Upgrade Prompt Readout

A free tool for product-led products, freemium or free trial. Every upgrade prompt and paywall is its own small funnel: users hit the limit, see the prompt, click, and pay. Enter those counts for each prompt to see which converts, with an honest range on every rate, whether one really converts best, and which prompt and stage to fix first. It runs the engine of the plg-upgrade-prompts agent skill in your browser; nothing you enter is sent anywhere.

What it reads

For each prompt, unique users in one period: how many hit the limit (optional, for nudges shown without one), saw the prompt, clicked its upgrade action and paid through that click. Conversion is paid divided by saw, with a Wilson score interval at 90, 95 or 99%. The engine also gives each stage's rate: delivery (hit the limit to saw it), click (saw to clicked) and checkout (clicked to paid), because each has a different fix: a bug or frequency cap, the moment or the message, or the pricing and checkout flow.

Which prompt to fix first

For each prompt and stage, the engine works out the upgrades you would gain if that stage matched the lower end of the interval of your best prompt at the same stage, and ranks them. Prompts reach different users at different moments, so this says where to look, not what a fix will cause; test the fix. Every pair of prompts is compared on saw-to-paid with a two-proportion z-test and Holm's correction for testing several pairs. The tool names a prompt as converting best only when it differs from every other prompt after that correction.

Why I built it

Teams tend to look at one free-to-paid number and argue about the pricing page. The upgrades are usually won or lost at a particular prompt, at a particular stage. I wanted each prompt read on its own, without a benchmark, and without ranking prompts the data cannot tell apart.

Limits

No benchmarks: the engine compares your prompts with each other and holds no "good" conversion rate.
Count unique users, one period, the same way for every prompt. Upgrades that did not come through the prompt's click belong in a separate analysis.
A prompt that trails your best may reach different users. Fix it, then A/B test the fix.

To test a fix, use the A/B Test Significance Calculator. To set how much a free user can use, see the AI Free-Tier Cost Limit Calculator. For the strategy, read the Product-Led Growth Playbook.

Run this in your agent

The free AAJ agent skill PLG Upgrade Prompts does this job inside a coding agent, on your own data. Install it with npx skills add sarojkjha/aaj-marketing-skills --skill plg-upgrade-prompts.

Using Claude or ChatGPT instead? Add the AAJ engines as a custom connector, with no install: https://aaj-engines-mcp.sarojkjha10.workers.dev/mcp

Frequently asked questions

How do I measure an upgrade prompt?

As a funnel of unique users in one period: how many hit the limit, how many saw the prompt, how many clicked its upgrade action, and how many paid through that click. Enter those four counts for each prompt or paywall.

Which upgrade prompt converts best?

The one whose saw-to-paid rate differs from every other prompt's after correcting for testing several pairs at once. If none does, the tool says no prompt converts clearly best yet, and does not rank them on that data.

What is a good free-to-paid conversion rate?

This tool does not say, because the engine holds no benchmark. Compare your prompts with each other, and this period with the last one.

Is my data stored or shared?

No. The readout runs in your browser and nothing you enter is sent to AAJ or kept after you close the page. If you copy a link to your results, your numbers travel inside that link, so anyone you send it to can see them.