What's wrong with personalization? (Part 1: Money on the table)
May 2019
Should you be doing personalization? Probably not, despite the hype. Let’s look at some simple decision criteria for whether it makes sense on your site.
Why do you want to do personalization?
Is it because you’ve read that “77% of marketers believe that real-time personalization is crucial“?Or because you heard that “Website users now expect personalized experiences”?
Maybe it’s because you’ve got a killer idea for a message to show visitors who’ve browsed > 2 categories of women’s shoes and then viewed a handbag product page on a rainy day. Or because “it’s just one of our key initiatives this quarter.”
These are all reasons I’ve encountered, and if they’re your reasons, I probably can’t change your mind. Have fun, do cool stuff!
If you’re considering personalization for the sake of getting more value out of your site, let’s talk about that.
When does personalization make sense?
Personalization is a good strategy when you have optimized your existing site and funnel to the point of diminishing returns.
So if testing 6 versions of your form layout would yield, at best, an additional 100 conversions per month - but adding a personalized home page heading and popup widget for Returning Visitors who have Visited the Solutions Page will yield an additional 105 conversions per month … you should opt for personalization.
Of course you can’t know the outcome of these campaigns in advance, but you have a better sense than you might think. First of all, have you tested 6 versions of your form layout? And a dozen home page designs, and 8-10 home page hero headings, and half a dozen hero images, and … you get the idea. If you’ve optimized to the point of diminishing returns, you know it.
Optimizely calls this out in the story of their ultra-personalized home page:
We had reached a local maximum on our current site. After four years of iteration and conversion optimization, we had achieved the best possible version of the existing design. New A/B tests on the page returned insignificant results.
Is this you? If not, please keep testing!
To add a bit more perspective, math out the size of your proposed personalization audience. How many Returning Visitors who have Visited the Solutions Page do you get per month? What’s the current conversion rate? How much of a lift would it take to get this audience to yield an extra 105 conversions per month?
Optimizely’s own results are sobering - and they personalized their home page for everybody, based on 5 different attributes, not just one audience:
The personalized site did not affect lead conversion rate immediately.
They did see in increase in engagement, solutions page visits, and CTA clicks. (See here for a judgmental perspective on these metrics.)
These results are interesting, but only if you’ve abandoned hope for further optimizing your site. Otherwise, you’re working hard and staying busy, but leaving money on the table.
If it doesn’t make sense, but you still want to try it
It’s fine! Go ahead, in the context of an experiment. Similar to Optimizely, you might do a 50/50 split with non-personalized vs personalized. Pay attention to how much effort goes into the creative, implementation, and QA for this particular experiment. Measure something that matters. See what happens.
Tomorrow we’ll look at some of the human and technical costs of maintaining a personalization program. In the meantime, if you’ve made more money from personalization than you have from experimentation, please hit Reply and set me straight.