How I read it, and what I found

How I read the assignment

I'd own talkiatry.com as a product and run it as a system: both the conversion work and the day-to-day webmastering.

• Hands-on Webflow: pages, components, content publishing, redirects.
• Technical SEO/AEO: schema, metadata, crawlability, internal linking.
• Site health: uptime, page speed / Core Web Vitals, broken-link and QA sweeps.
• Accessibility: WCAG 2.2 AA held as an ongoing standard.
• Analytics & tracking hygiene: tag management, events, consent.
• Roadmap & experiments: prioritization, then the conversion work on top.

The lens I bring: a conversion path is a chain of promises. Conversion is what happens when each one is strong, accurate, and kept.

Most of the plan is that one idea applied:

• An honest promise about who we can help (eligibility clarity).
• A promise the click keeps (CTA message-match).
• Promises backed by evidence (proof, footnotes).
• The tracking to see where promises break.

The metric, decomposed

• Start = visitor begins the assessment at app.talkiatry.com/assessment (a different domain than the Webflow site).
• Qualified = actually serviceable: covered state, accepted insurance (no Medicaid), in-scope condition (not eating disorders / schizophrenia), age 5+, virtual-only.

Two honest levers:

1. More qualified visitors start (clarity, trust, CTA, speed, mobile).
2. Fewer unqualified visitors start and stall (eligibility clarity up front), so the numerator is real and downstream booking/show improves.

Method note: built without live analytics (GA4, Lighthouse field data, Search Console, session replay), so it's framed as hypotheses plus the instrumentation I'd use to confirm them. Building that instrumentation is itself a big part of the first 30 days.

Audit: what I observed on the live site (June 2026)

Verified strengths (keep and amplify)

Checked in a Screaming Frog crawl and PageSpeed Insights.

• Proof assets: 13,000+ 5-star reviews, 3M+ visits, "86% better after 2 visits," "~$30 out of pocket," in-network messaging, HIPAA / LegitScript / FSA-HSA badges, clinician credentials.
• SEO/AEO structured data: all ~1,955 pages carry schema, 0 errors / 0 warnings (MedicalWebPage, FAQPage, Physician). Plus medical-reviewer bylines, on-page FAQs, deep internal linking.
• Provider directory at scale: ~1,403 indexable profiles, all with meta descriptions. The site's largest programmatic surface, well-optimized.
• External E-E-A-T: hundreds of citations to authoritative sources (PubMed/NCBI ~425, plus FDA, NIMH, Mayo, CDC, APA). Exactly what a YMYL brand needs.
• /insurance: copay estimator + names major carriers (Optum, BCBS, Aetna, Cigna, UnitedHealthcare).

Verified issues (the openings)

• CTA message-match, not just variety. Several primary buttons imply a different action than the real next step (a ~10-min assessment): "Book a visit" reads as scheduling, "Find a psychiatrist" as browsing. Test the gap; confirm variants are a measured system, not drift.
• Eligibility is fragmented and late. Disqualifiers (no Medicaid, no eating disorders/schizophrenia, age 5+, virtual-only, covered states) are spread across FAQ, insurance, and condition pages. Many learn if they qualify only inside the 10-minute assessment.
• Mobile speed + accessibility have headroom. Mobile Lighthouse 79 vs 90 desktop; accessibility 88. Root cause: systemic color-contrast (WCAG AA) failures across templates, so a few design-token fixes propagate across hundreds of pages. (SEO 92, best-practices 96.)
• Site-health gaps. 5 broken internal URLs (3×404, 2×500), 104 help-center pages missing meta descriptions, some duplicate titles. Routine hygiene this role owns.

Known risk to check first (not a confirmed issue): the conversion event lives on app.talkiatry.com (~776 links to it), a different domain than the Webflow site. Cross-domain boundaries are where GA4 attribution and funnel continuity commonly break, so I'd confirm it's clean before trusting the metric.