Ad copy testing is one of the basic tenets of PPC advertising. The ability to test several ad variations at once is valuable to advertisers.
Yet, a surprising number of PPC advertisers don’t take advantage of ad copy testing at all. Of those that do test, many do so in a haphazard, disorganized manner.
I used to be one of those disorganized ad testers. I’d create a couple ad variations and rotate them evenly to see which performed better. Once I had a winner, I’d think up something new to test against it. I’d review results whenever I thought about it.
Lucky for me, this approach did result in better ROI for clients. With large accounts, though, haphazard testing becomes unwieldy in a hurry.
That’s why I started using an ad testing matrix. And you should, too. Here’s how.
Do Your Homework
Probably half the articles I’ve written about PPC have goal identification as the first step. There’s a reason for that.
Ad copy is often your introduction to prospects – it’s the first time they’ve encountered you. And you only get 95 characters to make a good first impression. That’s why goals are so important in crafting your ad testing matrix.
Sit down with your client or boss and talk about the goals of your PPC campaign. The discussion should include some or all of the following topics:
- Your unique selling proposition (USP)
- Special offers or deals
- Key products to focus on
- Key copy points and messaging direction
- Calls to action – what do you want people to do?
- Priority of the above – what are the must-have’s for this campaign?
Talking about goals not only helps to reassure your client or boss that you have their best interests at heart, but it also makes ad copywriting a whole lot easier.
Develop Your Ad Copy Variations
Now it’s time to start writing ads. Take the talking points from your goals conversation and whittle them down into ad variations. Put as many ideas as you can into a spreadsheet – this is a brainstorming exercise first and foremost.
Once you’ve exhausted all your ideas, list the ads in order of priority as identified by your goals discussion. And of course make sure to get any necessary approvals from the client or your boss before launch.
Map Out the Test Matrix
You’re finally ready to put the matrix together! There are a few different ways to organize ad variations. Here are a few that I’ve used:
- Price in ad vs. no price in ad
- DKI vs. no DKI
- Buy call to action: “Buy Now” vs. “Limited Time”, for example
- Company name in headline vs. USP in headline
- With and without “Official Site” in copy
- Keyword in display URL vs. no keyword in display URL
You get the picture. Group your tests in a way that makes sense and will help you learn and take action on the results.
When creating your matrix, isolate each element that you’re testing – ideally by line of ad text. For instance, one test group should be a headline test, another could be a Description Line 1 test, and another a display URL test. This makes it easy to evaluate results. While this isn’t a hard and fast rule, and there are good arguments against it, the more you can isolate the test elements, the cleaner your test will be.
I use Excel for the matrix. When you’re finished organizing all the variations, it will look something like this:
The test matrix is your roadmap for executing PPC tests – it becomes a living record of all the tests you’ve done. When you think of something new to try, add it to the document. Keeping all current and former tests in one place is helpful for those times when a client or boss asks about the results of an old test!
Test One Thing at a Time
It’s tempting to load up all the test variations in your matrix at once. Don’t make this mistake! Work systematically through one test group at a time. This way, you’ll know what really works the best.
Standardize the Test Across Ad Groups as Much as Possible
The beauty of PPC is its granularity: best practices dictate small, tightly themed ad groups with distinct ad copy variations in each ad group. However, this practice can make ad copy testing a time-consuming nightmare.
Evaluating thousands of different tests not only takes hours, it also takes a long time to get statistical significance. Often what ends up happening is that one or two high-volume ad groups get good test data, and the rest of the tests never finish because they never get enough volume for statistical significance.
If you really want to learn what headline, call to action, or offer performs the best, test it across the board as much as you can.
I’m not saying use the exact same ad, though. For instance, if you’re testing whether DKI improves performance, just use DKI in all your ad headlines. The headlines themselves can be worded according to the ad group – you’re testing the feature, not the specific words.
Purists may argue that this isn’t a good test, but I disagree. Sometimes practicality has to outweigh perfect methods.
This is where good judgment comes in. Use your knowledge and expertise to determine whether it makes sense to standardize the test across ad groups, or whether it’s better to run independent tests in each ad group. Think about the time it will take to set up the tests and analyze the results, along with any possible skewing that may happen by being too general in your ad copy, and adjust accordingly.
Analyze Your Results
This seems like an obvious step, and yet I’ve audited countless accounts where ad tests have been running long past the time when a winner could have been determined. Someone set them up and forgot about them!
The whole point of testing is to learn and improve – and allowing a bad ad to keep running after it’s been beaten just hurts results.
Set a schedule for ad test review, and stick to it. A monthly cadence works well for most accounts. For very large accounts, weekly reviews may be called for; for very small accounts, you may only need to review results quarterly. Whatever the schedule is, set it and stick to it.
There are many ways to analyze test results. I like to use Excel pivot tables. Some bid management platforms will actually do the statistical analysis for you.
Whatever your method, make sure that the tests have reached statistical significance before stopping them. There are several free tools out there that will help with this; I like SplitTester.
Don’t forget to record the results in your matrix document. Remember, this is your point of record for all ad tests. Include the date the test concluded, highlight the winning ad, and note the start date of the next test.
Rinse and Repeat
When you’ve completed one round of tests, start testing new ads against the winner. And keep on testing.
At some point, you may want to conduct a back-test: swap in one of the “losers” from previous tests and test it against your current winner. The results may surprise you. I’ve often seen “loser” ads outperform current “hero” ads, for whatever reason.
By following a systematic approach to ad testing with a testing matrix, you’ll give yourself the best chance at great ROI from PPC.
Twitter Canada MD Kirstine Stewart to Keynote Toronto
ClickZ Live Toronto (May 14-16) is a new event addressing the rapidly changing landscape that digital marketers face. The agenda focuses on customer engagement and attaining maximum ROI through online marketing efforts across paid, owned & earned media. Register now and save!