Writing ads that converts the most clicks to purchases can be one of the most impactful optimizations you can make to your entire Google Ads. The challenge is that the ad that you wrote when launching the campaigns (where you had a lot of different things to focus on) might not be the best ad.

And with so many other things to optimize you might never get into the habit of testing your ads.

That’s a major mistake though.

When you find better performing ads you’re able to improve all your Search Ads with one click. It sounds too good to be true, but effective ad testing processes can yield big results (until a certain point).

But testing ads individually across tens or hundreds of campaigns is not only simply impossible but an incredible waste of time.

The solution is multi-ad group ad testing.

In other words, test the same ad theme across several ad groups and compare your results. This will give you more data faster and you can conclude on ad tests.

This article is just as much a primer on how to think about ad testing as it is a guide to implement multi-ad group testing.

Batch Your Thinking

The way many PPC specialists work with ad testing is that they open an ad group and then come up with ideas to test. As with anything you do casually, it’s rarely the best way to approach ad campaigns. Plus, it’s not efficient at all.

I first shared my process for coming up with ad testing ideas back in 2015 on SearchEngineJournal.com, and it’s really this process that I’m describing here:

  1. Set aside a good chunk of time. Focus on nothing else than coming up with ad testing ideas.
  2. Brainstorm potential ad themes (unique selling proposition, benefits, features, etc.).
  3. Write out your various ad themes.
  4. Write themes specifically for Google Ads (character counters work awesome here).
  5. Create 10-20 ads to test.
  6. Test the best 2-3 ads at a time.
  7. Rinse and repeat every quarter to build up an inventory of ads.

The idea behind batching your work with ad testing—and really anything related to individual Google Ads tasks—is powerful. You get to do nothing else for a set amount of time other than preparing for ad testing.

For PPC specialists, the hardest part is coming up with the actual ideas to test. We can upload ads quickly. But the trick is to upload something that will actually make a difference.

By batching the time you spend on ad testing you specifically carve out time for each step. In this case the step of coming up with meaningful ad tests is of the biggest important.

It means that you will consistently have a bank of ideas to test next.

Note: By batching, I mean that you spend time once a quarter coming up with ad tests. You don’t try to do it weekly or monthly. No, you spend a great deal of time once every few months to come up with what you need to test. And then you implement it.

It takes time to get into the zone to come up with ad tests that are meaningful. Testing “Buy Now” vs. “Buy Online” is the epitome of inefficiency. It doesn’t matter.

However, testing the impact of including pricing, savings, or features in your ads— now,that’s meaningful!

Find Ad Themes That Perform Best

The goal with ad testing today is not necessarily to find the individual words that convert the best. It can be a lost cause to try to test individual words. You can do that when you know whether “price” vs. “savings” work the best.

So, instead of trying to test words against each other, test themes against each other.

Here are some typical themes that we test for eCommerce clients at SavvyRevenue:

  • Savings
  • Quality
  • Newest
  • Selection
  • Price
  • Feature
  • Benefits
  • Seasonality (being current)

Let me give you a couple of examples of what I mean by themes:

Different Ad themes you can test

Good chunks of the ads are the same, but the meaning changes so you can test what resonates best with people.

Know that different messaging will perform better or worse at different times of the year, and it also depends on your competitors. So plan for that.

Should You Copy Your Competitors’ Ads?

Yes and no.

It depends on how good you are.

If you are good at PPC, then you should be able to come up with better ads than your competitors, and therefore, win the race.

If you’re not that good, then you can just copy what seems to be working for your competitors and live with campaign performance that equals your competition.

Our advice: Test the opposite of what your competitors are doing.

Test Ad Themes Across Several Ad Groups (aka Multi-Ad Group Ad Testing)

The whole idea of multi-ad group ad testing is that you initiate the same test across multiple ad groups at the same time. You take the aggregated data across all the ads and find the winner.

There are two main benefits of this:

  1. You get statistical significance much faster
  2. You implement new, better performing, ads across more ad groups, which leads to better performance on bigger parts of your account much faster

To do ad group testing you need to make sure that you leave the same ad assets (e.g., headlines and descriptions) across all ads and only change the assets that you want to test.

Here is an example of this kind of multi ad group ad testing. The champion ad is the current ad, and the challenger ad keeps most of the same assets, but changes a few things we want to test:

Test your ads by changing different assets

The “Testing Headline 2” and “Testing Description 1.” are the changed parts.

What About Keyword Relevance? We Can’t Just Run the Same Ad Across All Ad Groups

So, we come back to the idea of testing ad themes.

The “Headline 1” theme includes the keyword and some sort of modifier (e.g., “Buy Patio Furniture or Patio Furniture Online”). It does this across all the ad groups, so even though the value is different, the theme is still the same.

