopter success we're seeing just simply because the markets are not fully commoditized with it? And if that begins to happen, and all of a sudden, it's fully Google and the machine-run and everything, are there going to be issues there rather than just simply the first mover? Sometimes questions like that, which I just don't know. Sometimes it's hard to even know that until you're 10 years past something. 

Definitely, we have seen success with Smart Shopping Campaigns. I'm putting the finishing touches on a Udemy course for Smart Shopping Campaigns. We like it. We see success. So I want to make sure if I come across any concerns and hesitations, that that that's not coming from someone who is looking at that saying, "This is trash; we'd never use it." I have friends like that. 

That being said, probably the main concerns that I would still have with Smart Shopping. I would probably be able to be one of Google's most vocal proponents for Smart Shopping if they would do two simple things. If they would begin showing data regardless of whether or not we can do anything with it because of some of our things that we've discussed in other ways riding on that. You do have some level of utilizing that for broader business strategies, things like that, too. 

So different people use different data. The data that we have been accustomed to and even built accounts and built strategies and that around, show that again. And then also, I do think the audience thing is still a big concern for me, although even that's incredibly complex right now, with all of the things that are changing with privacy laws, as well as cookies being wiped out off the face of the planet. All that stuff is impacting that, too. So, by audience, I mean things like new versus returning at a very simplistic level. 

Some of that is as simple as saying if a machine requires very specific data inputs, and then we say this is our target, and the machine says, "Okay, great, we'll aim at that." And those revolve specifically around what we are giving to that algorithm. So in the case of smart shopping, it's solely maximized conversions or maximized conversions with ads. And that does not always align with some of our clients' more complex, broader goals. 

So that typically is a place where we'll still utilize both standard and smart, and this especially works its way into brand versus retail retailers a lot of times. We might have a brand that comes in for shopping ads and maybe they're newer in a specific space, maybe they're introducing a product in a specific space, maybe they want to control that space. We don't have the history, we don't have the account history, we don't have the conversion history, we're not even necessarily expecting a certain return at that point because they want to get in and just own the coffee grinder space if you will. 

So there are certain things like that where I would, "Hey, if there was a way to think through that, to be able to divide those up to say, here's more of our direct performance-based Smart Shopping objectives, maybe this is our separate objective, maybe there are different ROAS targets we can give to different types of campaigns that target different types of audiences, things like that. I think that that would give us a little bit more of that complexity. I think it reveals itself a little bit more in the broader business world, as opposed to simply just what we have in just the Smart Shopping type as it is. 

And the final thing I'll say, and then I'll shut and let someone else talk would be, as we've been testing Smart Shopping, and this is props to you, Patrick because a lot of this is as we've been chewing over your book as a team, too, some of this has helped us even become more aggressive in thinking through some of this stuff. It is interesting, we have begun to discover what you all have. I mean, you're ahead of us here in this regard. When you do start to say, "Okay, let's start utilizing something automated like Smart Shopping," with those caveats that I noted, then as you start to utilize it, you do start to determine, "Hey, there are different things we can do with this that we never even had thought of before we actually started getting into the system and working with it."

 And so, there are different strategies and things that we try tips and tricks with Smart Shopping that I never would have thought of before we got into it, that are still more complex than just simply a single campaign. So we are also doing things to try to figure out and to guide the machine. And I think there are ways that you can do that even with Smart Shopping. So, we are utilizing more than we used to. I see things differently than I did a year ago. I still have some of those same concerns.

Patrick Gilbert: I would add to that. I think that's a great way to cover a lot of the different questions and issues people have. There's the commoditization feature of is there a first-mover advantage? Honestly, for a long time, I felt that that was a large driver of the success we were seeing. I've since changed my perspective on that. And then there are all these other things about rights to data, and does it actually work as well as it says? We could be doing a better job. 

On the note, though, that there's really only two levers or really only one lever that we can pull, which is what bid strategy we're optimizing for when a client might have another goal or other goals. I think that's interesting because, at face value, you're completely right. But I do think if you step back and say, "Well, is there a way for us to use what we do have access to still achieve that goal?

And an example that I give in the book that I talk about all the time; we have a client, they have a large product category, and different profit margins for every product. So it's really impossible to say to the algorithm, "Hey, listen, we have a target return on ad spend of 500%" because that's not really what it is. We want to earn a profit; return on ad spend is not equal profit, particularly if some of our products are white-labeled or private label, and therefore, we have a larger profit margin, and some of them are national brands, and we have a tighter profit margin. 

So what we'll do in that case is we'll work with our development team to use the data layer variable to code in the profit margin. And we'll pull that number through as our conversion value variable and optimize for that. So we're actually tricking the system to understand our actual profit goals. To your example about we want to own the coffee grinder space, that's a little bit more challenging, but I think it could probably be resolved with a bit of creativity, like, "Hey, maybe we can fire conversion actions for engagement on the coffee grinder page on our site." I think that there are ways to get to that. 

And that's something that we're trying to do more with our clients is figuring out all these different things that we can optimize for. And it gets complicated because then you have to get developers involved and you have to have the right people in the room that can have this conversation. But I do think it is possible. 

Kirk Williams: So you're saying basically create micro-conversions around specific things that would help you communicate to Google, "This is what we're aiming at," and then help guide the machine in that way?

Patrick Gilbert: Yes. So I think what you had said before, and I'm paraphrasing, but it was something to the extent of Google doesn't understand what our real goals are. They're binarily really focused on here's our budget, and here's our return on ad spend goal. And there are usually more layers to that. It's like we have excess inventory in this area, and we need to push this product, and we want to own this space. 

And I'm saying that there's probably a way, and it's not as clear as it used to be with the tools that we had access to three years ago. But I do think that there's a way, if you get creative with how you're counting conversion actions in your account, and what you're choosing to optimize for, that you can achieve that same level of success.

Mike Ryan: I agree. I think back to this idea of technological commoditization and strategic commoditization, it's not a dead-end with Smart Shopping Campaigns. There are three main areas where you can be active here: and that's your feed, especially custom labels, your conversion tracking, or offline conversion imports, and then it's your audiences. And the more sophisticated that you can be with all three of these areas, and they can be complementary, the better results that you're going to get for sure. 

We have a client, for example, talking about owning the coffee grinder, however, we can build segments based on their impression share. And if they're not satisfied with a certain impression share, they send it over to Smart Shopping instead of standard shopping. And you can segment that further and pursue it more aggressively using ROAS as a lever. 

