Signalpost · 2024
A pricing experiment that lifted paid conversion by a fifth
Three rounds of tests on a product analytics company's pricing page, including the round that lost, and the plan structure that finally made sense to buyers.
- Client
- Signalpost
- Year
- 2024
- Role
- Product designer, growth
- Duration
- 16 weeks
- Services
- Experiment design
- Pricing UX
- Customer research
The problem
Signalpost sells product analytics priced by tracked events. The pricing page converted 2.1% of visitors to trial, and sales calls kept circling the same question: how many events will we use? Visitors could not estimate their bill, so they either picked the free plan and churned or booked a sales call they did not need. Sales estimated that 40% of call time went on explaining event volume.
I combined three months of page analytics, session recordings, 30 sales call transcripts and 8 interviews with recent buyers and lost prospects. Two findings shaped the work. First, buyers thought in monthly active users, not events, and translated badly between the two. Second, the comparison table listed 34 features, and recordings showed visitors scrolling it end to end without pausing, then leaving the page.
Process
- 01
Test one: a calculator (it lost)
Our first hypothesis was an event calculator: enter your monthly users, get an estimate. It lost to the control by 9% in trial starts over four weeks. Session recordings showed why. The calculator produced a precise number that alarmed people, and many left after seeing an estimate above their budget. We had answered the question correctly and still made the decision harder.
- 02
Test two: plans by company size
I redesigned the plans around user ranges buyers already know, such as up to 10,000 or up to 100,000 monthly users, with generous event allowances in each tier and overage explained in one line. The comparison table shrank to the 9 rows that actually differ between plans, with the full list behind a link. Finance accepted the allowances after modelling showed fewer than 6% of accounts would exceed them.
- 03
Test three: proof next to price
The final round tested what sits beside each plan. Short notes on which teams usually choose each tier, plus a demo dashboard showing what 50,000 users of data looks like, beat a generic logo strip. We ran each variant for a full billing cycle so that quick, low-intent trials would not flatter the results. That slowed the programme but kept the numbers honest.
Selected images
Outcome
The final page converted 2.6% of visitors to trial, a 24% relative lift, and trial-to-paid conversion rose from 14% to 17% because buyers landed on a plan that fit them. Pricing questions on sales calls roughly halved, and the sales team now uses the plan descriptions in its own decks. The losing calculator lives on as an internal tool for account managers.
- +24%
- visitor-to-trial conversion
- 17%
- trial-to-paid, up from 14%
- −48%
- pricing questions on calls
Keep reading
All work