Many marketers still believe that a) they are better than algorithms at making decisions and b) they can hack the algorithms to give them a competitive advantage. This is a backwards way of thinking.
Marketers should be working alongside technology, remembering it is humans ultimately on the other side of their marketing campaigns that will determine whether an advertiser turns a profit.
Marketers need to embrace automation and stop placing manual constraints into their Google Ad campaigns, restricting the algorithm’s ability to drive positive results. A strategic, modern approach to your Google Ads account structure will provide the necessary foundation for automation to work.
This article outlines the modern Google Ads auction, shows how our team approaches Google Ads account structure, and provides a new perspective on developing your own algorithmically driven, customer-centric approach to Google Ads strategies.
Your primary responsibility as a PPC account manager is to profitably grow your client’s (or your own) business. You should be less focused on secondary outputs like cost-per-click, average position, search impression share, and Quality Score.
You should avoid practices that only serve to give your work the appearance of value, and instead, pursue strategies that actually have concrete, positive effects on performance.
For example, an overly segmented and complicated Google Ads account structure may seem impressive to your clients… but times have changed. You should be willing to exchange that short-sighted playbook (which is doomed to fail anyway), for consistent, profitable growth over the long term.
Today, profitable growth with Google Ads can only be achieved by training Google’s algorithms to make better decisions on your behalf.
Forget everything you think you know about Google Ads’ account structure best practices. Automation can succeed under any account structure. Ultimately, the decisions that lead to your automated strategies will determine your success – not your match type strategy.
A simplified account structure could make your workflows easier and allow you more time to invest in your overall automation strategy, but the structure itself will not make you any more or less profitable.
For example, implementing SKAGs is a waste of your time because SKAG’s won’t impact profit. Testing inflated conversion values that you feed into a Target ROAS bidding algorithm is a much better use of your time; that’s something likely to impact profit.
So you can make the right decisions for your own account structure, you need to relearn your fundamental understanding of Google Ad’s backend systems, starting with a fresh understanding of search behavior, the classic marketing funnel and Google’s most recent updates to their auction algorithms.
Unfortunately (and to my interminable frustration) search marketers still believe search queries that include “best” and “for sale” need separate ad groups. Yuck.
I recently had to order a new golf club. I accidentally let go of a wedge during my follow-through and sent the club sailing into a lake. The club flew further than the ball.
I had no interest in experiencing the sensation (or giving the onlookers the viewing pleasure) of diving to the bottom of a lake fully clothed. So that night I performed a Google search for 52-degree wedge on my phone. Google served me a Shopping ad from Wilson (a brand I trust), showcasing a wedge in my price-range. I bought the wedge in three clicks..
(I've traditionally purchased clubs manufactured by rival brand TaylorMade, but the unique variables of this shopping experience lead to purchase a Wilson Harmonized Wedge.)
Here’s what’s important: I didn’t search best 52-degree wedge. My search was not long-tail, nor did it contain a specific brand or model number. Does that mean that my search wasn’t considered to be of high value by Google or its advertisers?
Whether I like it or not, Google knows that I play golf often and that I’m a sucker for impulsive golf-related purchases. Google probably knew I played earlier that day (I used Google Maps to get to and from the course).
I’m not sure if golfers tend to have a higher conversion rate on golf-related purchases within 24 hours of finishing a round, but if that trend exists, I’m sure Google’s algorithms factor that into their predicted conversion rate analysis.
The team running the ad campaigns for Wilson Golf were willing to bid on my broad search because they were following modern best practices. They won my business as a reward.
Advertisers who follow best practices of account structure and automation will win the best traffic … regardless of the search query.
If you win the best traffic, leveraging Google automation to properly bid on that traffic (based on predicted conversion rate analysis), then you’ve unlocked the secret of profitable growth.
Many PPC accounts were traditionally set up using the following marketing funnel approach. This hypothetical funnel was used to assume purchase-intent based on what users were physically typing into Google.
In the graphic below, CVR represents Expected Conversion Rate. If a user’s search query was more specific, we assumed a higher conversion rate… and we would manually bid more aggressively on that search.
