Loading leads…
A variant can bring more requests but worse leads. The form carries the visitor ID in a hidden field; your CRM later sends back “qualified” or “signed” for that ID, with no name, email, or phone number. The report puts forms, qualified leads, and signed deals side by side.
The corporate gifts quote form (company, number of gifts, budget, date) against a short form, “Devis gratuit en 2 minutes” (free quote in 2 minutes). Three readings of the same test, corrected together. Fictional data. Differences, ranges, windows, and durations are calculated by Probant.
B · Free quote in 2 minutes wins on forms but loses on qualified leads: more requests, fewer good leads. Rolling it out would cost qualified leads. Qualified leads: -42.9%, from -61.7% to -18.2%.
7,200 visitors exposed more recently than the qualified lead window are waiting. They’re out of the calculation until their window closes.
Exposed visitors whose windows have all closed (day 45), each in the variant of their first exposure. The step-to-step rates describe the funnel without being compared, because who submits a form depends on the variant. Differences are shown above, per exposed visitor. 10,800 visitors still waiting for their window to close.
Duration of each primary metric at the page’s traffic (to detect 30%, baseline taken from the original): a CRM conversion’s window adds to the time it takes to collect the sample.
More forms don’t mean more leads. A variant can bring more requests and fewer good leads (a field removed, a “free” promise). Decide on “Qualified lead” (30 more days), or add it as a secondary metric: the report will flag a variant that gains forms but loses qualified leads. Track “Closed sale” as a secondary metric or guardrail: it confirms the result without making the test longer.
Observed delay for “Qualified lead”, from first exposure to qualification: median 6 days, over 187 conversions · 97% arrived before day 30.
Tests that decide on a qualified lead or a signed deal: their planned duration, including the CRM window, and the day they’ve reached. Each one’s report is on its page.
What the demo site pasted into its quote page.
Read-only demo. In your workspace, these settings are saved and the script picks them up within a minute.
<!-- Marked form: the script adds the hidden field “probant_id” --> <form action="/devis" method="post" data-probant-lead> … </form> <!-- Form sent with JavaScript: at submit time --> <script> const id = typeof probant.leadId === "function" ? probant.leadId() : ""; // id is "" without the visitor’s consent, or if they opted out: send nothing. </script>
The three steps that a test report puts side by side, with the “wins on forms, loses on qualified leads” signal.
By key, with the observed delay from first exposure to conversion, over 180 days. A conversion from an unknown visitor or one who opted out is ignored.
curl https://getprobant.com/api/v1/conversions \
-H "Authorization: Bearer sk_…" \
-H "Content-Type: application/json" \
-d '{ "visitorId": "0190f2c4-…", "metric": "lead-qualifie",
"occurredAt": "2026-11-08T10:30:00Z", "externalId": "deal-1042" }'In a real workspace: HubSpot or Pipedrive, read-only, with the chosen stage turned into a conversion for the hidden field’s ID, and no workflow or relay to write.