55% to 18%: When More
Clicks Meant Fewer Registrations

The Free WHY to Paid WHY.os funnel runs a large share of its volume through a joint venture with Ryan Kohler, whose audience is job seekers. Those people subscribed to Ryan. They read his emails, they watch his videos, and they click his recommendations because they have been following him for a while. Most of them had never heard of WHY Institute.

The flow had three surfaces, not two. Ryan's email went out, the click landed on Ryan's own landing page, and that page handed the visitor off to the WHY Institute registration page. That middle page was a bridge: it was in Ryan's voice, on Ryan's brand, and it explained what the offer was before anyone was asked to fill in a form. In that shape, registration was converting at 55%.

Ryan's team suggested taking the bridge page out. Their reasoning was sound and it is the same reasoning most of us would reach for: every extra page sheds people, so removing a step should cut friction and push more traffic to the registration page. They know their audience better than we do, the idea was worth running, and we ran it.

It shipped inside a batch of changes that all had the same goal. Two of them ended up mattering. The bridge page came out, so email traffic went straight from Ryan's send to the WHY Institute registration page. And the registration form got pushed below the fold, which on mobile meant a real scroll before anyone even saw a field. Most of this traffic is mobile. A third issue is worth naming: there were too many words standing in front of a free offer, and an entry point like that should be close to frictionless.

On the metric the change was aimed at, it worked. More people reached the registration page than before. Underneath that, registration conversion fell from 55% to 18%, and a 37-point drop on that volume meant the funnel was handing us fewer completed registrations than it had been when the extra page was still in the path. More clicks in, fewer people out the other end.

The reason this got caught at all is that registration conversion and completed registrations were tracked as their own numbers, not folded into traffic to the page. Clicks to registration were up, and by themselves they read as a win. The step everyone was watching had improved while the outcome it was supposed to produce had gotten worse.

I keep an experimentation and change log for this funnel. Every change to a page, a form, a sequence, or a routing rule gets written down with the date it shipped and who shipped it. I built it for exactly this situation, so that when a number moves there is a record of what moved with it.

One. I found the window the number broke in, then worked backward through the log and pulled every change that had gone out inside it. Without it, a swing this size on a live funnel would have had a dozen plausible explanations and no way to rank them. With it, I had a dated list of what had actually changed and could work down it.

Two. Two entries stood out. The bridge page was gone, which meant a warm click from Ryan's list was now landing straight on a registration page from a company the visitor had never heard of. And the form had been moved below the fold, which on a mobile-heavy audience meant the ask was no longer on screen when the page loaded. Both changes had taken away something the visitor needed at the moment they needed it: context in the first case, the ask itself in the second.

Three. I mocked up corrected pages against that hypothesis rather than reverting everything blind. The context-giving step went back in ahead of registration, the copy standing in front of the free offer came back down, and the form went back to the top of the page. Then I took the mockups to the dev team, who implemented the form position fix so it held above the fold on mobile and desktop rather than only on the desktop breakpoint where it is easy to get right.

Going back to Ryan's team was straightforward because there was nothing to argue about. Their idea was reasonable, we ran it properly, and the funnel numbers said what they said. I brought the conversion data and the log entries, we agreed on what to put back, and we kept working together. Evidence is what lets a disagreement between partners get settled in one conversation instead of turning into a standoff about whose instinct is better.

Within a few days of the corrected pages going live, registration conversion was back at 55%. The funnel returned to the level it had been holding before the batch shipped, on the same traffic source and the same offer, and the number of people actually completing registration came back with it.

Here is the part I will not dress up. Several changes shipped together, and several changes were reversed together. That means the recovery cannot be attributed to any single one of them. I do not know whether the bridge page did most of the work, or the form position did, or the copy weight did, or whether it took all three. This was a regression I diagnosed and reversed, not a controlled experiment, and reading it as one would be reading more into it than it can hold.

I am also not going to attach a revenue figure or a significance number to this, because neither was recorded at the time. I do not have a figure for how much traffic to the registration page rose either, only that it did. What was recorded is what changed and when, what conversion did on either side of it, and what went back in to recover it. That record is what made the diagnosis possible in the first place.

The bridge page had been counted as a click and nothing else. It read like a tax on the funnel: a full page standing between a warm email click and the form we wanted completed. It was doing two jobs. It carried Ryan's brand, so the handoff never felt like a change of hands, and it supplied the context that made the registration page legible: what the assessment was, why he was recommending it, what happens after you enter your email. Take it out and the click still works, but the visitor arrives at a form from a company they have never heard of with no idea what it is for. Fewer clicks, worse outcome.

Branding was part of the mechanism and not all of it. If the only thing the bridge page did was put a familiar logo in front of people, a logo swap on the registration page would have covered for it. What was carrying the weight was the explanation that came before the ask, and the partner's brand was what made that explanation credible.

The measurement lesson is the one I would put first for anyone running a partner funnel. Traffic to the registration page is the easiest number to see and the easiest one to improve, and it went up here exactly as predicted. Completed registrations went down. If the only thing on the dashboard had been clicks to the page, this would have been logged as a win and left in place, and the funnel would have quietly gotten worse for as long as nobody checked the step after it. Measure at the level of the outcome you are actually buying.

There is one clean data point on this funnel that points the same way, and it was a genuinely controlled test. On refer traffic, a landing-page-first flow beat sending people straight onward, 39% Step 1 drop-off against 76%. A single change, one segment, a clear read. It does not explain the regression, and I am not stretching it to. It independently supports the same idea: context before the ask does more work than a shorter path.

The last piece travels furthest. Teams ship in batches because batches are efficient, and the cost of that efficiency shows up when a number moves and nobody can say which change moved it. A dated record of what shipped is what separates a diagnosis from a guess, and it is also what lets two organizations settle a disagreement in one conversation. None of that depends on WHY Institute, on Ryan's list, or on this offer. It is the practice I would bring to any funnel with more than one team shipping to it.