My problem with ROAS is that I always think of the law of the instrument, which is when the only tool that you have is a hammer, then you start seeing a lot of nails. And not everything is a ROAS problem. And this is this bias that we have. And by the way, Google added another goal type here: new customer acquisition, but I have something else to say about that. But you can be more sophisticated with Smart Shopping Campaigns. Those are your same three levers: feed, conversion tracking, and your audiences, those are the same three main levers in standard shopping too.

Kirk Williams: So let me ask you this, in my example, the coffee grinder space thing, are you actually taking those micro conversions and translating them into some level of determined revenue goal?  Let's say that you know that people aren't necessarily going to buy. You want to make sure that you're owning the search term space for a coffee grinder, that sort of thing. Are you building out some landing page, "Well, let's get interested by maybe an email, things like that."? And then as that gold fires, are you determining, as you've done your math and figured out your research with your client, "Okay, let's call that up a $5 goal value for that email."? 

And then is that part of how you're feeding that back in because, at some point, that's still, to me, the system that is based as Mike was saying, with the hammer as being the ROAS model. At some point, it's got to be translated into that. Is that how you guys do it? Do you work on conversion value and those things? 

Patrick Gilbert: Yes. And so I will say this coffee grinder example we want to own the space is a little bit more complicated. And I don't think that any of our accounts or optimizing Smart Shopping for an engagement-based metric like that, although I think conceivably it could be done. Your search impression example, I have no idea if this is actually possible, but conceivably, you could pull that search impression share metric into Google Doc. 

If you create a report and then create a script, that would then fire up a micro-conversion based on what your search impression share was. And every time your search impression share for a day was above 90%, you earned a conversion from that. Maybe that probably wouldn't work. Okay, sorry. So let's put that aside for a second. 

This is more in line with what we do for a lot of our lead gen clients, though. And I think that's part of the advantage. We don't necessarily specialize in e-commerce or retail. We have a broader client base. And I think that's helped shape our perspective on a lot of these things.

The number one thing that we do with lead gen clients is we won't want to spend any advertising dollars at all until we have proper feedback mechanisms in place. So whatever your CRM is, let's take Salesforce example, which has a very easy plug-in. Gone are the days where we're going to optimize for quantity of leads, quantity of phone calls, or form submissions. At the end of the day, anyone can get your form submissions, and a lot of them are going to be bots and they're terrible. And that's where it becomes very dangerous when you're optimizing for that, especially with automated bidding and saying, "Hey, go crazy, give me as many conversions as possible," and then none of those leads convert into deals. 

So what we've done, and this has really been our focus over the last two years, has been connecting those additional conversion actions of a salesperson interacting with Salesforce and saying this click led to a conversion, which then led to a marketing qualified lead. And of those 100 marketing qualified leads, 50 of them were sales qualified leads, and of those 50 sales qualified leads, 10 of them became closed customers. And we pull all that data back into the system. And eventually, what you want to do is slowly but surely pull away the less valuable conversion actions. 

So this is kind of the holy grail of lead gen advertising right now. At some point, as quickly as possible, I want to become confident where I'm saying, like, I don't want to tell Google to count a conversion when we get a lead. I only want them to count a conversion when it's a marketing qualified lead, or so on and so forth down the funnel. 

And if you take that sort of methodology and apply it to e-commerce, if you're trying to own the coffee grinder space, you got it. You just need the systems in place to be able to track those conversions. And then some sort of feedback loop to know that, ultimately, that's resulting in revenue down the line.

Mike Ryan: Yeah, and I think in e-commerce, it tends to go in the direction of customer lifetime value modeling, and it's a super complex topic in the end because it's like tracking margin, but it's aggregated margin over the lifespan of a customer. And you best case want to be doing it in somehow a more predictive or forecasting way to try to understand what a customer will be worth it could be worth or is worth so far. There's a lot of approaches here

Patrick Gilbert: Mike, you mentioned the new customer acquisition feature within Smart Shopping. That works on the basis of saying, "What is a new customer worth to you?" And you artificially inflate what the actual revenue value is to train the system? That's what it comes down to. Calling it smart is hilarious because it's not actually smart. And I'm not even just saying that in a nefarious way. Like, these algorithms don't ever think in the way that humans do. They're just given instructions, and they try to meet that. 

So what they're really being told, they're being tricked to say, "Well, this is more valuable so I'm going to go off and try and find more things like that." And if you can train it, and trick it to go do that, then over time, it gets smarter, conceivably, to go and replicate that at scale. And you could, theoretically, do that with a lot of different things.

Mike Ryan: Totally, I mean, algorithms, they don't have common sense. It's not general intelligence. It's this kind of very focused, savant intelligence on a given task. Yeah, it's definitely important to remember that.

Patrick Gilbert: Definitely. So everything we've talked about so far on this topic is really about how to drive success with Smart Shopping and all the different things. And I'm really excited about this because one of the reasons I've been concerned about topics like this has been because we have not had that query-level bidding breakthrough that the industry has been really excited about and we've all been working together to figure out like, "How are we really going to drive the success?"

I think this next big thing, the query-level bidding phenomenon that is going to take this entire industry forward is going to come out of conversations like these. Maybe not this one directly, but we need to have more of these. But I want to put this on hold for a second as important as it may be and talk more about the threats of Smart Shopping and why there are criticisms and concerns and continue that part of the conversation.

In the articles that we have passed back and forth, we had this concept of a black box system and there are all these threats that come with it. To me, there are two different parts of the black box system. There's the point that we covered so far, where it's, "Okay, I need this data to be able to be successful." Or, "I deserve this data because it'll help me be successful and other channels." And that's one category. 

And the second is if it is a black box system, who's to stop Google from taking advantage of advertisers. And I actually want to start there, or broadly speaking about the BlackBox. Kirk, how has your perspective changed, if at all, on the threat of the black box, so to speak?

Kirk Williams: One thing I'd like Mike to maybe speak on after this; I love his black hole versus black box distinction. I think that helps here. I'll let him talk about that because that was his idea. I think you put it well; those two primary concerns. As I think about data, and data rights, I do think that that's going to be a pretty big thing that people are discussing. And by people I mean, courts and such, governments, more than just simply a few people talking about on a podcast. 

There are some great questions and great ways of seeing it. Someone can say, "Hey, look, we don't have any rights." So there are at least three groups, and there might even be more than this. You have the user, you have the platform, then you have the advertiser. The advertiser might be agencies or the advertiser might be the user; they might be the same people. 

And this is where I look at tech and where we are right now, and we're a highly new and highly unregulated industry, which is cool and great because that actually does, a lot of times, for innovation. I'm not a huge regulation guy as it stands. But sometimes, unfortunately, what that is is you look at a lot of these other places, Wall Street, stuff like that, the reason why they're regulated is because, eventually, someone came along and Bernie Madoff-ed it. So I think it'd be wise of us to at least consider what are the possibilities of stuff that could go wrong here.

