Over the last few years, I’ve been asked the same question again and again:
If digital advertising is leaning more and more into automation, what will agencies actually do? What value will they bring to their clients?
The short answer is that modern agencies (or in-house teams) must adapt to a new role in which their primary focus is to manage the automation.
In my book, I analogize the role of a marketer to someone riding on horseback: the horse is capable of navigating a course without running into walls or aimlessly off a cliff; the jockey can therefore focus their attention on the horizon, using haptic signals to get a feel for where the horse is headed, whether it’s speeding up or slowing down, or whether it’s starting to feel anxious.
For what it’s worth, I’m terrified of horses.
And while I love this analogy, it’s not practical enough to answer the original question. That sounds great… but what do you actually do?
When I started in this industry, agencies were valued by the number of changes they made to an account. The Google AdWords Change History Tool was the best proxy for agency value. There was a time long ago when AdVenture Media billed clients based on the number of keywords, ad groups, and campaigns in an account. This would be an arbitrary and useless standard in the industry today.
Today, less is more. We make far fewer changes to each account, but each change holds significantly more weight. The decisions we make today are much more complex; ten years ago they were mundane: this campaign receives a higher conversion rate after 5 pm, so we should increase the bids by 15% after 5 pm.
Years ago, you could reasonably optimize a Google AdWords account while multitasking—in 2015, I listened to the first season of the Serial podcast while performing routine account work. You can’t do that anymore.
Modern optimization requires intense focus, creativity, and careful consideration. The tasks are no longer mundane; we are attempting to analyze complex marketing data while also training fickle algorithms to make decisions on our behalf.
… and horseback riding can be dangerous.
Here’s an inside-baseball example of what our account management teams actually do.
Last week, we sent the following email to a client regarding a single change in the account: we switched the bidding strategy for one campaign from Maximize Clicks to Target ROAS.
This one change was carefully constructed. It was based on the account data as well as the collective experience of multiple people on our team. Not only did we have sufficient evidence to support this decision, but we also knew what to expect after we made that change.
This is an ecommerce business that has struggled to scale its Search efforts beyond the very bottom of the funnel. The majority of our Google Ads budget is dedicated to Performance Max; specifically, the Google Shopping placements within the Performance Max campaigns.
With a small portion of the budget, we tested Middle-Funnel Search campaigns in an effort to profitably grow top-line revenue.
The campaign originated under a Maximize Clicks bid strategy for two reasons:
The test ran for one month. We monitored the campaigns but didn’t make a single change for 30 days.
Here’s how we explained it all by email to the client. I hope this helps articulate what we actually do.
Last month, we started an initiative to ramp up our Search campaign efforts. Historically, we've struggled to retain profitability as we've attempted to scale Search, but we're determined to make it work.
Search campaigns focused on Bottom-Funnel terms have always been a part of our strategy; however, we wanted to test Middle-Funnel terms.
We launched this campaign on March 6th (30 days ago). The quick stats:
Here are the top 10 Search Terms, ranked in order of total clicks:
Only #8 generated an actual sale. #5 generated an add-to-cart, but not a sale.
Most of the revenue is attributed to Search Terms outside of the top 10 in total traffic. Ironically, a Search for: [redacted] generated a sale worth over 13K in revenue.
Revenue was also earned from the following Search Terms:
This proves the original hypothesis that we can convert Middle-Funnel Search Traffic. The concern is whether we can do so profitably...
IMPORTANT: Until now, the campaign has been running on a Maximize Clicks bidding strategy.
We just changed this to tROAS... I'll explain why in a minute.
The Maximize Clicks strategy was set to gain as much traffic as possible with a low/stable cost-per-click.
We also wanted to force Google to explore Middle-Funnel search terms. Historically, Search campaigns did not profitably convert Middle-Funnel terms; therefore, if we started with Target ROAS bidding, the algorithm would have likely turned its attention toward Bottom-Funnel searches, where it was already confident it could generate a positive return. This, though, would have defeated the purpose of the test.
We also set a bid cap of $3.75. Interestingly, our Actual CPC has been $3.38. For a week during the middle of March, our Actual CPC was above $3.60.
We typically see a wider delta between our Bid Cap and our Actual CPC. The small delta (less than four percent) suggests we were bidding very close to the market price for the clicks we received. If we had raised our Bid Cap, we likely would have entered significantly more auctions and gained significantly more traffic.
