Regressing to the mean. This is Adolphe Quetelet at his very best. Complex systems – like credit unions, or indeed the broader cooperative system – will not change and will continue to operate with the same inefficiencies until underlying conditions are adjusted. They may peak, they may trough, but ultimately, they’ll normalize at the same level, every time.

What is regressing to the mean?

See if this sounds familiar: You kick off a new vendor relationship, let’s say you’ve brought on somebody to help you with your email marketing, and it works. All of a sudden, email traffic to your website rockets up! You’re thrilled … Until it somehow comes right back down to where it was, pre-engagement.

So you escalate. You bring in the team and say, “Hey, I thought we agreed that X was our target, but we’ve gone back down to Y.” It’s a somewhat uncomfortable conversation but the vendor makes some changes, they come up with a new campaign and – BOOM! – email traffic jumps up again. Until it doesn’t. Again.

You see, the intervention tweaked the plan rather than adjusted your organization’s underlying performance criteria; the issues that were there to begin with – in fact, the very ones you brought that vendor in to address.

The problem is, it’s got nothing to do with the vendor, nor is it down to your email technology. It’s not even anything to do with the credit union! Nope, it’s math. Until you adjust the baseline, you always regress to the mean. Always.

How to fix a drop in loan applications

In the last part of this series, we explored a situation where a credit union was facing a 65% drop in loan applications. If that credit union had hired an SEO company to try and fix the problem, sure, maybe a new SEO program would have sent more members to the loan application, and the credit union would have seen the numbers increase a bit. But what would have happened shortly after? Numbers would have dropped right back down again Why? Because the loan application was broken, and sending more people there with a revised SEO program isn’t solving the underlying issue.

If the credit union had hired an AI abandonment bot that follows up with users and says, “Hey, come back, finish out your loan application,” the credit union would have seen a bit of a spike. And then it would have dropped back down. It would regress to the mean.

It’s real easy to spend your members’ money chasing spikes, but until you adjust the underlying issue, you’re not wasting every penny.

When a win is worth celebrating

Remember a few paragraphs ago, when I said you can cut your vendors a little bit of slack because your baseline is what’s flawed? Well, don’t cut them too much, because some third parties live for those performance spikes. They’ll come to you and  say, “Look, we did the thing! We fixed it! How great are we?” And just like every trough doesn’t require panic, a single spike doesn’t deserve a parade.

Here’s how the conversation goes when your vendor comes to you and says, “We increased website traffic by 35%!” A lot of leaders might say, “Wow, that’s great. 35%,” and leave it at that. What you should be asking for is the corresponding numbers. Did call center volume decrease a corresponding amount? Did online loan applications increase accordingly? Without those corresponding numbers to validate performance, you should be skeptical. You’re likely looking at a short-term spike that will inevitably return to baseline. Nothing has fundamentally changed, and nobody needs a pat on the back.

Time for an operational deep-dive

Of all the homework assignments I’m setting through this series, this is probably the heaviest lift, because it requires you to really know how your credit union operates generally, broadly. It requires you to know what you’re bringing in when you’re not running your HELOC campaign. It requires you to know how your members engage with you online and in person when you’re not running an auto promotion or you don’t have a certificate special.

You have to find those numbers. That’s this article’s follow-up.

Just as you know what your cost of funds is, just as you know where your expense ratio needs to be, you need to know what your digital traffic looks like. You need to know the fundamental numbers of how many people use your app and what they do on it. 

All this deep-dive research pays off during strategic conversations with vendors to make sure they are bringing true value to the relationship. You can challenge them to say, “Are you actually adjusting longitudinal behaviors; are you improving processes, are you driving value, or are you simply chasing spikes that will inevitably regress back to the mean?” 

Because, ultimately, you have to know what your baseline is before you can change it, before you can identify and address the underlying condition and stop regressing to the mean.

This article has been adapted from a presentation Sam Plester gave at America’s Credit Unions’ Small Credit Union Conference, New Orleans, May 2026. To discuss a speaking opportunity or book Sam for your next session, please contact us.