Some of that would be looking at rights. The person paying for the data; do they have any rights, some rights, no rights? The platform, the one providing the data; do they have some rights, any rights, no rights? I know that there are people who will fall into that camp of saying, "Google gives, Google takes away, Google does as they please. We figure it out." Then there'll be the people who say, "No, no, the users have 100%, everything, what they want to do?"

So then you have the users in there as well. And we're starting to see that more with GDPR and CCPA in that, basically coming in saying they have the right to be forgotten. This is their data. They have the right to ask the platform to remove me. So that courts, more and more are saying platforms might not have much rights here at all, actually. 

And then you have the ad, the person paying for it. What does that mean? Money is changing hands; goods and services, something being received for that? Okay, well, are there some rights that go along with and I'm paying for something? Some might say, "Oh, no, you're simply paying to utilize the platform."

 Again, this is where people smarter than me are going to be figuring this out. I would say, and I had referenced this, at some point, Google talks about that with agencies. There seems to be some level of money changing hands. So at some point, and this is how I would define it, the data needed to do the job well is probably falling into that goods and service exchange of money has changed hands. And part of what is purchased is the data to at least give them a shot at making their business successful. 

Again, this is in-the-mind Kirk. This is not any sort of legal whatever; I have no authority. These are just meanderings and thoughts that I have. So if that's part of it is if we're paying for not simply access, and I realized there's a ton to unpack, we're not simply paying for access. But we're also paying for some level of data as well, to do that well. 

Part of why this is complex is because Google has in the past given more data. So they've also helped set the expectations of this is what you receive when you pay us. But there seems to be some level, in my opinion, of justice of right or wrong in terms of, "I'm paying for this meal. I'd like to be able to see the itemized bill of what I've paid for in terms of the big stuff. Someone might rightly come along and say, "Geez, there's no way you could possibly figure out all of that data." You can even store all the possible data. 

At some point, I also don't think that the advertiser has or even the user has every single necessary write to every single possible data because that's almost not realistic. So, again, that to me goes back to realistic data. What is actually needed? What's important? So that's probably then where a lot of the arguments will be because in some ways, some data might be important to one advertiser that utilizes it more, some might be not. 

So that's some of what I look at and I say, I have some concerns on that level of does Google even have the right to do this, to not in any way give anything back? So there's some question there. That's almost more of the pondering thing. 

The flip side of the fraud thing would be, again, and I try to communicate this whenever I've talked about this, which is I'm not in any way accusing Google of saying, Hey, here's a black hole system. You're a fraud?" I truly believe, as I've talked to them, you've talked to Google ads as well. What we've always heard, I truly believe this is the motivation of hiding the data is exactly what Patrick, you've said. They want to protect their system so it can perform as well as it can. 

So I attribute that to, it's like they're the doctor looking at the patient and saying, "No, no. I know you don't want to but sorry. You got to take the medicine, you got to get the chemotherapy. I know it sucks. It's the way it is. I know best." And so I think that's at least the core motivation. So not necessarily nefarious.

I do have concerns in a broader bigger picture, marketing, ideology, philosophy, concerns, that if you pretty much look at history and human behavior, you get billions of dollars going on. And you get the people who are accepting the money, who are also the ones that determine how that money is doled out and taken in between the various competitors and all that and you have these completely closed black hole systems, controlled only by the individual company who is taking those billions of dollars. And they are also completely obfuscating all of that. 

I don't see how, eventually, that's not going to lead to some sort of fraudulent behavior and regulation and that sort of thing. And to be honest, I'd like to keep the regulation off. I think some of the ways to do that is just by being transparent. So that would be the question I'd have is, even if we can't do anything with that data, show the data.

Patrick Gilbert: Correct. Something will inevitably go wrong or someone will inevitably take advantage? There are two things. I would say that part of that responsibility falls on the actual advertiser paying the bill because this is traditional, right? So traditional advertising, it was always a black hole or a black box where, say, you hired Ogilvy in the 60s. And you said, "I want to spend $100,000, and we're going to pay you 12% of the media. So I'm going to wire you $112,000." And you're essentially trusting that they're going to spend $100,000 on media. 

And they might take a little off the top. Maybe they're only spending $99,000 on media, and they're taking a 13% cut, even though their contract says otherwise. And from my understanding, there really was no way to really regulate that. But that fell on the brands and the advertisers to at the end of the month reconcile and say, Okay, well, we know that we earned X amount of revenue, and the lines don't completely draw in the way that we're used to it from like a click attribution standpoint because this is before all this stuff. 

We know that this is generating positive revenue. We can tell that it's profitable. We're going to continue our relationship. That's the first thing I'll say. The second thing I'll say is if there is a bad actor, and they are being transparent with even some of this data because I think I would agree, and I think all of us would agree that they can't be a completely open book, they'll have to be hiding something because, again, now we're on top of abuse of rights.

If they are actually working nefariously, what's to stop them from manipulating the data that you have access to anyway? If we're operating on the assumption that Google Is eventually going to take advantage of the system and the advertiser and take more money and obfuscate this data. What's to stop them from obfuscating the search term report? And then it's a conspiracy theory spiral out of control, where it's like, "Okay, well, if we're going to go down that route, then maybe we shouldn't be spending any money on Google, to begin with."

Isaac Rudansky: I would actually take Patrick's point one step further because I think that a lot of, and Kirk your point is very well taken. It's like, Look, you break it out. None of us are legal experts. As far as I'm concerned, forgive me if I missed something in anybody's resume. So we don't really know the ins and outs of antitrust laws. And I think the more interesting discussion anyway is not legal? What should it be? If we were the legislators. What would we do? How would we set policy?

I think that's really the perspective you're coming from. It's like if you were in charge, what do you feel is the right thing to do? And that, to me, is anyway, a much more interesting discussion because then it's up to our own feelings and It's also fun to imagine being a legislator. So that's the first thing. It's, well, what would you do and what should we do in these situations? 

To Patrick's point; it's interesting. I think a lot of the data that we've gotten has been a crutch, in fact, because if you read the biographies of the early advertisers, the Claude Hopkins, the Albert Lasker is the Ogilvy's, they had to come up with real methods of testing results, keying every individual newspaper in a hyperlocal market. They didn't have Microsoft Excel, but they did it by hand, match tables and regression models, and mixed market match tests to actually see if there was a profit on the spend. 