We also would expect the quality of that traffic to increase. In general, higher-quality traffic costs more to acquire. A rise in CPC generally correlates with an increase in time-on-site, conversion rate, AOV, or a combination of all three.
The campaign had a 27% Search Impression Share. This suggests that, with higher bids, we could have doubled or even tripled our total impressions/clicks. However, Search Impression Share does not factor in the auctions that you did not enter entirely. There are likely many auctions where our $3.75 Max Bid prevented us from exceeding the Ad Rank Threshold.
For comparison's sake, the Average CPC on our Bottom-Funnel Keyword Search campaign is above $10—the "market price" for these keywords is much higher than $3.75—and the time on site and conversion rate here is much greater than we've seen in our Middle-Funnel campaign with lower bids.
Therefore, we are reasonably confident that an increase in bids for our Middle-Funnel campaign will not just generate much more traffic but also drastically increase the quality of traffic that we acquire.
As such, we switched the campaign from Maximize Clicks to Target ROAS earlier today.
Will this improve ROAS right away? Probably not. We need our Conversion Rate to increase at a greater rate relative to our CPC (holding AOV constant) for us to make gains in ROAS.
We assume this will improve ROAS over time.
Also, since our historic ROAS in this campaign is just 0.71%, we're setting our starting Target ROAS to 0.85%. We do this so that Google continues to optimize within similar auctions to what we've recently been bidding. If we were to set an aggressive 1.50% ROAS target, then Google would likely exclusively bid on Bottom-Funnel terms. We must ease into it and train the algorithm to find the profitable needles in the massive Middle-Funnel Search haystack.
In the short term, we expect an increase in on-site engagement metrics (time on site/pages per session). We'll take our first measurement after one week.
We expect CPCs to rise... possibly above $10. If it reaches $15, stays above that rate for three consecutive days, and does not generate a better ROAS, we will want to adjust our targeting, lower the budget, or increase ROAS Targets.
Ideally, CPCs would stay under $10... but we'll see.
Total clicks could rise, fall, or stay the same: it all depends on how confident Google's algorithm is under the new bid strategy. If we see a sustained increase in total clicks, this will suggest that Google is confident that it can quickly achieve this 0.85% ROAS at scale. That would be ideal.
The more likely scenario is that we'll have volatility in traffic from this campaign. We might go up one day, then down the next, but the algorithm will likely settle in after about ten days.
Again, this is the next step as part of a longer process to profitably scale our Search campaigns. We don't expect this to be 100% profitable this month, but with more testing, we can continue to improve the quality of the traffic that we bid on, as well as Google's understanding of the kinds of signals that help lead to profitable conversions. As we fine-tune that effort, ROAS should follow suit.
We'll keep you posted on the progress. Let me know if you have any questions.
The second answer to the What do agencies do? question is: manage client expectations. We’re clear that this campaign will not be profitable in the search term; however, it does not suffice to simply say that “it will take time.”
Instead, we attempt to convey other metrics that we can track in the short term to get a sense of how the new test is going. In my horseback analogy at the top, I referred to these as haptic signals. Metrics like CPC, time-on-site, etc. will help us understand if there is hope, or if we were dead wrong from the onset.
We also use second-order logic to help describe what decisions we might make given different possible outcomes. We have no crystal ball, but we can at least predict some of the possible outcomes and have a predetermined game plan for each.
At the time of this writing—only two days after we made this change—our initial hypotheses have been correct:
We only have two conversions, but technically our conversion rate is up 87%! (I wouldn’t read too much into this… it’s too small of a sample size).
Clicks are down 60%. We were originally unsure about what would happen to clicks, but this is the most likely outcome; after all, moving away from a Maximize Clicks strategy would naturally take you away from generating as many clicks as possible…
Most importantly, we still don’t see any Bottom-Funnel search queries in the Search Term report. We’re hoping this will continue, as this would prove that Google’s algorithms are confident in its ability to drive conversions for Middle-Funnel traffic within a Target ROAS framework.
All of this … just for one tiny change logged in the Change History Tool.
Less is more, and as a result, experienced PPC agencies are more valuable today than they have ever been.
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.
Woodmere, NY, 11598
View On Map