A/B Testing, or A/Z?

A/B testing is a statistics term where you create two versions of a particular variant and test a user’s reaction to one versus the other. Common convention now has extended this term to include tests across many more variants, so it might more properly be called A/Z testing.

If you will recall the hardware store chain example from our previous post, you have some sense of the complexity of managing Facebook advertising across hundreds or thousands of locations in diverse geographies and demographic mixtures. Our data platform watches up to 293 variants on every ad unit every hour to determine what is working and what is not and to modify a campaign accordingly. We let the data tell us what to do; this level of decision making should not be made on intuition.

First you must create your ad units. Will you use photos and/or videos? The title and copy are obviously critical, but so are the wording and button placement for the call to action. When that action leads your target to a landing page, how will you make that page effective in converting a user from a spectator into a participant in your offer?

Facebook provides literally millions ofad options. For example, potential customers can be targeted based on interests, behaviors, socioeconomics, geography (from a whole continent to a defined radius around a store), and even those you want to poach from your competitors. You can pick age, gender, type of device, and choose placement across Facebook in-line feeds, Instagram, Messenger, and Audience Network. And, you pick the day and time to expose ads for each of your locations.

We augment Facebook’s information with our business intelligence tools to bring into play other targeting data. For example, we can use a geo database to let you pick exact neighborhoods where your customers are more likely to be found. (That’s really important, and we’ll cover it in more detail in a subsequent blog post.) We can also consider where else your online visitors have taken action on the web.

Let’s look at practical examples using our hardware stores. If spring flowers are starting to bloom in some of your markets, you may want to create an offer for plants, garden tools, or fertilizers and have that appear when it’s time for Saturday morning yard work. A well-placed coupon at the right time may prevent a drive to the nearest big box competitor and result in a number of add-on sales once you have that customer in the store. Ultimately we’ll know when a customer is in one of your locations, if he or she is a Facebook user and is carrying a phone (both highly likely), and we can spawn instant ads for whatever you are promoting in-store at that moment.

The continual A/Z testing is what makes all this most relevant to your customer base. Your patrons will appreciate getting offers that are highly tailored to them at the right time and in the right place. Instead of buying thousands of glancing impressions from a billboard on the main highway, you’re letting data science provide you engaging impressions at the exact moment of purchase potential. Mass media is for brand building; precision media like Facebook augmented by our tools is for making sales and lifting revenue. And, when you get your report from a campaign, you’ll appreciate exactly what this means to your bottom line.

Leave a Reply

Your email address will not be published. Required fields are marked *