Are You Making Any of These 5 Ad Testing Mistakes?

pushing-needle-backwardPPC allows you to easily split test ads. Better ad copy is a big competitive advantage. Google rewards people who write better ad copy with higher positions (more often) and a lower CPC.

However, the ease of split testing can lead many people astray, leading them to pause the better ad and use the under-performing ad. We all intend to move the needle forward, but sometimes, we inadvertently push the needle the wrong way.

Here are five things to be aware of while testing.

Mistake 1: Pausing an Ad Within the First Couple of Days of Campaign Performance

In the first couple of days, even with all other rotation settings set properly, Google tends to rank new ads either higher or lower, depending on the historical ad performance in the account (quality score).

We have clients who spend thousands of dollars a day on specific keywords, so the data would look “statistically significant”. However, the initial couple of days of data can be anomalies.

This is especially relevant for click-through rate (CTR), which varies significantly based on whether you’re on the top or side of the search results. In the initial couple of days, Google fluctuates what percentage of the time your new ad is on top or on the side.

Mistake 2: Using Ad Performance From a Period Where Only the Control Ad Was Running

The only way to test is to use statistics from when both ads are running simultaneously, not a before and after. Conditions in the market could have changed, you may have changed bids on your top performing keywords previously, etc.

If you’re having trouble figuring out when an ad launched, you can use labels to filter and identify when an ad went live.

Mistake 3: Not Running an Ad Long Enough

Similar to the first mistake, many advertisers will let the ad run for a month or so, negating the initial irrelevant data. However, in this case, advertisers will evaluate performance off of too few clicks or impressions etc.

This is simply fixed by plugging your performance numbers into an ad copy split testing tool. There are a number out there, but the one I’d recommend is from Chad Summerhill.

Mistake 4: Leaving Google’s Default Rotation Settings on

Google by default will show the ad that has the highest CTR. In order to properly test your ads, you can’t let Google decide when to show which ad, as your test results will become statistically invalid.

All you need to do is change the default settings on the campaign tabs (as you can see below) to rotate ads evenly. Otherwise, your data can get skewed.


Mistake 5: Not Comparing Top vs. Side Performance

Most people don’t realize, but showing up on the top of the search results (top 3, with all of the ad extensions etc.) can get you 10x the CTR.

At a glance on many ads, it will look like CTR on one ad is statistically better, when in reality, it was just shown in the top position more often than on the side. (When you use your statistically significant calculator, make sure to put in stats for either just top or just side. Hint: Using top is way way easier, as you have more click data.)

Related reading

Five ways to target ads on Google that don’t involve keywords
small budget guide to testing ad copy, landing pages, and more
keyword research tools you can use for free
Five ways PPC customer support can help SMBs