Back to Insights
TeardownApril 18, 2026

I used Microsoft Clarity and Claude to find why 62% of users abandoned a SaaS signup form

A synthetic walkthrough of a Clarity + LLM audit on a B2B SaaS signup flow — where the drop-offs actually happen and why AI alone doesn't get you to the fix.

Most signup-form audits start and end with "shorten the form." That's the advice. It's almost always wrong — or at least incomplete. This post walks through a real methodology: Clarity captures the behaviour, Claude interprets the patterns, judgement writes the fix list.

The setup

Describe the client (composite/synthetic or anonymised). One-paragraph product context. The headline number: 62% abandonment between step 1 and the verification email. What the team had already tried (form shortening, social login) and why it didn't move the needle.

What Clarity actually showed

The specific Clarity signals that mattered: rage clicks on a non-clickable "continue" label, dead clicks on a tooltip icon, scroll-depth stalling at the pricing anchor mid-signup. Include 2–3 concrete screenshots (or placeholders) and explain what you were looking for before each one.

What I asked Claude

The actual prompts I ran against session replays and the Clarity aggregate data. Why I didn't just ask "why are people abandoning" (LLMs bullshit when the question is that open). The structured prompt pattern: here is one user's path, here are our three hypotheses, score each 1–5 and tell me which signal in the replay most supports each.

The three root causes we actually found

One paragraph per cause. Each names the Clarity signal, the Claude interpretation, and the human judgement call that turned it into a fix.

  1. Trust collapse at the pricing moment
  2. Ambiguous "continue" affordance at step 2
  3. Verification email delay creating a dead zone

The fixes we shipped

The exact copy changes, the component move, the email provider swap. Before/after screenshots. The 15-minute vs. 3-hour fixes, separated — most wins are 15-minute.

What this methodology can't do

Where it fails: new products with <500 sessions/week (no signal), highly gated B2B flows (sample too small), anything where the drop-off is off-site. Honest about limits.

Takeaways

Three bullets. The transferable pattern, not the specific fixes.


Want this run on your signup flow? Book a Teardown — £1,500. We install Clarity, run it for 7 days, and ship the top three fixes. If we can't project £1,500/month in recoverable revenue, you pay nothing.

Get More AI Infrastructure Insights

Subscribe to receive weekly articles, case studies, and expert strategies delivered to your inbox.

No spam. Unsubscribe anytime.