A/B Testing For PPC
Pay Per Click advertising is often a crucial channel towards an online marketing campaign and in turn, can be one of the most expensive. Because of this, it becomes vital to keep testing your ads to make sure you are getting the most return on your spend, and not leaving money on the table.
Before we start our testing, there are a few things we need to consider:
Decide Our Success Metric
This is working out what metric we want to improve, and how we will do that. If we want to improve our click through rate, we are likely to test our ad copy. Defining our success metric will help us develop our test hypothesis and help us determine the winning formula. Other success metrics we can and should be testing include:
- ROAS – Return on Ad Spend
- CTR – Click Through Rate
- CPC – Cost Per Click
- CR – Conversion Rate
- CPA – Cost Per Acquisition (or Action)
Define Your Hypothesis
In order to have a successful AB test, there needs to be a clear hypothesis. This is where we are making a prediction as to what will change once we have implemented the test. So, we will be saying “Changing [success metric] to [what it will be] will lift the [success metric] to [outcome]”
Generally, to achieve our success metric objective we will be testing any of the following:
- The Headline
- Body Text
The headline of the advert is the part that shows up in blue text and is a link. This usually tells the user what is being offered. It needs to be to the point, short and inclusive of a keyword being searched on Google.
The body text is essentially the same as your pages meta tag in an organic search result. It gives us an opportunity to give the searcher more information regarding the product or service. We could test this by having one ad set focussed around information about the product and another with urgency on pushing the sale.
Using different links is going to be very important because that is where we will send the customer. If we are offering an advert about pink shirts, and then send them to a page about black shoes, that isn’t going to convert very well. We need to ensure that the pages that the customer will go to will follow their expectation of what you’re offering.
Testing different ad sets against different keywords is a great way to see what combinations work best with different customers. For example, a certain customer profile is going to resonate with different words better than others. Some people will be more likely to click on “Private School” over “Independent School” because of the exclusivity the word Private suggests. Some people like that, and some do not.
It is important once we have set up our test that we don’t simply forget about it and we need to ensure we track and analyse our results. We will want to see how many click throughs get and how many conversions those clicks lead to. We don’t necessarily want to just try and improve one metric, because getting 100s more clicks that do not convert is pointless and actually makes us worse off.
We should be running the test for ideally a couple of weeks, but a minimum of 5 days and we want to try and keep the targeting as close as possible for the two ads, with the same bids and same placements to give us the most accurate data possible.
Here are some top tips and best practices for running AB tests on your PPC campaign
- Test early and test often. Don’t allow your account to sit dormant for too long leaving money out there
- Only test one thing at a time. We want to be able to conclude for sure which test is giving us the best results
- Test the ads simultaneously. If we are running two different adverts, they need to be running at the same time otherwise our results may be skewed with seasonality, buyer intent, sales etc
- Use the data, not your gut. The results will speak for themselves and we need to utilise this information