If you were an ecommerce business that sold golf clubs, here’s how your campaigns may have been structured:
Top Funnel (TOF) Search Campaign
Middle Funnel (MOF) Search Campaign
Bottom Funnel (BOF) Search Campaign
All three of these example keywords are targeting users that are likely in the market for the kinds of products that you sell. As you travel down the funnel, the search becomes more descriptive.
The MOF search includes the genre of club we are searching for, while the BOF search includes a genre and also includes a specific brand, model and year.
If we determine expected conversion rate based on the context of the actual search query and nothing else, our bottom-funnel keyword, [TaylorMade M2 2017 52-degree wedge], will always represent a user with the greatest intent-to-buy.
(Intent-to-buy is synonymous with predicted conversion rate. Classically trained PPC strategists would assume that the conversion rate for a bottom-funnel keyword will always be greater than a top-funnel keyword, holding other variables like location and device constant).
It is completely foolish to operate by that principle. Who are we to predict relative conversion rates based on nothing except the context of the search query?
My search for 52-degree wedge could be considered a middle-funnel keyword, but my intent-to-buy was through the roof, due to variables not made clear in my search query.
(The thought of potentially hearing “Gee, Patrick, I bet you wish you had a 52 in your bag, huh?” from my buddies as I lined up an approach shot during my next round was enough of a motivation to buy a new club as quickly as possible. In this micro-moment, I was less sensitive to price and more sensitive to quick shipping.)
That said, this traditional funnel structure is still a safe bet. If profitable ROAS is your only goal, then this strategy will likely help you achieve it.
There is a finite amount of BOF inventory available, so to achieve profitable growth, we will ultimately have to expand into MOF and TOF keywords (and other broader targeted campaigns such as Display and Video).
However, we would logically deduct that it will never be profitable to aggressively bid on all users that perform TOF searches for “golf clubs,” but if you leverage Google automation to find the desperate souls that perform that search, just hours after they accidentally chuck a wedge into a lake… then you’ll find your competitive edge.
The traditional funnel methodology limits the size of your potential market. Without automation, you will never be able to profitably scale your campaigns beyond bottom-funnel search.
The primary benefit of automation technology is its ability to predict the conversion likelihood for a specific ad auction with greater accuracy that humans can, regardless of the search query.
Not all top funnel searches for golf clubs are worth bidding on… But automation will be able to make that call.
Therefore, if your goal is profit AND revenue growth, then automation is for you.
Google has the power to consider millions of signals (data points) to determine the likelihood of a conversion. Search Query is just one of these signals.
Beyond search query, Google will consider information not just about the user performing the search, but also specific conditions of the moment in which the search is taking place.
But Patrick, some of these signals have always been available to advertisers! I can manually make time of day and device-level bid adjustments! Why is automation any better?
Earlier I outlined how all users performing the same keyword search are not created equal. The same logic applies to any manual bid adjustments that have been available in the Google Ads UI.
It is illogical to assume that all users performing keyword searches at 2 pm are more valuable than all users performing searches at 10 pm.
It follows, that:
I remember streaming the 2016 Google AdWords Performance Summit to the conference room TV in our old office. Sridhar Ramaswamy, Google’s Senior VP of Advertising and Commerce used the phrase micro-moment about two thousand times.
I was skeptical. I even wrote a blog post poking fun at all of it. The micro-moment seemed like a tall tale – great on paper but impractical and out of reach for most advertisers.
Four years later, I stand corrected. Automation helps our clients reach millions of micro-moments every day, and profitably convert those users into customers.
Here’s a handful of possible micro-moment signals Google used on behalf of Wilson, helping them win my business:
Disclaimer: I have no way of knowing if any of these specific data points exist about me. This is a simplified, speculative example for the sake of added context.
Google’s algorithms now use account-level signal data when making decisions. Your performance data is no longer just tied to the specific keyword or campaign where it is attributed to.
For example, if a user converts after performing a branded search, Google will register all the signal data related to that conversion and use those learnings to make smarter decisions for future auctions across the account.
In this example, the branded search query itself is just one of many signals that Google strips down and analyzes.
