A/B Testing is a constant paid search practice with the process looking like this:
Step One: Try something new.
Step Two: Let it run for an appropriate length of time.
Step Three: Determine the winner.
Step Four: Repeat.
But what happens when we can’t conclude a test because the campaign does not get enough conversions? Do we conclude tests early without significant data or do we try to wait it out? In most situations, we extend the timeline.
For questions of when to pause a keyword or add negatives, we open the timeframe to incorporate 90 days or 120 days. However, ad testing is a little more complicated. In most campaigns, ad testing can run for 30-60 day intervals, but if the campaign only gets ten conversions in 60 days, it may not be enough data to conclude significance. For me, significance depends on the type of campaign and how it relates to the rest of the campaigns within the account.
Scenario 1: Odd Man Out
This campaign is the ugly duckling of your account. While all of the other campaigns convert regularly, this campaign’s conversions trickle in. Furthermore, all campaigns point to the same goal. For example, a lead generation client may have the same form completion goal for each of its campaigns. By creating a tier system to test your ads, where ads are segmented by anticipated volume, you will be able to find takeaways for your low-converting campaign from a high volume one.
In this scenario, I take the ad copy which is working in my highest converting campaigns and use it in the low-converter. I will not need to ad test separately because I already know what works well. Every time I conclude an ad test I refresh my lower tiered ad copy with the winning messaging. The new ad copy will need to be adjusted for the ad group, but the messaging has proven to generate conversions, allowing the best ad(s) to show in my ugly duckling campaign.
Here is a visual representation of how this testing works:
Scenario 2: This Campaign Stands Alone
This campaign might also be your ugly duckling, however, the conversion goal is different. It may be a different product or an ecommerce campaign in your lead-generation heavy account. Conclusions that you can draw from the other campaigns cannot be of assistance here.
In this type of situation, I re-evaluate which metrics or analysis will best suite this campaign. First I’ll decide which type of campaign I have. There are two traditional types of campaigns that fall within the definition of a low-converting campaign:
- If my campaign has a high click volume, but few conversions I would look to update ad copy or my landing page. Ads have a high click-thru-rate, but conversions should be the focus. The ad is the first stage of convincing people to visit my website, but when they arrive they don’t see what they expect. In the example below, Test E has a better CTR, but CPA is higher while conversion rate is lower.
- On the other hand, I may see few clicks and a lower CTR, but a higher percentage of those clicks are converting. In this case visitors are getting what they expect from the landing page, but the ad isn’t convincing enough to click. By optimizing ads for CTR, more people will visit the site and this should help utilize the performance in the campaign that already does well. The table below demonstrates how low CTR sometimes leads to higher conversion rate.
Scenario 3: All Your Campaigns Are Low-Converters
Tying strongly to the campaign that stands alone, in this situation the account needs more time. If you cannot reevaluate your metrics, you may just have to wait it out. I would test an account wide overarching ad copy message. Though aspects of the test should vary by campaign and ad group, the message will remain similar. This test will allow you to eventually draw some conclusions after the appropriate length of time.
While ad testing might seem like a tiny aspect of PPC, it is extremely important. Finding the right approach to ad testing, even if the campaign has low volume, will guide you in the right direction of better performance.