Brad Geddes / PPC Geek
Official Google Ads Seminar Leader.
Author of Advanced Google AdWords.
Co-Founder, Adalysis.
(703) 828-5811‬
Brad Geddes's Theories on Marketing PPC Ad Testing: Discovering what to test and how to...

PPC Ad Testing: Discovering what to test and how to do it

Why You Must Test Ads

Testing ads is an essential part of any marketer’s life. Your offers change, your competition changes their offers, people change how they interact with messages, and without ad testing, you will quickly fall behind the competition.

By testing ads, you are ensuring that you are constantly trying to improve your marketing. When we look at accounts that do a lot of ad testing, especially in competitive markets, we generally see one of two trends emerge.

The first trend is where you see improvements; and then your account declines, then you test more and see improvement, and then you see declines.

This is common when your competitors are constantly changing ads or when the market conditions are always in flux; for instance, in the mortgage industry, there are market changes due to mortgage types, interest rate changes, and parts of the year when people are more or less likely to move.

In this instance, you need to start testing before you see the decline so you know the next message to use for your ads.

The other common trend is a general improvement over time. This usually occurs in accounts that have done very little testing the past; so gains are easy to be had by just getting started in their ad testing.

Not every test will be a winner; so, don’t expect constant improvement. On occasion, you’ll try a test, it won’t win; and while the worse performing ad is running; it will drag down your account for a while. This is why you need to monitor your ads (or use an automated system like Adalysis to monitor your tests for you) so that once an ad is a loser; you can pause it and force the impressions into your top ads.

Where to Start Testing

In ad testing, you can test anything:

  • • Geographies vs non geographies for local companies
  • • Prices vs discounts vs free shipping for ecommerce companies
  • • Branded ads vs non-brand ads for non-brand keywords
  • • Keyword insertion vs static ads
  • • Features vs benefits
  • • Different calls to action
  • • Etc

The possibilities are limitless. Since the options are so varied; let’s look at where you should start ad testing.

When we examine ads; these are the components:

  • • Headline 1: Almost always related to the keyword or ad group’s purpose
  • • Headline 2: The 2nd more prominent part of the ad; and is commonly a benefit or call to action
  • • Description: A long description, in smaller font than the headlines, that talks about the products & services you offer and why someone should want to learn more.
  • • Paths: A description of the type of page a user will see after clicking your ad

The two headlines are by far the most prominent parts of the ad. Headline 1 is generally related to the keyword or ad group; and while testing it is important; you don’t have the creative freedom you do with headline 2 since you need to show relevance to the user in an ad; and that’s generally headline 1’s job.

That means, headline 2 is the best place to start your ad testing. While all the ad components should be tested, the headline 2 is the best place for your initial ad tests.

Ad Copywriting Elements

When we look at ideas for ad testing; there’s a few pieces of jargon we should define that go along with ads:

USP or Unique Selling Proposition: This is what is unique about your company. It can be a true statement, a marketing slogan, or what you want people to know about your company.

  • Slogans
    • • Saddleback leather: 100 year guarantee; They’ll Fight Over It When You’re Dead”
    • • FedEx Corporation: “When it absolutely, positively has to be there overnight.”
    • • Target: “Expect More. Pay Less.”
    • • Enterprise: “Pick Enterprise. We’ll Pick You Up.”
  • • Facts:
    • • Everything taxes: local, state, federal and even international
    • • Plumbers: We’ll be there anytime, anywhere, day or night.

Features are true facts about a product. While they aren’t sexy, if a user is comparison shopping, they just want the facts.

  • • Price
  • • Screen size
  • • Memory
  • • Square feet
  • • Etc

Benefits are what someone gets from using a product or service. A product or services benefits is the reason we pull out our credit card to make a purchase. Benefits can be hard for some people to think of; but there’s an easy trick to this.

Features are easy to list out. So make a list. For instance, my computer is:

  • • 2.8 lbs
  • • 16gb memory
  • • 512 SSD drive
  • • 9 hour battery life

Unless you’re comparison shopping; that’s a boring list of items. There’s no context to what that means. Your job is to build the context – the story – with benefits. To do so, just ask yourself ‘so what’ and finish the sentence.

  • • The computer weighs 2.8lbs so you can carry it around all day without back and shoulder pain.
  • • The computer features a 9 hour battery life so you can be outlet-free all day. No more hunting for an open power outlet.

CTA: Calls to action tells a user what you want them to do. When your landing pages features that same call to action; you can often see higher conversion rates. Example calls to action are:

  • • Call Now
  • • Shop Now
  • • Download our whitepaper
  • • Etc

See another trend? They are boring. A good call to action combines a benefit statement with a call to action. Consider this:

  • • Subscribe to our newsletter
  • • Receive powerful marketing tips

In the end, the user will get an email sent to their inbox. However, no one wants yet another email. All marketers want powerful marketing tips. By combining benefits and calls to action you can help stand out from the crowd.

Single Ad Group Testing

Testing ads is easy – just have two or more ads in an ad group. Testing ads within an ad group will let you find the best marketing message for the targeting in that ad group. The targeting could be keywords or audiences.

The hard part is coming up with ideas to test and monitoring the results.

After analyzing a large number of ads; we’ve devised a few templates to help you start testing.

When you first start testing, you want to vary your headline 2 with a call to action, consumer benefit, or USP. You can simply follow these three templates to get you started down your path to testing.

There are some downsides into testing ads within an ad group:

  • • The ‘best ad’ might not apply to other ad groups
  • • If your account doesn’t get a lot of data, then you might never achieve statistical significance
  • • It can be highly time consuming in enterprise sized accounts to write ads for every ad group

For your brand terms and your top terms, you should test within an ad group. For other targeting types, you can use multi-ad group testing as this will aggregate the data across ad groups together.

