The traffic threshold problem
CRO has one inviolable rule: statistical significance requires volume. If a site gets 5,000 visitors a month, you can't run meaningful A/B tests. The agency that takes those clients on anyway is selling theatre.
The fix: filter out under-threshold prospects, and price the rest by test velocity + research depth.
The 2026 CRO retainer benchmark
Foundation — entry-level program
- Setup: $5,000
- Monthly: $3,000
- Min commitment: 6 months
- Heuristic + analytics audit
- Heatmap + session recording setup (Hotjar / Microsoft Clarity)
- 1 A/B test launched / month
- Statistical analysis + write-up
- Monthly experimentation report
Growth — real experimentation cadence
- Setup: $10,000
- Monthly: $7,500
- Foundation + user research (5 customer interviews)
- 2-3 A/B tests / month
- Personalization experiments
- Funnel analysis + drop-off interventions
- Bi-weekly strategy calls
- Custom dashboard with experiment library
Scale — enterprise-grade program
- Setup: $25,000
- Monthly: $18,000
- Growth + dedicated strategist + designer + dev
- 5+ tests/month (including server-side)
- Custom segmentation + ML-driven targeting
- Quarterly user research sprints
- Roadmap reviews with leadership
- Integration with experimentation platform (LaunchDarkly, Optimizely)
When to walk away
Decline CRO engagements when:
- Site has <50K monthly visitors (not enough for stat-sig)
- Conversion goal is undefined or has <30 conversions/month
- Client wants guaranteed lift % (you can't guarantee experiment outcomes)
- Decision-maker doesn't understand basic stats (they'll fire you on the first inconclusive test)
What pricing structure won't work
Per-test pricing. Tempting but bad:
- Doesn't account for research that has no test attached
- Incentivizes shipping low-quality tests just to bill
- Makes commitment look transactional
% of revenue lift. Sounds aligned but:
- Hard to attribute lift to one agency
- Punishes you for bad test outcomes you can't control
- Creates accounting nightmares
Flat retainer + setup fee is the only sensible model.
Where Growth justifies its 2.5× price
The thing that lets Growth charge $7,500 vs Foundation's $3,000:
- User research — 5 customer interviews per month surfaces hypotheses no analytics tool finds
- Personalization — segment-based experiences (returning vs. new, mobile vs. desktop, geo)
- Funnel analysis — not just landing-page tests, but checkout-flow + onboarding tests
- Custom dashboard — agencies who own the data view own the renewal
Free template
The CRO retainer template with three tiers, test velocity defined, and traffic threshold filter built in.