Additional signal information, including the demographics and interests of the user that converted, previous Google search history, device, YouTube history, and other factors are also stripped out and will now be used by the algorithms to predict conversion-likelihoods of any other auction taking place in your account.
Therefore, this branded conversion will help Google more accurately bid on other campaigns, including non-branded search campaigns, Smart Shopping campaigns, YouTube campaigns, and more.
Tying this back into account structure, here’s two questions we often get from clients:
Yes, you should bid on branded terms. No, you should not set up multiple accounts unless you absolutely need to, because you benefit from having as much accurate conversion data in the account as possible.
*Note, after originally publishing this post, a colleague at Google pointed out that signal data can in fact be shared across accounts when you use MCC level conversion tracking. If you have to setup multiple accounts, make sure you also create a MCC account, and setup your conversion tracking pixels from that level.
Wilson Golf likely had thousands of branded conversions that helped the system accurately bid on my auction. For starters, Wilson isn’t exactly in the top-tier of golf brands like Callaway or TaylorMade, but I am not in the top-tier of golfing hobbyists, nor am I the kind of person that would be willing to spend more than $100 on a replacement wedge.
I have a lot in common with others that have purchased this exact golf club, and Google knew that.
The modern marketing funnel doesn’t look much like a funnel. In fact, let’s throw out the funnel concept altogether because what we really need is a framework.
Below is how we currently approach Google Ads management. This applies at the macro-level (account-wide) and the micro-level (individual campaigns).
A few examples of these stages:
Best practices and new features
Leverage Attribution & Audience Signals
*Note: the graphic above references “Google (and other ad platforms).” This framework should be applied to every advertising initiative. Facebook and other platforms also leverage machine learning, and so you must be mindful of best practices and the data inputs that you are using throughout your marketing channels. This is not to say that your Google data will impact performance on other platforms.
This framework challenges us to constantly seek larger scale business solutions, forces us to be methodical in our approach, and ensures that we are running proper tests and making decisions based on statistically significant, accurate data.
With properly implemented automation and account structure, you should be more hands-off with your negative keyword lists.
PPC managers used to brag to clients about the number of negative keywords they added to the account. However, negative keyword overload will most likely have a negative impact in the modern Google Ads ecosystem.
If you take more of a hand-off approach to negative keywords, the system will learn which queries will not result in conversions and will use these learnings to make proactive, smarter decisions at scale.
One of our clients manufactures custom picture frames. This is an extremely competitive arena with massive amounts of search volume and lots of opportunities for growth.
I’ve personally been working on this account for more than three years. The digital advertising landscape has changed a lot in the three years, as have my own habits and perspectives about how to drive success and grow their business.
A few years back, my colleagues and I dedicated time each week to groom through search term data and look for negative keywords. It was extremely monotonous work that I dreaded doing, but it was essential to maintaining a positive ROAS at the time.
People search for all kinds of picture frames. I remember sitting in our old office and coming across the search term Ohio State Buckeyes Picture Frame. My client doesn’t sell Ohio State Buckeye frames, or any gimmicky frame for that matter, so I added “Ohio State Buckeyes” as a phrase match negative keyword and kept moving down the list.
A week later, the search Ohio State Picture Frame appeared in the search term report. I realized, duh, I should have just added “Ohio State” as a phrase match negative keyword the prior week. It was a helpful lesson in how to rationally think about negative keyword implementation and I vowed to improve my processes moving forward.
Soon thereafter, I stumbled across the term, “Penn State Picture Frame.” While I would assume that the user performing that search is of much greater intelligence and is probably better looking than the losers looking for Ohio State Buckeye frames, alas, we do not sell these types of frames either. "Penn State"was added as a negative keyword.
Trying to be a more proactive account manager, I spent the next few hours compiling a list of every major US university, sports team, vacation destination and holiday that I could think of. My negative keyword lists were bursting at the seams with phrases containing “engagement” and “corgi” and “I love you grandma” (that last one is a real thing, but my client does not sell it).
There is an unlimited amount of weird, personalized picture frames that people want to buy that my client does not offer, but there’s no way for me to proactively negate all these search terms. There are also negative keyword limits to each account that we came close to running into.