The same goes if you have three different categories with three different savings offers:

  • Outdoor sofa: Save 25%
  • Hammocks Save 20%
  • Outdoor furniture set: Save 40%

We are still testing whether or not having the percentage of savings mentioned in the ad performs the best. The individual savings percent doesn’t really matter (unless it’s an irrelevant amount like 2%).

Recommendations: Use Ad Variations in Google Ads to Perform Ad Testing

A feature that was bad when it first came out was Ad Variations. It didn’t really work, and, therefore, many of us skipped using it. That’s generally a challenge with new features from Google. We already have processes in place, so when Google introduces a new way of doing it, then it better be good.

For brand new PPC specialists wanting to perform ad tests across multiple ad groups, I do find that Ad Variations is the easiest way to get started.

It works just like the changes you would normally make in Google Ads Editor. Here’s how to set up an ad variation in the Help Center.

So, why would you use any other way of testing ads than Ad Variations?

Technically, there is no reason not to use Ad Variations. It’s just that us PPC people get used to a way of doing things, and if it ain’t broken, don’t fix it. Especially, when it comes to new features from Google, as they are often in beta and buggy when they first come out.

BIG NOTE: Ad Variations is the only way to test ads if you use Smart Bidding.


Ad Variations is the only way to test ads if you use Smart Bidding.


Otherwise, Smart Bidding will override your ad delivery settings to “optimize”. This means Google will primarily show the ad that it thinks is performing the best. You might think that this is the whole point of an ad test, and that’s true.

But Google can sometimes determine an ad test winner within just a couple hundred impressions. That’s insane. Unless it can predict performance, that would be insane across multiple languages.

How to Determine Ad Test Winners with Multi-Ad Group Ad Testing

We use three different ways of determining ad test winners:

  1. Ad Variations
  2. Reports in the Google Ads interface
  3. Adalysis: The ad testing platform

I’ll go into details about the three options below:

Ad Variations

This one is really self-explanatory. You get a nice interface that actually tells you which ad test is the winner:

pasted image 0 17

Just make sure you choose “View by Campaign” otherwise, you won’t see the aggregated performance below. You do get the aggregated performance at the top, but it might take a couple of days.

Reports in the Google Ads Interface

This is the easiest, fastest and cheapest way to determine ad test winners if you are not using Ad Variations.

The reporting part of Google Ads doesn’t get mentioned enough, but I use it a ton.

You can use two different methods to measure performance directly in Reports:

  1. Based on the ad asset that you are testing for
  2. Based on the Ad Label

Here’s how to do it:

  • Choose the Table format
  • Add the columns (data) you want to see
  • Add the asset you are testing (e.g., the description or headline)

The final report can then look like this:

pasted image 0 16

Make sure you only choose to see the data for the ad assets (e.g., headlines or descriptions) that you are testing. Otherwise, you will not get the aggregated data based on the right assets.

Ad Labels in Reports

The other option is Ad Labels.

Selecting the headlines you want to compare works great when the assets are exactly the same, but it doesn’t work when you have different assets under the same theme. This could be different savings for different ad groups. Your data would be split on multiple headlines, and the report would not be useful.

That’s where labels come into play.

You have to ensure that you add the labels on the ads when you create them:

pasted image 0 18

And then you switch out the “Headline 2” asset with the label in the Report:

pasted image 0 21

It’s super easy, and fast to do once you get the hang of it.

Adalysis: The Ad Testing Platform

Adalysis can also be used. We are using the tool less and less because we are using feed-based ads, so our ad testing process is streamlined already. We can get the insights we need by using reports in the Google Ads interface.

I have described Adalysis in more detail in our article on Google Ads tools, so I will not go into more detail in this article.

Bonus option: You could use SuperMetrics (affiliate link) to pull the labels or headlines into a Google Sheet where you could apply formulas to determine performance. We don’t need to go to that extent, although our Director of PPC would love to set it up 🙂

What to Do When Your Tests Are Inconclusive?

Let’s say that you are testing a Savings message against a Quality message.

You notice that in four out of ten ads your Savings message performs best, but overall your Quality message performs best.

What to do?

That’s the whole idea with multi-ad group ad testing. That you aggregate the data across multiple ads and find the overall best-performing ad message.

You will most likely always have a couple of outliers here and there, but that’s ok. Don’t worry too much about them.

One thing to keep in mind here is if you are testing ad messages across completely different keyword sets.

Like the keywords “KIDS’ NIKE AIR FORCE 1 LOW CASUAL SHOES” and “running shoes”. These searches appear at different parts of the buyer’s journey so people will react best to different ad messages.

