(It won’t, but this is a LinkedIn post.)

Our attribution data is a seething mass of “Unassigned” and “(not set)” and “direct / none.” GA4 shows people are converting. It gives no indication of where they came from. Like the pasta sauce I’m fermenting. I looked at it this morning and there’s a thin white film of kahm yeast on top. No idea where that came from either.

A jar of fermenting pasta sauce with a thin white film of kahm yeast on top

Here’s what we were trying to do. Every campaign we run: Meta ads, Google Ads, social posts, email, all tagged with UTMs. Source, medium, campaign, and content. The kind of granular attribution that tells you which ad drove which registration, which email drove which membership signup. My pasta sauce has Campari tomatoes, shallots, garlic, and fresh basil leaves. A little over 2% salinity by weight.

Daxko is the platform where most of those conversions actually happen. Memberships and program registrations. Purpose-built for nonprofits: YMCAs, JCCs, Boys and Girls Clubs, etc. I cleaned and sanitized everything. Closed the mason jar off from oxygen exposure.

It is also very good at stripping UTM parameters from URLs.

We watched our carefully built attribution structure collapse. We escalated to Daxko. We got sympathy and not much else. I watched the ferment very carefully, worked hard to minimize oxygen exposure. Still got the kahm yeast.

What we have now is directional data. We can see trends.

Kahm yeast is harmless. You scrape it off and work with what’s underneath. The sauce is still good and the data is still useful. Both tell me something. Just not everything. But what I learn is directional.

Which is why I’ve been investigating server-side Google Tag Manager. Not because it’s the obvious answer. Because it might be the right one for us. I wrote about that investigation here.