Yesterday we looked at how personalization can tempt you to focus on lower impact (but complex) campaigns. Another risk with personalization at scale is that it becomes mentally taxing and time consuming to maintain. Let’s stare this danger in the face.
Personalization campaigns are complicated to set up
Not the first one. The first one is a liberating act of creativity and deep empathy. Obviously if your online shoe store shows galoshes on the home page to Returning Visitors who have Not Purchased Galoshes when It’s Raining, they’ll appreciate it. And probably purchase more. Click, click, click, done.
The issue arises with your second, third, tenth campaigns. You’ve divided your audience into two segments – “Returning Visitors who have Not Purchased Galoshes when It’s Raining” and “Everyone Else.”
So if your second campaign targets “Loyalty Club Members who have Abandoned Cart”, technically that audience is now “Loyalty Club Members who have Abandoned Cart but who are NOT Returning Visitors who have Not Purchased Galoshes when It’s Raining.”
It gets crazy, fast. Your personalization platform will have features to make this more manageable – the option to quickly exclude an audience from a campaign, or campaign prioritization and mutual exclusivity.
That’s handy, but your brain still has to process the fact that there are 3 home pages – one for Returning Visitors who have Not Purchased Galoshes when It’s Raining, one for Loyalty Club Members who have Abandoned Cart but who are NOT Returning Visitors who have Not Purchased Galoshes when It’s Raining, and one for Everyone Else.
If you didn’t bother to finish that last sentence, I don’t blame you. If it’s painful to read, think about how painful it is to manage.
More effort for less reach
Your personalization efforts will have your team creating designs, writing copy, writing code, and doing QA for experiences that will reach ever-smaller segments of your audience.
Very simply, if the ROI that comes from personalization for a particular segment is worth more than your team’s time invested, you’re okay. (If it’s worth more than any other way you could’ve spent your time, you’re really winning.)
So if you’ve identified a sizable segment that’s already converting, and see an opportunity to greatly improve their experience or better communicate the benefits of your product, awesome. If you’re idly theorizing about what Firefox Users who have Abandoned Cart might find compelling, maybe think bigger.
Performing QA on a given personalization experience is no different from QA on A/B test experiences. Get a preview link, a checklist, a cross device testing platform, do your thing.
It gets fun when you start trying to QA the campaign activation. Did you set everything up right? Are Returning Visitors who have Not Purchased Galoshes when It’s Raining really seeing that new home page design? You can easily mimic a returning visitor who has not purchased galoshes, but can you make it rain?
This is such a complicated topic that Brooks Bell has a best practices article on the topic. It’s thorough and well thought out; you should give it a read. Pay particular attention to stuff like this:
… this introduces new challenges: it requires a higher level of effort from your developer … Ultimately, this may require you to weigh the pros and cons of testing that segment.
Please do identify valuable audience segments and figure out how to delight them while increasing revenue. Do that all day long, and get a giant raise, and tell me about it, and I will send you an Internet Hi Five.
But when trying out personalization, be sure the opportunity cost and the complexity it introduces to your testing program don’t outweigh the potential benefits. It’s easy to paint yourself into a corner with fancy code and a pile of AND / OR statements. Be careful out there.