There's no one today that has that level of sophistication. There are but like your typical marketer, they don't because, "Look, I have a search term report. I can go into Google and I can change my columns. I could see my conversion value divided by cost. I know that's ROI. So now I'm in the top 1% of intelligence agency team members.

But people really don't have the skills anymore to take a look at, let's say you look at incrementality. It's a big question. It's like, I'm bidding on my branded search terms, or I'm bidding, typically, on Google search, how many of these conversions that are being reported in the dashboard would I have, in any way, received had I not spent any advertising dollars? That is not a simple question to answer. 

It's a legitimate question that advertisers should be able to answer, but it takes real training. And you need to know a lot more statistics than I know to actually get to the bottom of that. But that's a skill that I think moving back towards a model of, Look, we're Google. You have billboards, you have Google, you have print, you have all these different channels, we'll take your money, we'll run your ads. You go figure out if this is making sense to you. 

I think there's a silver lining or an opportunity that's going to winnow the field because you're going to need to have very intelligent, sophisticated, real thinkers to be able to sort this stuff out. That's one point. I think there's, potentially, a serious potential advantage in the future to removing what I think are a lot of these crutches, and that Google's going to say whatever they're going to. They're going to say, "Well, we have to go with privacy. We have to remove search from data privacy." That's probably hogwash. It's most likely what we're saying that they feel that this is, in aggregate, a system that will probably harm the performance of certain accounts, but in aggregate, will make Google more money. That's most likely the ultimate goal Google has is to make more money. 

Now, there's always a little bit of bad acting, but it's unlikely that, fundamentally, this is a nefarious play because we've just seen through history that at certain points, sycophantic thieves are usually found out. It might take some time, but they're typically discovered for what they are. You cannot be the largest advertising platform on planet earth and fleece a million advertisers for that long. You maybe could do it for a little bit. You can't do it for that long. You'll be discovered.

If some sub-Reddit could swing Wall Street in the most effective way in the actual Occupy Wall Street movement in 24 hours, there'll be a subReddit that exposes Brin and Page for the sycophants that they are, if that's what's going on. But really, they're looking to make more money. Anyway, my point over there is I think there's an interesting challenge and an interesting opportunity for marketers. It might force the good agencies, especially the performance agencies, which all the original advertising agencies were.

If you go back to the Ogilvy era, it was all performance marketing. They all came up with innovative, really interesting statistical models to figure out what's working and what's not working, and not just what's working, but incrementality and liquidity and how to run tests and do it within the client's budget; really complicated questions that take real minds to answer. And my hope is that if Google removes some of these artificial crutches that make it seem that we have all the answers, but we really don't, it might force us to get smarter collectively 

Patrick Gilbert: Mike, what do you have to add to this? And if you can also tie this into your black hole analogy because I do think that that's still a very relevant piece here.

Mike Ryan: Just speaking toward Isaac's last point here, though, about kind of improving our collective fitness in this area, I do think this will happen because David Riniski wrote about this already in 2016. He had this article that the SEM agency is dying, and he talks about the need to move upstream. 

And we've seen at a recent Google Marketing Live, I think, in 2019, that Google was predicting disintermediation as a result of these new technologies that are developing. That agencies are going to start getting squeezed out and there's debate about to what extent this is an unintentional strategy. But they also talked about moving upstream. 

Now, let's imagine for a second, in our minds, a kind of classic strategic framework, like where you've got going from kind of a vision at the top in a pyramid, all the way down through to tactics and daily work. You've got a vision and then you've got strategic initiatives, or rather goals and then initiatives, and then down to tasks and so on. 

If we think about that, then if you kind of map a river delta on that, you can see that there are all these streams spreading out as you go down from a single stream at the top, all the way down to these little tributaries, and so on out in the Delta. And we have a lot of folks who are very specialized in these little individual tributaries down here at a more operative level and this imperative is to go upstream. 

And that means to capture more of these little channels through integration. It means also to vertically integrate and to move upstream more. And the thing is that, as you do that, there's going to be a crowding effect, as more people all try to move upstream. And also, you're going to have the stream working against you. 

The point of this metaphor is that it's like this natural selection battle to move upstream. We're not all going to get upstream. And I think that's to your point, Isaac, that we're going to have to improve our fitness to get out there. And some people might not make that journey. But to bring it back to the black hole topic, I think that Google brand themselves as a technology company. They brand themselves as organizing the world's information. They no longer brand themselves as "Don't be evil" famously, but I don't think that they are evil.

I think that they're not actually really a technology company. I think that they're a sales and marketing organization, fundamentally. And their main revenue source comes through these advertising channels. There's increasing pressure on them from Amazon. They're losing the share of intentful search. A lot of that search is being transferred over to Amazon search. This is why they're also increasing their ad inventory as much as they can, with new offerings like the surfaces across Google. They're working on that problem. And the thing is, I view this whole topic of the full-funnel coverage, or you guys are mentioning liquidity, this is a dechanneling strategy in the end. They're taking a channel like Google Shopping and other separate channels or channel flavors like search text ads, like display, YouTube, Gmail, and they're saying, "Don't worry about the channel, worry about the audiences." You'll see marketing materials from them saying that people need to shift from a channel focus to an audience focus. 

But what's a bit ironic about that is that they removed that kind of data from us too. They make this data soup. They remove this kind of data. So this is a product strategy. It's not something that is unavoidable. A black box is something that needs to get, in my perspective, destigmatized a little bit. It's a challenge. It's a technical challenge that arises often in machine learning settings or in AI settings, where you have an input, for example, ad spend, and then you have an output like conversion value. And you might not be able to see everything that goes on in between in terms of the bidding and the auction environment, and so on. You might not know how a given bidding model is making its decisions. And this might actually be unknowable, or very difficult to figure out. 

But that's not exactly what's occurring here. This is what I call the black hole. They're actually moving beyond this and they're sucking away these data points that there's no technical reason why they can't report on these things. Or if there is, then they should just explain it to us. We know that it's a product strategy. They want to reduce agitation because people might not be comfortable with all the decisions that the machine is making. And they want as much freedom as possible for their algorithms and models to work. 

And if I talk to my data science team, from a raw performance perspective, they have a similar mindset. They also don't want that. But for me, as a product manager, I have to make different kinds of decisions because I have to consider the wants and the needs of my users, and I respect my users a lot. And not to say that the product teams at Google are not respectful. I always have great conversations and relationships with them. 

But what I've gathered is that there are people in the product organization there who don't particularly like Smart Shopping but they felt there is a lot of pressure coming up from higher in the organization on this as a business case because when you're developing features, you need to think about the feasibility: can we do this? The desirability: does the market want this? And the viability, which is how is this going to work for our business? 