Multi-Ad Group Testing

Multi-ad group testing is when you test a pattern or line across multiple ad groups at once.

For instance, you might be wondering if you should use prices or discounts, offer free shipping, or focus more on benefits or USPs.

When you have these high level questions, multi-ad group testing can give you answers.

With multi-ad group testing, you will create at least 2 ads in an ad group that share similar characteristics for every ad group you want included in the test.

For example; you might want to test two different calls to action to see which works best for your non-brand terms. To do this, you will have two ads in every ad group for those terms:

  • • The headline and description will still be related to that specific ad group
    • • Ad 1: Use CTA 1 in headline 2
    • • Ad 2: Use CTA 2 in headline 2
  • • The headline and description will still be related to that specific ad group

Then you will replicate this across all the ad groups in the test. What you end up with is 2 ads in all the ad groups where the headline 1 and description are still related to the ad group; but the headline 2 is consistent across all the ad groups in the test. Once you have a winning call to action; it will be appropriate to all the ad groups in that test.

When you want to ask big marketing questions or find insights in your data; multi-ad group testing is wonderful.

To find which line does best, you can use software like Adalysis or run pivot tables to examine the aggregated data.

There are some downsides with multi-ad group testing:

  • • The ‘best’ ad line might not be the best for every single ad group as you are looking at aggregated data
  • • You must be consistent in your usage. These two lines are different (casing and punctuation):
    • • Call Us!
    • • Call us.

For enterprise accounts, accounts with little data, and market research questions, multi-ad group testing is the best way to test across your account.


For example, Adalysis won the ‘Best PPC Management Software Suite’ at the US Search Awards and we wanted to see if showcasing the award in our ad copy would increase our conversion per impression metrics.

There were a few ways we wanted to test the hypothesis that showcasing the award would provide better better results:

Single ad group tests:

In our account, we have several search ad groups. 3 of these ad groups lead to more than half of our conversions; and the other 50+ ad groups (long tail) contain less than half of our conversions.

This means we wanted to find:

  • • The absolute best ad for our top 3 ad groups.
    • • This relies on single ad group testing
  • • The best ad headline 2 (as the headline 1 is related to the ad group’s purpose) for the long tail ad groups.

For our top ad groups, we created a new ad to test our current winner and had these two ads in the ad group:

We found that for two of the three top search ad groups; not focusing on the award did better than focusing on it. Presumably, these ad groups are triggered by some specific searches and those conducting these queries have done a lot of research and know the ins and outs of the industry; and therefore, focusing on action did best in 2 of these 3 ad groups.

Multi-ad group tests:

We ran three sets of multi-ad group tests: one for search and two for display.


For search, we have ads that are setup like this:

  • • Headline 1: related to the ad group (such as Optimize Quality Score; Diagnose Keyword Conflicts, etc)
  • • Headline 2: our playground to test new calls to actions, benefits, and statements
  • • Description: related to the tool and the ad group

Here’s an example of our ad for Quality Score:

As all the ads are slightly different; we just wanted to focus on changing the headline 2 across all these ad groups and then see which line performed best. Here’s an example of the ad for the Quality Score ad group.

We created a second ad in all of the ad groups using this new line. So now we can test these two lines across all the ad groups:

  • • Take a Free Trial
  • • US Search Awards Winning Tool

We found that focusing on the awards for these ad groups did better than a call to action for a free trial.

So now we know where to focus on trials versus awards for our search ad groups.


We run a large number of display ad groups and campaigns and spend a lot more marketing dollars on display than search.

With display campaigns, many users aren’t actively researching the product; so you need to pull their attention away from their current action and have them focus on your offer.

What we wanted to know was two-fold:

  • • For similar audiences; people who have yet to be on our website, did focusing on award winning do better or worse than the current ads
  • • For remarketing audiences; people who have been on our website, did focusing on award winning do better or worse than the current ads

With display, we don’t have anyone ‘most important ad group’; all the ad groups convert; and none of them represent a huge percentage of the total display conversions.

For our large number of similar audience ad groups; we used two responsive ads:


Then we aggregated the data at the ad level; instead of the ad group level; across all the ad groups.

In this case, we found that focusing on award winning did much better than focusing on automation for similar audiences. However, for remarketing, focusing on automation did best.

With this knowledge, we can now test tweaking the image within the responsive ads for similar audiences to show an image of the award and move on to testing other types of information within the ads.

Taking Action

Once your tests are created, then you need to monitor the results.

Wait at least a week before taking any action as search behavior changes over the course of a week (your Monday morning and Friday night searchers are not the same); and it can be OK to wait longer.

You want to be at least 90% confident in your results before you take any action; and for brand terms, you might wait until you are at 95-99% confident in the results.

Once you have achieved statistical significance, then you will pause your losers. Examine why some ads lost and the other ads won. This examining phase is crucial to pull insights from your testing results. Make sure to save a snapshot of the data along with any notes so you can review the results at a later time when you are setting up your next ad test.

If you are using multi-ad group testing, you might also inform other departments of the results. We often find that the best calls to action for multi-ad group tests make good tests for email buttons, landing page buttons, and other places where you want a user to take action. Multi-ad group testing can give you great insights that can be leveraged by others within your company.


Testing is essential. If you aren’t testing, your marketing will stagnate and you will see a decline in results.

For ad testing, you can test anything. However, the most creative freedom you have is with the headline 2. It’s a great place to start testing.

Think about your top benefits, calls to action, and unique selling propositions. List them out. Examine them. Think how you can make them better; and then test them to find which one resonates best with your customers.

By continuing to test, pull insights, test again, and repeating the process, you PPC accounts will continue to grow and achieve your marketing goals.

Leave a Reply