Overview
Where we came in
A developer tool billing on API calls with three tiers and an overage rate. Traffic to pricing was healthy and intent was high, but the page was the last thing a large share of visitors ever saw. The support inbox held the answer: the same question, phrased forty different ways, all of them asking what a normal month would actually cost.
The baseline problem
- Pricing was expressed per 1,000 calls, but customers thought in requests per second and monthly active users.
- 31% of support tickets originating from the pricing page were cost-estimation questions.
- Session replay showed repeated scrolling between the tier table and the overage footnote, a pattern of people trying to do arithmetic on the page.
- Exit rate on pricing was 58%, well above the site average.
Hypothesis
What we expected, and why
Written before the test was built. Because we observed X, we expect Y, measured by Z.
Because visitors cannot convert usage they understand into a monthly bill they can approve, we expect that an interactive estimator that outputs a single monthly figure will increase signups from the pricing page and reduce cost-related support contacts.
Experiment plan
How the test was set up
Sample size and stop date were both fixed before launch, so no result here was called early.
Two-arm split test on the pricing page. Control kept the static three-tier table with the overage footnote. Variant added an estimator above the table: a request-volume slider that named a monthly figure and highlighted the matching tier as the slider moved.
- Primary metric
- Signup rate among pricing page visitors
- Traffic
- ~9,700 visitors per arm
- Duration
- 28 days
- Guardrails
- Average contracted tier value, to catch the estimator talking people down into cheaper plans
- Pricing page load time, since the widget adds JavaScript
- Trial-to-paid conversion downstream
Results
What actually moved
| Metric | Control | Variant | Change |
|---|---|---|---|
| Pricing to signup rate | 11.15% | 13.26% | +18.9% |
| Pricing page exit rate | 58.0% | 49.4% | -8.6pp |
| Cost-question support tickets | 84 / mo | 39 / mo | -53.6% |
| Average contracted tier value | $318 | $326 | +2.5% |
| Trial-to-paid conversion | 22.4% | 23.1% | +0.7pp |
The worry going in was that an honest calculator would talk people onto cheaper plans. It did not. Visitors who could predict the bill picked slightly larger tiers, which is the usual pattern when uncertainty is what was capping commitment.
What changed
Shipped to production
- Interactive estimator added above the tier table, defaulting to the median customer volume rather than to zero.
- Overage rate moved out of a footnote and into the estimator output, where it changes a number people can see.
- Tier names rewritten from Starter, Growth and Scale into the volume bands they actually represent.
- A worked example for the three most common integration patterns added below the table.
Next experiments
Already in the queue
- 01Test an annual toggle with the saving stated in dollars rather than a percentage.
- 02Test surfacing the estimator on the docs pages where evaluation actually happens.
- 03Test a self-serve enterprise path against the current contact-sales gate.