And I think when you're Google and you have the resources like Google does, there's not much that isn't feasible for you. That's not much of a factor. That desirability, I think it's clear that a lot of things they implemented here are not desirable for advertisers. I think they focus most on viability. And I'm not saying that they're an evil company. I'm not saying that they're going to rip us off. I'm saying that they want as much freedom as possible to control what's going on here in the auction, and what's going on with their profit margins, too. 

I mean, they're a business. Any business has an obligation to be profitable. And I don't want to challenge that too much. I just think that they could have built a different and better product here.

Patrick Gilbert: Mike, you bring up a really good point when you refer to them as a sales and marketing company, as opposed to a tech company. And this also was similar to the conversation about are they trying to phase out agencies. This is something that's been talked about forever. And that conversation by itself has always started and stopped with agencies as if we're all the same. And you've noted that that's not really the case. There are some that'll swim upstream.

If you put yourself in Google's shoes, they want to be profitable. They have a responsibility to their shareholders. And they're the largest advertising company on the planet. And I think that they've gone full into this whole automation piece. And they've created a product. imperfect, of course, they created a product that meets all the checkboxes as far as like, "Hey, we should be able to deliver profitable revenue for clients at scale, and this is going to be great." And they released it in 2018, and nobody embraces it despite the fact that they're also doing support, all these premier partners that have all the support teams that work with them. And the training module is just coming out of their pocket. And nobody embraces this product. We're among the very few people that do. 

Meanwhile, throughout 2019, they watch market share go to Amazon, and Facebook, and other ad channels. And then all of a sudden, 2020 takes place. And people start freaking out that they're losing their Google support team. Of course, you are because nobody was listening to any of the Google support strategists that were saying, "Hey, listen, try this tool out and don't touch it. I promise you, it'll be better than standard shopping." And not enough people are willing to listen to it. 

So I don't think that it's that Google wants to do away with agencies, I think that they're just okay if some of those bad agencies fail. They'll never admit that, but I think that that's what the directive is. And they absolutely should want agencies to exist. All of us here are unpaid Google salespeople. Think of how much revenue that we brought to them. We're here promoting Google's products right now. I've never received a dime in commissions from that company. 

Mike Ryan: Yeah. I can't really say the same because our business model works on collecting a percentage of the ad spend. And yeah, it's not through Google. Google would probably view that as different from them. 

Patrick Gilbert: I agree. I think that there's some middle ground there. Either way, I think the black hole analogy is very good; very strong. And there really isn't a way to get to an answer to it. I do agree that that's what's happening. Then it comes down to, is it right? Is it correct? Or are they breaking the rules? And that goes back to legislation.

On what do advertisers have the right to conversation; this is where I think all of us got most passionate about. What do we deserve? And if you think about it, like Kirk, you had mentioned earlier, Google kind of set a precedent or an expectation that some of this data would exist for advertisers. And now they're taking it away.

And as we've discussed, they were really the first advertising platform to be that transparent about anything. If you've been running Facebook ads at any point ever, you never get that level of segmentation. But even going before digital, nobody running advertising had anything close to what we had access to, or even have access to now. And regarding these tools and levers that we're saying here help us drive success in being able to pull out individual search queries or audiences or this product category is more profitable and generates more lifetime revenue in that product category. That's the sort of stuff that has been icing on the cake for what is otherwise a pretty robust advertising platform. 

If you were to go to the New York Times and buy an ad placement, look, advertising is not new. If you were to go within the last 100 years and say, "I want to buy an ad placement in New York Times, but I want the condition of saying, I only want to buy an ad placement in New York Times that are sold in these neighborhoods, or for these kinds of people, or only when you run these sorts of articles," You're pulling levers. The New York Times won't have that. 

Now, we could all agree that if you could pull those levers, it would help your campaigns be more profitable. But the New York Times or anyone really has never given advertisers that option. It's just, "Okay, do you want an ad in New York Times, it's going to be printed everywhere. It doesn't matter if we're selling papers in Antarctica.

Isaac Rudansky: It's not even really standardized pricing because you have the right to negotiate. If you have an inside connection or you have a better negotiation skill set, you could get cheaper rates, right? So it's a whole different world.

Patrick Gilbert: It's totally different. And so, even if they did have that sort of product, try to envision that for a second, it would absolutely come at a premium, where you would say, "I'm only gonna pay for customized..." Newspapers absolutely have the ability to print customized; everything is just print on demand. They could do this, and they could charge a premium for it, but they don't. Google is actually giving you the ability. And actually, just take a step back, you would say, "I only want to serve in New York Times ad in these neighborhoods because I believe that the folks reading this paper are a higher quality audience to me. And I'm making that decision based on predicted conversion rate. I'm saying these people are more likely to buy my product, which is more likely to make me profitable." Great. And we can all agree that that's how this works. 

Google is giving you the ability to do that. They're just not letting you be the ones to pull the levers. So instead of just saying like, "Okay, well, I'm just going to place an ad on Google. And I hope my audience finds it in the same way that it's like, "I hope my audience is reading the New York Times today." Google is saying, "No, we actually have all the data that says that it's not just this neighborhood, but it's three people that live here, one lady that lives over there, one guy over here, and we're going to handpick these individual things because we know a hell of a lot of information about all these different folks. And we can actually optimize this in a much more effective way than my New York Times example, everyone would have been profitable. And we're not even going to charge you a premium to be able to get access to them. So I think that's a different perspective. Do we really have the right to this data? And does it really matter?

Kirk Williams: It's interesting. One of the things you noted, Patrick, that made me think of something; Moore's Law. Every two years, you guys probably know better than I do, but something about the processing doubles, I think, the chip capability, something like that. Basically, the idea of machine learning actually gets smarter in part simply because there's more processing power. Also, machine learning is getting smarter as it's learning and teaching itself. 

So one interesting thing, you had said in 2018, no one embraced it, right. 2018, we were absolutely testing Smart Shopping, absolutely. To your point, we could have simply been not letting it run long enough things. We had a few times, though, where we were letting it go quite a bit. And it was interesting that, for the most part, in a strictly personal just from what I have observed, which can be erroneous, back in 2018, when we first started with smart shopping, most standard shopping campaigns were still kicking its butt. And that's just not happening now for me. 

So some of that could be that I'm figuring out the system better. I think some of it, in my defense, if I can, some of it might be the Moore's Law thing as well. It's been two years now. It would make sense to me that their machine actually has gotten a heck of a lot smarter in two years. How could it not basically when it's learning and upon learning?

