Copperline tested pricing without breaking trust
A B2B payments company ran its first pricing experiments with guardrails on churn and support volume.

A risky kind of test
Copperline sells payment tools to mid-size businesses. Pricing was set by instinct and revisited every year. The finance team distrusted pricing tests because a bad one can anger customers who see two prices for the same plan.
Designing a safe test
Kerf let the team constrain the experiment to new visitors only. Existing customers never entered it. Visitors were bucketed by account, not by device, so a person who returned on a phone saw the same price.
- Only new sign ups were eligible.
- The bucket was sticky for the full trial period.
- Guardrails watched refund requests, support tickets and trial to paid conversion.
- Finance had read access to the live results.
What they tested
The team tested the order of the plans and the anchor price shown first, before touching any actual price. Showing the middle plan as recommended, with a clear annual discount, lifted upgrades without changing what anyone paid.
We assumed a pricing test meant a price change. Most of the gain was in how we presented the plans we already had.
Head of Growth, Copperline
Results
Plan upgrades rose 11% and no guardrail moved. The AI readout flagged that the effect was concentrated in teams of 10 to 50 people, which became the basis for a dedicated plan.
Pricing is now reviewed every quarter, using the same test setup each time.
More customer stories

Northwind cut checkout drop-off by 18%
A retailer with 2,400 employees moved every checkout change behind a flag and a test, and found the fix was a smaller form, not a redesign.

Halcyon ships weekly instead of monthly
A health scheduling platform replaced a manual release process with gradual rollouts and automatic rollback.

Brightpath replaced three tools with one
An education platform consolidated flags, experiments and session replay, and the arguments about whose numbers were right stopped.