So make sure you are testing messaging on similar keywords.

Metrics for Determining Ad Performance

Most important Ad Metric Determine the ad performance

There are a couple of different opinions on what metrics to use for determining winners of ad tests:

  1. Some believe that all an ad is responsible for is getting clicks, making CTR the most important metric.
  2. Some believe it should be ROAS.
  3. Some believe it should be a custom metric like revenue per 1000 impressions, which would be equivalent to a mix of CTR, Conversion Rate and Average Order Value.

I believe that an ad is equally responsible for making users click on the ad, and also converting afterward.

The messaging you use in your ads must find the best balance of both. As an example, you can write “GET FREE TVs” and get a massive CTR, but no sales.

We like to use conversion value, BUT the ROAS must be above our target.

So if we have the following results:

Winner Ad against Loser Ad

Even if our ROAS target is 400%, we will go with the ad that produces more revenue.

It’s simplifying it a lot, but you shouldn’t overcomplicate these things. It will never be an exact science anyway.

If you’re tracking profit, pay attention to whichever ad produces the biggest amount of profit.

Ad Testing vs Recurring Promotions or Seasonal Messaging

Can you test ads when you have a new promotion every couple of weeks that require you to update the ads?

Seasonal Sale influencing Ad testing Ad testing vs. recurring promotions

If you are running a lot of promotions, then it can seem hard to find a good time to test ads because you are changing them all the time.

If that’s the case then you can test the following:

  • Test to see if ads that don’t mention the promotion works best
  • Test to see if mentioning %-savings, $-savings as the only words works best

The idea is that you find out two things:

  • Find out if it makes sense to run promotional-heavy messaging
  • Find out the best way to have promotional messaging in your ads

Who knows, it might be enough to just run your Promotion Extension.

But just know that you can still run ad tests even if you are changing your ads a lot. Using labels to aggregate performance across ads will help you get the right information faster.

It’s OK to Run Different Messaging At The Same Time

By definition, you will lose revenue if you are always running ad tests.

Because as you are running new ad tests, there will always be a chance that your new ad is performing worse than your existing ads. That means you will get less revenue out of the potential searches.

So, it’s perfectly fine for you to decide that you will not be testing any new ads for an amount of time. This might mean running with two different ad themes at the same time:

Ad with selection message vs. Ad with Sales message


Ad Testing with Responsive Search Ads

Google released Responsive Search Ads a while back, and they are meant to be the end-all for ad testing. That’s partially true. They work by adding X amount of assets (e.g., headlines and description lines) to a Responsive Search Ad.

Google will then test these assets against each other to find the best-performing assets.

pasted image 0 19

Credit: Wordstream

On top of that, Google will try to determine whether Asset A performs best with some people and Asset B performs best with other people. Then Google will show the highest performing asset for specific people.

pasted image 0 20

Credit: Wordstream

Just like Facebook Ads.

There is one big difference between Facebook Ads and Google Ads. In Facebook Ads, you can set an objective that Facebook optimizes for. This hardly works in Google Ads if at all. So it’s up in the air what Google actually defines as “highest” performance. I might not be interested in a high CTR, but instead in overall conversion value.

Or, if an asset combination has a higher conversion rate, but lower ROAS, which asset combination is Google determining as the best?

If we could choose what Google should optimize for, then I would be a lot more open to RSAs. Until then, I don’t like them.

Test Themes to Find The Best Wording

But if you have to run RSAs (and the rumor is that ETAs will be replaced by RSAs completely) then I suggest running ad themes and testing them against each other.

So, instead of simply running an ad test that is Ad A vs. Ad B, then run Asset Group A against Asset Group B.

  • In Group A you only add savings messages.
  • In Group B you only add seasonality messages.

Once you have found the best-performing assets across the different themes, you can try to combine them into a single Responsive Search Ad that Google runs with its own algorithms.

But just know that savings-messaging often outperforms everything else, so Google might go for this asset every time—even when it’s not ideal for you.

And again, we come back to the point: what is Google actually optimizing for?

Ad Variations Also Work for Responsive Search Ads

It works exactly as with Expanded Text Ads, so I will not go into further detail.

Ad Testing is Crucial for Any Google Ads Account

It’s crucial that you identify what kind of ad message works the best at what times of the year. Is it best to run a promotional message or a new selection message?

Is it best to run a promotional message highlighting percent savings, price savings, or something completely different?

If you just go with what you have always done, or what you think works best, then it’s time to start testing. When you have your keywords and bids dialed in, ad testing should be your biggest focus area.

If you need one place to start, then choose your top campaign and test the ads. Once you start seeing better ROAS from these ads, you will be addicted to the constant improvement and momentum you can create.