Ad Rotation & AB Split Testing (part 2)
In part 1 of this series I discussed the value of narrowing the number of variables you change in AB split testing. Let’s do a quick recap: “What’s the point of AB split testing?” To determine which lines of text work best to generate relevant traffic to your website. “How do I determine which ad is best?” Welcome to part 2 of the series! I’m going to review the core pieces of data to consider when deciding which ad is “best” among the ads you’re AB split-testing. We’ll use the example ads from part 1. There are 7 columns within AdWords that you should enable when reviewing your ads (Clicks, Impressions, CTR, Conversions, Cost/Conversion, Conversion Value/Cost, and Conversion Rate). For those unfamiliar with the column options, there’s a dropdown menu titled Columns –> Customize Columns that allows you to edit which statistics are displayed within the AdWords interface (see below): I recommend starting with these columns and slowly branching out as you begin to feel more comfortable with AdWords. Let’s get started using our previous ad rotation example: Clicks: amount of times your ads have been clicked
Getting a statistically significant amount of clicks is important for any conversion rate testing campaign. In this case, you need to know whether or not the data accurately represents how these ads will perform over the long-run. My general rule of thumb is “at least 30 clicks.” This doesn’t mean that once your ads reach 30 clicks that you NEED to rotate them. However, I almost never rotate ads that have less than 30 clicks. The only exception for this rule is when you see a drastic gap in data indicating that one ad is worse than the other (eg. if one ad had 10,000 impressions with 1,000 clicks, and the other one had 10,000 impressions with 2 clicks, then I’d be confident that my ad with 2 clicks was not a winner). In the example above, there are certainly enough clicks to be confident that these ads will continue to perform in a relatively linear fashion over a long period of time.
Impressions: the amount of times your ads showed up on Google (but didn’t necessarily get clicked)
This statistic is useful for many of the same reasons that clicks are useful. Keep an eye on your impressions to know whether or not you have statistically significant data.
CTR (click-through rate): the percentage of impressions in which your ads get clicked
Google only makes money when ads get clicked. Because of this, Google allocates more weight to an ad’s CTR within their search query algorithm. Like it or not, Google will reward you for running ads with higher CTR’s because Google makes more money from them. Sure, sometimes an ad with a lower CTR will have better statistics in other categories, but keep in mind that Google will reward the ad with a higher CTR with better quality scores. Higher quality scores will lead to a cheaper cost-per-click and extra budget for advertisement in the long run. This doesn’t mean that the ad with the highest CTR is always the winner, but it’s certainly one of the most important pieces of data to consider when rotating your ads. Generally speaking, anything between a 0.10-0.50% difference is generally not enough to outweigh significant conversion data. However, seeing 1%+ differences in CTR’s should certainly have you leaning towards choosing the ad with the higher CTR. If an ad has 3% or more CTR than another, then I will almost always choose to keep that ad over the lower one (regardless of conversion data). In the example above you can see there’s a 2.34% difference in CTR. This difference is likely significant enough to outweigh most other variables, but it’s still good to consider everything before you pull the trigger on rotating your ads.
Conversions: the amount of times a person who clicked on your ad decided to convert
In a vacuum, more conversions are better (obviously). To further analyze conversions, we break our conversion data down into the following three conversion data statistics:
Cost/Conversion: how much you spent (on average) for each conversion
This is a good statistic to keep an eye on. However, it’s sometimes misleading when you’re running multiple types of conversions (in the example above, there’s no way of knowing whether the conversions listed are phone calls, form fill outs, e-commerce sales, etc). This is why it’s good to look at what types of conversions you receive to know exactly what you’re getting out of your campaigns. I generally only consider this statistic in campaigns that are running only one type of conversion.
Conversion Value/Cost: When you configure conversions in AdWords, you’re able to assign a dollar value to each conversion type. Many campaigns will have multiple types of conversions (phone calls, e-commerce sales, form fill outs, etc). These conversions are going to have different dollar values, so you should assign an estimate of the value of each conversion type in the conversion set-up process (for example, if the average sale you get over the phone is $400, and 25% of the people who call you end up converting over the phone, then a phone call is worth ~$100 to you). If you only have one type of conversion, then this category isn’t as useful.
This piece of data is very similar to Cost/Conversion, but it’s more accurate for campaigns with different types of conversions. Notice how in the example above the CTR of the top ad is lower than the bottom ad, but the Conversion Value/Cost of the top ad significantly outperforms the bottom ad. These are contradicting variables that you will need to weigh against the rest of the data to determine which ad is performing better (all other variables aside, in this specific example I would likely choose the ad with the higher CTR. The difference in conversion data would have to be larger for me to forego this large of a gap in CTR).
Conversion Rate: how often a click leads to a conversion (on average)
Ads with a call-to-action such as “Call Today For A Free Consultation!” or “Fill Out Our Free Quote Form!” will often lead to a higher conversion rate. It’s good to incorporate call-to-actions in your ads to let people know what they’re supposed to do once they land on your page. Applying this technique in your ad writing will often lead to higher conversion rates. You can see in my ads written above that I wrote “Call Today!” to encourage the reader to pick up the phone and dial my number.
Keep in mind that there isn’t always a winning ad. Sometimes I will let ads run for longer periods of time to see if their data will differentiate further. Other times, you will see that ads remain neck-and-neck for a very long time. If this is the case, it’s a good idea to rewrite one of them (regardless of data) to make the differences between the two ads more noticeable. These tips will give you a good start to rotating your first ads within AdWords. There will be a third part to this series focusing on how to make your ads look professional and stand out from your competitors’ ads. Thanks for reading, and stay tuned for part 3!