In this case study, we wanted to demonstrate our methodology and findings for a test we ran to determine whether manual bidding or target impression share was the best bidding strategy for this particular account. 

One of the reasons I wanted to write this up, was to demonstrate the sort of tests ZATO does on a regular basis inside of Google Ads accounts. In this particular instance, the business has been a client of ZATO’s for over ten years. We hope this demonstrates the sort of care and ongoing testing and analysis we invest in even our oldest clients. We do not simply take clients on and forget about them, but regularly invest in testing strategies and optimizations such as the following. Frankly, I wanted to walk you through that so you could experience that for yourself. Even if you end up disagreeing with our findings, my hope is you can respect our desire to investigate a potential solution within a campaign that has been run for many years.

‍

Okay, with that in mind, let’s look at some caveats:

‍

Statistical Significance:

While we want to pursue ideal testing and data analysis standards, the reality of PPC, especially with smaller budgets like this client, is that an ideal level of statistical significant often does not occur within a reasonable enough timeframe to make a decision within the account to allow for ongoing management and optimizations. In other words, there is a reason managing SMB Google Ads can be so tricky!

Because of this, I have found over the years that we need to pursue as much data as possible, then make the best decision we can based on that data and the additional context we have in the account, and then move on with life. 

‍

Individual Account Results Will Vary:

‍I find it crucial to emphasize that this is a single example, from a single campaign, in a single account of a test being run. This should NOT be viewed as saying more than it is. We are not here making a claim as to the ideal bidding model for all brand campaigns in every account. We are simply walking you through one account where we tested something and found a conclusion that made sense to us. Your accounts may differ because many things influence how you should think about setting up an account in Google Ads, even with your brand campaigns. If you have advertised for multiple brands, you understand that brand bidding strategies can vary widely based upon the nature of that vertical’s competition, the average cost of your brand’s CPCs (and more specifically, what that does to profitability), and a variety of other factors. So, learn from this case study, but then run your own tests to come to your own conclusions about what is best for each individual account you manage. 

With that being said, here is the test I (Kirk) ran in an account I personally manage: 

‍

‍

A Single Account Case Study on Brand Bidding with a Target Impression Share Test

‍

TEST PURPOSE: 

The reason I ran this test originally, was to determine whether Manual CPC is still the most effective way to manage Brand search campaigns. That quickly turned into a test between the Manual CPC and Target Impression Share bidding models.

The reason I have traditionally used Manual CPC for Brand search, is because Brand search tends to be limited in its scope of what it will match to, and more stable in terms of ongoing click costs.

In other words, aside from a rise over time, Brand doesn’t typically change drastically in competition and costs (at least on average for clients we tend to manage, certainly it does in some instances). Because of that, I often like to force Exact (and often Phrase) brand terms into their own campaign, and then tightly control the specific CPCs that we will set on those terms. This also includes managing all of the additional bidding modifiers, by the way, for maximum efficiency. We’re talking full-on old school. Checking in regularly on devices, locations, audiences, etc and making adjustments as we get enough data to prove a change should occur. Again, a key reason I find Manual CPC still works well in 2024 for Brand bidding is because of the stability of those search term costs and exposure potential (a variable not taken into account, would be a brand who has significant spikes due to things such as CTV spots, Shark Tank episodes, etc).

‍

Why go to all of this trouble with Manual CPC in the first place?

As I have tested smart bidding on Brand terms, I have consistently found that Google overbids for click volume that would have come through regardless. In other words, we have lower profitability on clicks that would have come regardless. I found I could keep efficiency higher while pushing for more traffic (in this account we tended to stay around 92% impression share on manual bidding for brand terms, which I am happy with).

‍

Why Target Impression Share?

In the past, I have experimented with conversion focused bidding strategies, which were unable to beat out my Manual CPC bidding. Part of that, is again, the fact that Brand terms tend to be so stable in many ways, which decreases the value of having a predictive model trying to make micro-adjustments. In other words, the gap between my ability as a human with a CPC cap and an advanced algorithm is a lot more narrow when everyone acts like we all know they’re going to act!

‍

Because of that, Brand is more about maximizing impression share. That is, because we know we are always going to be very profitable on brand terms in this specific account, we want to make sure we stay in as many auctions as possible. Since Brand auctions are so affordable in this account, we want to make sure we don’t lose out on any chance to convert people who are searching for their brand, while keeping things as efficient as possible since we don’t want to overbid on Brand (remember, Brand is simply people who have already been marketed to, so we need to ensure we keep our Return on Ad Spend very high so we don’t lose money through the entire customer journey).

‍

‍

TEST METHODOLOGY:

In order to run this test, I used the Experiments option within Google Ads. This has evolved over the years, and is really quite a handy tool for running, well, experiments! For this one, I ran into some trouble initially because I had some old paused ads (Google wouldn’t allow me to begin an experiment), but once I removed those I was able set up the experiment and put it into play. 

Within the experiment, I set the test to the Target Impression Share bidding model, and then I aimed at a 92% Target Impression Share (identical to the average impression share from the past 30 days of my Brand campaign, so we could have some parity) and I set a max CPC cap of around $2.51 (which is quite a bit higher than the traditional average CPC in this Brand campaign, but I wanted to give some room for growth if the machine saw it without over-bidding). 

I ran the experiment from February 3 - March 11 before finally ending the Experiment. 

Which won? 

‍

TEST OUTCOME:

Here is where things get interesting. 

Manual bidding technically won my experiment, but I turned on Target Impression Share after digging into the results.

Wait, what??? Why would I do that? 

Well, much to my personal pride, my complex Manual CPC bidding setup (remember, I was utilizing bid modifiers on various things such as device, location, and audience) actually performed slightly better than the Target Impression Share (TIS). 

The experiment collected 4,235 clicks in total, with the TIS campaign receiving 1% MORE clicks than the Manual CPC campaign. That being said, the TIS campaign received -2% FEWER conversions, and overall was slightly (-1%) less profitable than the manual CPC bidding campaign. 

‍

So in one sense, manual CPC won, right???

Well, keep two things in mind.

1. This is a brand campaign. The slightest level of efficiency is negligible compared to the amount of work required to continue managing this particular brand campaign with manual CPC bidding. 
2. I didn’t make adjustments to the TIS campaign. So by tweaking some of the guidelines, let’s say increasing or decreasing the max CPC, and/or increasing the desired Impression Share target (I have since done both of those things), we might actually be able to eventually tweak better performance out of the TIS campaign than the manual CPC campaign. 

‍

In other words, the difference was so minute, that enabling a smart bidding solution so I can invest my time managing this account into other testing ideas or optimizations rather than managing what it takes to keep that manual CPC running strong, is clearly the best option here. 

Like I said in the beginning, your results may differ, so run your own tests for sure. But in this case, it made sense to lean into Google automation. 

Hope that was informative to you, and keep testing!

‍