We got 152 ad clicks and zero signups. The bug wasn't the ads — it was 3 clicks away from the money.

The 31 who finished a scan already got the answer they came for if the summary says "mostly fine" or lists a count of issues. An email gate only works when the locked part is something they cannot infer from that summary. For an accessibility scan that is usually the exact failing elements plus a copy-paste fix, not a longer restatement of the score. I'd also split the logs by device. Name, password, and a captcha after an ad click quietly kills a lot of mobile traffic that already did the scan. If the email-only version still sits near zero, the free result already closed the question and the report is not the bottleneck.

Reading the raw logs instead of guessing is the real lesson here. For a free tool, I think showing part of the result before any gate matters more than email vs. account.

Love that you traced raw logs instead of blaming the keywords. Three clicks from the ad to the first bit of value is usually two too many. Letting people run a scan on the landing page itself, and asking for the email only when they want the full report, tends to fix exactly this.

The part I'd steal isn't the email gate — it's the log-first debugging. Ads buy you traffic; raw logs tell you exactly where conviction breaks. Most people would have rewritten the landing page three times before ever checking where the drop-off actually was.

One addition for the experiment design: lock in the success metric before the first week of data lands. With an email gate, unlock rate is the vanity number. The real one is what share of unlockers reply to the first email or click toward anything paid. You had 31 scan completers as a denominator — if even a handful of them answer a plain-text "what were you hoping the scan would find?" message, that tells you more about the offer than 100 unlocks ever would.

This is really close to what we went through with UtilitySEO. We run a free SEO scan — up to 2,000 pages, 100+ ranking factors, about 30 seconds, no signup at all. Your data basically proves why: 20% engaged with the tool, 0% hit /register. The wall killed intent that was already there.

On your question about email-gate vs full-account-gate: we went a third way — completely ungated free tier, then paid plans for ongoing monitoring and deeper analysis. The free scan is the demo. Anyone who runs it and finds real issues is already pre-qualified for the paid tier.

Your "email me this report" fix is the right move. It captures the moment the user has already seen value. Registration before value is asking someone to invest before they know if there's anything to invest in.

I turned it into a $147 one time report if anyone wants their store checked: https://workers.miodiollc.com/agent-journey/"

Same pattern shows up on Shopify stores. The traffic is usually fine and the leak is a step or two from checkout. I walk the store the way an AI shopping agent would, through search, product page, cart and coupon, and stop before payment. The stuff that breaks is boring. Size pickers that need a tap in a weird spot, coupon fields hidden behind a link, a cart that reloads and loses the variant. Humans push through that. Agents just leave.

31 scans and 0 visits to /register is a clean number. One thing I'd measure before comparing gates: how many of those 31 scrolled down to the locked part at all. If most left after the summary score, the free part already answered "is my site okay", and any gate will convert near zero. Showing the issue count by severity with one example fix, then gating the full list behind email, gives them a reason to want the rest.

With 31 people already completing scans, what result from the email-only gate would confirm the account wall was the conversion bottleneck rather than the report itself lacking enough value?

the nginx-log trace is the part most people skip, respect for doing it before touching the ads. on email-gate vs account-gate: email-only almost always moves the number, because you are matching the ask to the intent. someone checking "is my site okay" will trade an email for the answer, they will not trade a password and a captcha for it. the account ask only works when the value compounds with use: saved history, dashboards, recurring scans.

the thing to watch after the change is not the unlock rate, it is the downstream quality. email-gate converts more and captures worse emails, so measure scan-to-second-visit and the open rate on the report email, not just captures. if report opens are strong the gate is doing its job. if they are weak, you moved the bounce from the wall into the inbox.

the soft version for clean sites is the smart bit. people with nothing to fix are the ones most likely to share a tool that told them so.

This resonates a lot — how long did it take before you saw any real signal on it?