So in some ways, some of that might even be just this thing of it's 2021 now. Great. You tried smart shopping a couple of years ago, you should probably give it another try, just in terms of results. 

So some of it in advertisers' defense is some of it is I think we were testing it as well, and not necessarily seeing the results. And it wasn't simply that. But I know you guys have done and seen it. So some of that could be user error as well. 

One other thing that I still chew on and I wrestle with, one of the things I've loved about learning from you, Patrick, is this idea of how many signals really are used in an auction time way, and how many things Google is looking at beyond what we can. The flip side and this is still what I wrestle with a little bit is, I perhaps have an unhealthy level of the keyword. But at the very least, I see a person individually communicating as at the very least a much stronger signal than most of the others. And so some of that works into just my still wrestling and trying to think through you losing something like search terms. I'm not sure if that still is fair to apples to apples comparison. I'm not sure if that still is the same thing as the New York Times type thing because, at its core, keywords and search terms really are still, I'd say, the most important signal. 

And the reason I say that is because all of the other audience stuff and time, all of those other signals are absolutely helpful and used by Google in ways that we couldn't. But still, people are random. People do not always communicate their intent well in their behavior online. People can communicate a certain thing, they can act a certain way, and then they can change their minds. And so that's still where I put a decent amount of weight on someone saying, "You know what, I've decided to go to Google and to type in this specific, intentional thing right now."

And that might even be something completely different than I thought before because I learned something new and I've changed my mind on something. And I'm now telling Google, and that's what a keyword is, now telling Google, "Hey, I'm interested in this now." And that's where I still see the keyword and that search term as a signal, but not on the same par as a signal, "Hey, it's 8:00 am on a Tuesday." 

So I think that's still where I'll wrestle a little bit with losing that search term data because I still see it. But genuinely, what you've been communicating, some of this has been a long process for me, really wrestling with this stuff, and thinking through this, and understanding the way that algorithms work. And realizing in a humbling way, gosh, there's a lot happening out there that I could never do at an auction time thing too. 

So trying to keep all that in mind, but also wrestling with the true intent of search has always been around that individual communicating to us. And that's got to be a stronger signal. And so it's got to be a fairly important part of the whole mix, which is also why then having that data can inform other things so well. Because, man, if we can see that over the last six months, for all the other signal reasons, a million people typed in this keyword and purchased. Yeah, there's a lot of other signals, but that amount of people, they all typed in that search term, we probably better make a landing page for that. And so to that point of utilizing it for broader business applications. That's my unformed, still wrestling with some of that thoughts. 

Patrick Gilbert: I completely agree with that. And I think that what is actually happening is that there's still a lot of weight placed on the search term. I still think that's the primary signal. But I think what is also happening is that there's a lot of value in more broad, other keywords that you can't really tell how good they are for your business. And if you're experimenting with dynamic search, this is where this was a big eye-opener for me in seeing the value of competitor keywords, something that I'd really never seen success with in the past. 

All of a sudden, they start showing up in dynamic search and converting in a profitable way. And then thinking through what's happening on the back end. And it's likely that that person was searching for a number of different things over the course of a week, and then at that point, searched for a competitor and we showed an ad and they actually converted on our site. And normally, if I was just looking at that one individual keyword, I would probably say it was bad. But Google knows a lot more about what's happening when that specific keyword is being searched. And I think that's where the additional data comes in. 

I think we can all agree that life bottom-funnel keywords are the most valuable thing for us to be able to leverage. And your point about making landing pages; that's something that we've done. And losing that ability is something that I really have not even fully come to grips with because it's been so valuable to us. But I think that expanding reach is really where the value comes. 

Kirk Williams: Yeah, I think that's a great point. And we've been experimenting with exactly that. And I've been sitting surprised how well the DSA straight broad match when we are also given it more aggressively, "Hey, stick to this ROAS goal, things like that. And I think, to your point, that's because Google is wisely using all of those other signals to say, "You normally never would have bid on this keyword."

And to be fair with the old system, that's because it wouldn't have worked Because if we're just bidding on that keyword, we're just getting everyone losing a lot of money. And now Google's saying, "Sometimes when they if this person types in this because they want to buy 4,000, Kansas spindrift, or whatever it might be." To your point in the books. See, I did read the book, not totally done with it yet. 

But anyway, I think that's a great valid point. And that is especially where I love the idea of all those other signals being used because it is helping you find exactly keywords that maybe you never would have assumed worked, or the reason why they're working is that all that other stuff is included. I just want both.

Patrick Gilbert: I know. I totally agree. Mike, anything to add to this point?

Mike Ryan: Yeah, a couple of things or just one thing. This is almost trivia, but regarding Moore's Law. I don't think it gets discussed a lot, but there is a limit to Moore's law. And we're sort of approaching it and that's just the physical limits about how small you can make processors and working at an atomic level because basically smaller and smaller chips are more powerful, or that computing power can be contained in a smaller space. Actually, it'll be algorithms that extend our computing power. Already, it's largely algorithms using that computing power more effectively. And then the next frontier will be quantum computing; just a random thought. 

I just think like this topic here tying it together about what data are we owed or not and is this stuff valuable to us? I think an interesting thought experiment, I live in Austria now. And everyone here still drives manual transmission. And this, to me, is fascinating. I had to learn a lot of new things when I moved to Austria; new geography, new culture, new language, but the worst was learning manual transmission after driving for years and years with automatic transmission. And this is like switching from automated bidding back to manual or something.

But what's so fascinating to me is that many Austrians will never drive automatic transmission until they drive a fully automated car. And this is so weird; this cultural preference toward manual transmission or resistance against automated transmission has been going on for generations now. It's not a new technology. 

And I just wonder if we think about this kind of black box topic or the black hole, these self-driving cars will make decisions that we might not understand, and which can have life or death implications. And so just raising the stakes a little bit, what kind of data are we owed there? How urgent or imperative is it that these decisions are understandable? And when we're talking about PPC, the stakes are a lot lower. But it gets to this issue of trust. 

And I think that people, even if the stuff is not actionable, they just want to see it on a basis of trust and an understanding. And I don't need to have my intelligence insulted by saying that I'm just going to tamper with things or make these bad decisions and so on because I do believe, to this point, smart shopping is not standing still. It's a moving target, whether it's improving through Moore's Law, or better models or whatever. It's, of course, going to improve and over time.

But it's really tricky to say what we are really owed or not? I'm not convinced that Google always has this kind of data. But we're used to it and it's definitely really painful to lose it. And now, it's been an interesting week. There were a couple of things popping up where people like Harpal Singh were unmasking some data in Smart Shopping that hadn't been exposed before. Alena Romanova had a really interesting post about some reporting possibilities. 