In hindsight, if we were properly using automation at the time and allowed the system to naturally learn about individual search terms and their respective conversion rates, then Google would have eventually learned that we don’t sell any type of gimmicky, personalized frames. All that time could have been better spent better serving the client (or practicing my golf swing).
Therefore, the concept of wasted spend has been entirely replaced by the concept of investing in algorithmic learnings.
It is no accident that we cannot add negative keywords to Smart Shopping campaigns, or even have access to the search term data for these types of campaigns. Google doesn’t give us access to all that information because they don’t want us manually tinkering with, and screwing up, this learning process.
Back to my drowned wedge and my sunken self esteem. I eventually purchased a Wilson Golf Club, even though the rest of my clubs are of the TaylorMade brand. I would have loved to buy a matching TaylorMade club, but Wilson offered a much more affordable option.
Several years ago, any PPC account manager likely would have added competitor branded terms as negative keywords. Callaway, Titleist and TaylorMade would have all been account-level negative keywords.
This would have been a mistake.
Even if my recent search query included the term TaylorMade, there is still a very good chance that I would have ended up purchasing the Wilson wedge.. Google’s automation would have anticipated all that. Google would have likely (and correctly) shows the Wilson ad, even though I was specifically searching for a competitor’s product.
That said, there are still instances when you should be adding negative keywords (especially for Dynamic Search campaigns) but proceed with a bit of restraint. Be cognizant of how your decisions will impact algorithmic learning.
But enjoy it while you can. I’m betting that negative keywords (and likely all search term data) will be extinct within the next two years.
A search for Google Ads Account Structure Best Practices yields 85.7M results (compared to just 13M for a search for Google Ads Smart Bidding). There’s a lot of outdated (bad) content out there, and most of these hacks have since been addressed in updated support docs or other material from Google.
Important facts that have been addressed by Google, but overlooked by the general public:
As per Google, a consolidated account structure will not specifically improve performance from automated campaigns. However, if a consolidated account structure makes it more efficient for you to manage, then this will indirectly result in improved performance.
Our team uses as much consolidation as possible. It provides:
With consolidation, we have not forfeited any benefits of segmentation. Google’s reporting features are now advanced enough that you can find any data or spot any trends that we would have otherwise wanted to see if we were ultra-segmented.
As a rule of thumb, campaigns should be segmented based on budget requirements, and ad groups should be as segmented based on what logically makes sense given available ad copy and targeted landing pages. If there is an advantage to be gained with more relevant ad copy or a more targeted landing page, then it makes sense to segment out those keywords.
While we prefer consolidated accounts, there are times that we inherit an account that has previously been segmented at a very granular level. In this event, we won’t just shut everything off and build it the way we like.
Plus, automation can work with any account structure, remember?
Ultimately we’ll consolidate efforts to take advantage of the benefits listed above, but drastic changes in a short period of time are generally not recommended.
A good rule of thumb for new accounts that are overly segmented:
You should become extremely aware of how your time is spent during client meetings and physically working on accounts.
If you are spending the bulk of your time focusing on physical changes that are being implemented inside of an account, you’re likely not using your time effectively. It’s also likely that you are not achieving profitable growth at the same rate that you could be.
Futile optimizations and trivial projects have traditionally overwhelmed the PPC manager’s workload. Many of these tasks are, for lack of a better expression, a waste of time.
All that time that used to be spent over-segmenting campaigns, optimizing for Quality Score, adding negative keywords and making manual bid adjustments can now be reinvested in larger scale projects that will actually achieve profitable growth.
While Google automation will work with any account structure, you should use your new understanding of Google’s ad auction to make your own intelligent decisions.. Remember, we’ve exchanged the funnel for the framework. In a funnel, everyone enters and exits the same place. In a framework, there’s infinite opportunity for your creative applications, as long as you understand the inner workings of the machine, leveraging that knowledge to your advantage.
As for me? I'm searching Google for a local shop that can re-grip the rest of my golf clubs. Wish me luck.
We'll get back to you within a day to schedule a quick strategy call. We can also communicate over email if that's easier for you.