👈 Home

How to fail at Conversion Rate Optimization: Optimize metrics you don't trust

Your CRO efforts will fail if nobody wants your product, and they’re a big fat waste of time if you’ve already decided what to do.

Another way to fall flat on your face is to run experiments when you don’t have a reliable measure of success.

I hesitated to write this, since the above sentence seems blindingly obvious.


It’s totally a thing.

This practice pops up most often in lead generation, where the website’s job is to collect form submissions, which only turn into money when a sales team closes a deal.

Form submissions are trackable, and optimizable. From your homepage hero section right down to your Submit button, there are opportunities to better communicate the value of your offering, remove obstacles in the visitor’s path, and low key manipulate people into signing up. (If that’s your thing.)

The problem arises when nobody cares about form submissions.

The issue depends on the boundaries, and relative organizational power, of the marketing and sales teams.

In a company where marketing holds some sway, the following assertions are reasonable and accepted:

If this is you, congrats! Keep optimizing.

What’s distressingly common, though, is the following:

In this type of organization, the true metric of success is the sales team’s feelings. I don’t know of a way to reliably measure, much less optimize, that goal. Sorry.

If this describes you, and you’re interested in CRO, your best bet is to start searching for your next role.

Bonus! Another way to fail

In addition to “nobody cares about form submissions,” some marketers find themselves in the scenario where lots of form submissions are junk.

It might be spam, or it might be that the form in question serves multiple (too many) purposes.

“Oh, lots of existing customers use that form to ask support questions.”

These irrelevant conversions add noise to your data. If you’ve got a lot of them, the noise will easily overwhelm the signal coming from real leads.

Result: You’re more likely to detect impact where there is none. Did Variation C on your heading copy really convince more prospects to reach out? Or did random chance lead to a higher proportion of customer inquiries?

To know, you have to disentangle the two. Give customers a clear, distinct path to ask questions.

Making this action prominent and easy to take has the additional benefit of signaling that you actually have customers, and that you take care of them. It might even increase conversions by 0.7%!

Either way, it’s the right thing to do. And it’s the only path toward a metric you can trust enough to optimize.

    © 2024 Brian David Hall