And meanwhile, we've seen search term data going dark in standard shopping too. And it feels like people are going to hack away and there's going to be this data leakage and people find clever ways to find it and start lifting the black box a bit. Meanwhile, standard shopping; the Blackbox is descending. So I wonder if it'll meet in the middle somewhere.

Patrick Gilbert: The automatic transmission analogy is fascinating. 

Mike Ryan: I think so.

Patrick Gilbert: That's incredible. Yeah. And I think that's a really good point. Looking forward, though, I do think that it is inevitable that this is the route that we're taking, whether it's by choice or being forced. And I think it is important to think about what is next? And how can we survive and continue to provide value to our clients, and all that other good stuff?

To Isaac's point earlier, I think a lot of this has forced us to really think hard about where we sit in the middle. And I think some good will come out of that, and I think some already has. And I want to share with you a pseudo case study, that is hot off the presses that we've been working on over the last week or so here, to help quantify the real value of what smart shopping means. And I'm very excited about what this actually shows.

When looking at the merits of an ad campaign, in the traditional sense, you would have to do some sort of complicated regression analysis to be able to figure out what is the incrementality, as we've discussed. And we've been doing this a lot more with different types of non-traditional media like YouTube advertising. It's really hard to prove the validity of your YouTube ads if you're just looking at the Google Ads dashboard because, well, it counts on views and view-through conversions, but ultimately, you know that it's having an impact beyond that. 

What's not tracked in a YouTube ad is, for example, if somebody saw a YouTube ad, chose to skip the ad, so therefore, it didn't even count as a view, and you weren't charged for it as an advertiser. But now subconsciously, that user is now familiar with your brand, and it just might impact their conversion rate, or even a click-through rate further down the road if they're shopping for your product, a month from now. And that's very hard to measure. 

So we've done some awesome... We have a guy on our team that's spearheaded a very awesome list of projects for a number of clients, where we're trying to measure brand lift and incrementality at scale. And it's done some pretty fascinating work with it. Just last week, we had this thought of saying, "Okay, well, Smart Shopping, is not just the traditional search ads. It also displays YouTube and Gmail. Is it possible that we're getting an incremental impact that's not being measured in the Google Ads dashboard? And that was the hypothesis we worked with. 

We looked at one account in particular, which is an account that I myself have been managing for the last five years. Their spend is in the six figures. They have a pretty competitive space. And what makes this data set interesting is we embraced smart shopping early, but we kept standard shopping running for about a year. So there's actually a very, very clean data set to work with. 

And even though the actual realized return on ad spend that's being attributed in the platform is virtually the same, we found a significant incremental impact in overall site revenue, since the implementation of smart shopping. We found every single dollar spent, even if we're tripping, every single dollar spent in smart shopping is being attributed in the same way from return on ad spend as smart shopping. We're seeing a 76% incremental impact from smart shopping. Meaning, the massive reach that we're getting now from these ads is actually putting $0.75 more on the dollar into the pocket of the site than when we were doing the exact same thing and realizing the exact same revenue three years ago. 

The second part of the study that we did was to look at whether or not impressions by themselves had a statistically significant impact on revenue lift. Standard shopping did not, which is not a surprise. If you just see an ad on Google Shopping, it's probably not going to influence your conversion rate at any point in the future. What smart shopping did, meaning impressions, the sheer impressions from smart shopping, proved an incremental lift in revenue on the site by a pretty dramatic means.

So this is brand new. We're still diving into this concept, but I think as far as comparing smart shopping to standard shopping, it's not even an apples-to-apples comparison because we're seeing, and this is based on anecdotal evidence that I've had. The only change that we've had to this customer over the last year was that we've doubled down on smart shopping and overall revenues up 76%. What's being attributed to that? Because we're not seeing it in the dashboard. 

So that's an anecdotal thought and we're seeing that in a few accounts. But now being able to say, "Okay, well, we're actually seeing a ton of revenue that could be attributed to impressions, and it's like overall exposure. And we're now trying to map that back in a more sophisticated way, which makes me even more bullish on smart shopping as a product. 

Kirk Williams: It's funny you say that because I think I tagged you all in my LinkedIn question the other day, in terms of the free ad clicks, the free video views. So I've been wondering about that, with that branding type of a question. Sorry, for the sake of the podcast; there's a way in Google ads, you can look with your dimensions and that you're reporting to see free ad clicks. And if you do that with smart shopping, it'll show you...

We had a client where it's 104,000, I think, it was video plays. And then it was like, 104,000, video pauses. So that was the other event. Apparently, it was an autoplay ad. So I was trying to figure that out with some people because I'd just never seen that before. But one of the things I thought of was that. Man, what sort of impact is this having if in the last 30 days, whether or not even this has been clicked on? This product has been shown to people 100 100,000 times. And even if they've hit that pause button and moved on, we were all advertisers enough to know that there's some branding value there. 

And exactly, you said; it's hard to quantify that. That's where you're starting to figure that out with incrementality. And so props to you all for digging in, and really thinking through how to measure that. But that's interesting because I was literally just talking about that last week because I'd never seen that report before. So cool.

Patrick Gilbert: I really want to find a client that would be willing to put up a video ad, that in the first five seconds, they do their sales pitch and then stay like, "Now click the skip button, but I'm going to keep talking so that I don't get charged for this ad, and I'm passing on those savings to you." Something like that; I would love to do something very cheesy, just to show that to get as many people to skip the ad as possible; that's a free impression. But it's certainly valuable. So I think, one of these days, I'll convince someone to run that ad and we'll circle back as far as what the impact is.

Mike Ryan: Yeah, I want to see that case study too, but, totally, this is like the mirror awareness effect, just having these impressions is quite valuable. And you don't get that in a standard PLA. The branding factor is too small there because unless you are Nike when someone's searching for Nike shoes or something. But otherwise, if you're a retailer, for example, then they'll see it's not very prominent either way your brand is represented there. 

And then even now, in Europe, there are also the CSS brands represented there as well. So, it's really interesting and I'd love to see it. Please do share the case study once you've finalized it. It's not too surprising, though. It's definitely reach is the USP of this campaign type. And that's really exciting. 

I think it's definitely a valuable use case for it in an area where smart shopping is particularly strong. A lot of advertisers out there, they're not doing enough activity on their brands, and in the upper funnel. And it's super important because you're not going to make it in the long run on just these points of sale or these acquisition-based advertisements because your customer experience then has to be really special. You will get some new customers for sure, but that they're returning this stuff over time. Branding is so essential for supporting that relationship. 

