Headline Personalization for SaaS: One Homepage, Four Audiences
2026-09-13
Your SaaS homepage has one headline being read by four people who want four different things. Evaluator: does this solve my problem today? End user: will this make my week easier or harder? Executive: what does it cost, what if it fails? Developer: how does it integrate, where are the docs? One headline can't answer all four — so SaaS writes a compromise ("Powerful tools for modern teams") that persuades no one.
How it works, concretely
Prept = one snippet before
+ data-prept tags on headline/subcopy/cta. Existing text is the fallback — renders as usual, never blank; thin-profile visitors see your original. Reads privacy-safe context (referral, company-level firmographics, new vs. returning) — no names, no emails, no PII.
The four audiences, with actual headlines
Baseline for fictional Atlas (incident response): "Ship software with fewer incidents." 1. Evaluator: "Replace PagerDuty in an afternoon — incidents, on-call, and postmortems in one place." (name the incumbent, name the scope, kill the yet-another-tool objection). 2. End user (on-call engineer): "Fewer 3 a.m. pages. Alerts that tell you what's actually wrong." (write to the 3 a.m. pain). 3. Executive: "Cut incident cost per outage by 40% — see the math for your team size." Subcopy: "Atlas customers resolve P1s in a median of 19 minutes, down from 47." 4. Developer (from docs): "One API call to page, one webhook to resolve. SDKs for Go, Python, and TypeScript." (specifics only, no adjectives). Same page, four conversations. 10% control holdout on the original verifies the rewrites; not A/B testing — no fixed variants, no declared winner.
Trial vs. demo CTAs should differ by visitor
Startup engineer → "Start a free trial". Enterprise evaluator (2,000-person company) → "Book a demo". Churned trial returner → "See what's new since March". Personalize the CTA via data-prept="cta" — same 0.25-credit returning rewrite.
Pricing pages deserve it too
Small team (1–20): "Start free. Pay $12/seat when you're ready." Growth (21–200): "One plan that covers on-call, incidents, and postmortems." Enterprise (200+): "SSO, audit logs, and a 99.99% SLA. Talk to sales for volume pricing." Don't waste buying-mode traffic on "Simple, transparent pricing."
Credit math at SaaS traffic levels
1 credit/new rewrite, 0.25/returning, 0.1/passive identify; packs 10k/$29, 50k/$129, 100k/$249; deep enrichment 25 credits — reserve for high-value visitors. Mid-size SaaS: 30,000 visitors/month, 60% new / 40% returning, 2 elements on homepage + pricing: new 18,000×2 = 36,000; returning 12,000×2×0.25 = 6,000; total 42,000 → one 50k pack ($129) with headroom. 150,000 visitors: 210,000 credits → ~$527/month; one extra enterprise demo/month pays for the year. 5,000 visitors: 7,000 credits → $29 pack covers it. Not enterprise-only.
The honest limitation
Personalization doesn't fix a confusing product. If all variants beat the original, the original was the problem. If visitors bounce after a headline written for them, the value prop is the problem. Also: the tool serves the right line to the right person; you still write the lines — from real sales calls, not persona templates.
Implementation in an afternoon
Snippet before . 2. Tag h1/subcopy/CTA on the homepage. 3. Write 2–4 variants per element from real sales objections. 4. Wait a week; check holdout vs. rewritten traffic. 5. Expand to pricing and signup. Millie — the AI behind Prept — handles per-visitor matching; you handle audience definitions and copy.
