Ferret

About the project

A demo form submits but the lead never reaches the CRM. The tracking expected on a conversion page disappears after a deploy. A pricing page quietly gains a noindex tag, or robots.txt starts blocking an important product path. In every case the page still answers 200, so nobody finds out until the pipeline does.

Ferret continuously verifies three things for HubSpot-based B2B websites: that a synthetic lead submitted through a supported form actually arrives in HubSpot, that configured marketing tags stay present where they are expected, and that critical pages do not become unindexable or blocked.

The method is deliberately dull. An approved expected state, a scheduled observation, a deterministic comparison. A confirmed regression becomes an incident with enough evidence for a marketer to hand it straight to engineering, web, SEO or an agency.

False alerts are the main product risk, so trust comes before breadth. AI never decides whether an incident is real. Monitoring infrastructure that fails is reported as inconclusive rather than as a customer regression. Public scans stay passive and never submit forms or create CRM data. This is not an uptime monitor, an SEO suite or an AI website auditor.

The project is in its pre-MVP validation build. The first target is a one-time deep scan of up to 300 pages that produces a report and a fix brief, enough to prove the defects it finds are concrete and specific. Continuous monitoring follows once that holds up. Ferret is a working title.