Patrick Gilbert: Absolutely. I would say let's try and wrap this up. You guys have any final thoughts as far as where we're going, concerns, and of course, anything that you guys want to plug in or some next steps here? Mike, we'll start with you.

Mike Ryan: Sure. Well, I was thinking back on that thing about the self-driving car and now it's not life or death. I was talking to a client the other day where they've got an 85% revenue share with Google Shopping. And for their business, this is kind of life or death. So they just had a little gravity to that.

I think that it's a really interesting campaign type to watch. I think what we'll be talking about, I bet, next year will be performance max. I've mentioned this concept of truthful auctions in the past, which is how I think Google will manage these situations in the future. And I think this works through getting valuations about what conversions are worth. And to me, new customer acquisition is perfectly going in this direction, and also the way that they solicit Cost of Goods Sold data and as a feed attribute and so on. I think they're going to work on collecting as much data as possible to try to help model what's going on in your business, not just with your users, but with you as a business as well. 

And if I have a minute to just plug something, then I'm going to be speaking at SMX Munich in March. It's a virtual conference So you don't have to fly to Germany. Definitely check it out. I'll be talking about marketplace suffocation, basically. The trend of big retailers to start taking on marketplace characteristics and business models and the presence of big marketplaces like Amazon and eBay, and the Google Shopping auction and what this all means for businesses. 

Patrick Gilbert: Amazing. How can people find you on the internet?

Mike Ryan:

You can find me on Twitter @mikeryanretail, and you can find me on LinkedIn. There's a lot of Mike Ryan's out there. But I think my LinkedIn snippet is like Mike-retail-insights, I think

Patrick Gilbert: So if you search the article about smart shopping data, you can't handle the data, whatever the tagline was, I'm sure that it'll link back to your LinkedIn,

Mike Ryan: Yeah, search for "Would Google's own marketing team use Smart Shopping campaigns?"

Kirk Williams: It's a great article. I read it again last night. 

Patrick Gilbert: Thank you, Mike.

Mike Ryan: Thank you very much.

Patrick Gilbert: Final thoughts, Kirk? 

Kirk Williams: Maybe I'll make a plea to both my fellow PPCers and to Google. My plea to fellow PPCers would be where I'm at, which is I don't have all of my objections answered in my own mind. I'm still kind of thinking through this, still trying to figure it out. But also, at some point, I think that we need to join or die to use Patrick's book title. And in some ways, I do think that getting into the system and starting to figure it out is important. Again, I know people who haven't even really tried it; haven't tried Smart Shopping, haven't tried some of these more automated solutions. 

As we have started to get into it more, I am surprised by how many things we are figuring out, "Here's a lever I didn't even know was a lever." And we're discovering, as others have, that there's almost a more fun aspect to it because we're not spending our time setting bids, we're spending time talking in the office, what are the strategic goals that should shift for this client that we need to communicate to them?

And to be honest, there's a really fun aspect to that that I've been enjoying. I would recommend that if you're trying smart shopping, we have begun to do far more than just simply throwing a campaign live; but trying to find different ways of grouping products and matching that to the ad creative. That's something we didn't even talk about. But that ad is something that isn't even part of shopping ads. And it's actually a way for you to do marketing and have images and videos and things like that. It's its own thing. It's still an entity that is even an option for standard shopping.

So there's almost this other layer of that and grouping that with products. And there's a lot to do and to figure out. So give it a try. That was to my fellow PPCers. I think I would know about Google. I know that we can be frustrating sometimes, but that's because we love PPC like you Google. And I have noticed a concerning trend in communication and public orientation towards agencies that I would just like kindly encouraged to reconsider. 

We are, as Patrick said, your free sales team. And I know that it can be frustrating, but also, it is worth the time to help us to be patient, help us walkthrough, and help us think through these things. These sorts of conversations, to be honest, I would have expected to have you know more with a Googler. And it's even been harder and harder to even just have those conversations, to be frank, with them. 

So I think there'd be some level of, hey, we've had plenty of stick. Now let's also have a carrot; carrot and stick maybe both. So yeah, there you go. I guess the thing I'd plug would be like I said, I'm finishing up a course hopefully next week on Smart Shopping. It ended up being a lot longer than I thought. It's taking me forever to edit because I wanted to edit myself, but I think it'd be good and helpful. And hopefully, I'll just keep editing it and changing things over time as I learn more about it, too.

Patrick Gilbert: That's amazing. Looking forward to that. Isaac, do you have any final thoughts? 

Isaac Rudansky: Well, I want to thank both Mike and Kirk for joining us today. It's an honor. I think it was a really interesting conversation. What I get out of it is totally different or not totally different than what anybody else is getting out of it, per se but outside of the topic of paid search, that we can have conversations where people are essentially arguing that Google or this platform, or this person is fundamentally, probably good, but there are still a lot of things that I don't like about it. '

That is already an elevated type of conversation that most people cannot have today because people have become ideologues. And when you're an ideologue, you're irrational. Your target is either all good or all bad. And we see that in the political sphere today; we see that even with friends. It's like, "Well, I can't talk to you." You can't have a sophisticated conversation about something. 

So I think that this conversation was just the opposite. It was listening to different perspectives; orienting yourself in the conversation with the belief that someone else actually has something valuable to say, which it's also very, very rare. I'm sure we're all guilty of this, to some extent, in our own lives. I certainly am, like listening to somebody, and then I'm catching myself, "Wait, like, I actually don't think that I'm going to learn something from this conversation." And then you have to sort of switch your brain be like, "Wait, maybe I can learn something from this conversation. Maybe this person does have something to teach me."

So I recommend you Mike, Kirk, Patrick for having this type of discussion. I think the four of us probably have all four different perspectives, to some extent. But I feel that we were listening to each other, trying to actually understand a different perspective in the hopes of enriching and deepening our own perspectives. And that's the opposite of being attached to an ideology or being an ideologue. And I think that's a great thing. 

And, in general, more conversations to be like this. So, to me, that's something I think listeners should take out of as well. I don't have much to add as far as what you could take out of as far as smart shopping and Google and strategies. We've covered all of that good stuff. And hopefully, we'll have a lot more of it. 

So, again, Mike Ryan, Kirk Williams, Patrick Gilbert, you've been listening to the How to Hide a Dead Body podcast. Check out Ponderings of a PPC Professional by Kirk, check out Join or Die: Digital Advertising in the Age of Automation, by Patrick, check out all the great work Mike's doing over at Smec and Orbiter. We look forward to continuing this discussion. And we'll catch up with all of you guys very soon.