Artifacts · The review
Supporting artifacts
Three concrete deliverables the plan produces: the first experiment brief, the tracking plan, and the QA checklist that holds quality as a standard.
Experiment brief (the first test I'd run)
- Title: Eligibility-clear CTA vs. current hero
- Hypothesis: visitors who can self-qualify (coverage + fit) before the 10-minute assessment will start at a higher qualified rate.
- Variant: hero with a scannable eligibility strip (states, plans + Medicaid note, in-scope conditions, age 5+, virtual-only), a coverage-led CTA ("Check your coverage and get matched"), and proof moved adjacent.
- Control: current "Find a psychiatrist" hero.
- Primary metric: visitor → qualified application start rate.
- Guardrails: total starts, downstream booking/show, bounce, mobile start rate.
- Audience/split: 50/50, new visitors, all devices; segment-read by device and source.
- Duration/power: run to a pre-calculated sample size before calling it; monitor guardrails to stop early on a regression.
- Decision rule: ship if qualified-start rate improves with no guardrail regression; else iterate on the framing.
Tracking / tagging plan (core events)
Events the prototype fires today (open its Events panel to watch live); the prototype is the source of truth, this documents it.
| Event | Trigger | Key params |
|---|---|---|
page_view | page load | page |
cta_click | any primary CTA click | label, text |
eligibility_start | first interaction with the checker | location |
eligibility_answer | each question answered | question, value |
eligibility_incomplete | submit with a missing answer | first_missing_step |
eligibility_result | checker returns a verdict | eligible, reason, who, concern, insurance |
assessment_start | click through to app.talkiatry.com | eligible, cross_domain, from |
eligibility_reset | "Start over" clicked | (none) |
Every event also pushes to window.dataLayer for GTM. Production additions beyond the prototype: cta_view impression events via IntersectionObserver, GA4 cross-domain linker to the app, consent mode for health-data compliance, UTM hygiene, and GTM preview/DebugView QA before publish.
QA + accessibility checklist (per page/template)
- Functional: all CTAs fire correct events; forms submit; redirects resolve; no broken internal links.
- SEO/AEO: unique title/meta; one H1; logical heading order; FAQ/MedicalWebPage schema valid; reviewer byline + date present.
- Performance: LCP/CLS/INP within target on mobile; images sized and lazy-loaded; fonts non-blocking.
- Accessibility (WCAG 2.2 AA): keyboard navigable; visible focus; contrast pass; labeled form fields; descriptive alt text; tap targets at least 24px.
- Mobile: primary CTA reachable; no horizontal scroll; no layout shift on load.
- Cross-browser/device: spot-check iOS Safari, Android Chrome, desktop.
All site facts in this write-up were observed on talkiatry.com in June 2026 and labeled as verified vs. to-be-confirmed. Anything depending on internal analytics is framed as a hypothesis plus the method to validate it.