Skip to content

Explain a drop in in-app subscription MRR, from ChatGPT or Claude

MRR is falling: your AI reads Purchasely, isolates churn by plan, platform and country, then looks for the cause in App Store and Google Play reviews. No Purchasely account, no export.

Published on November 1, 2026

In short

  • With Persyst’s Purchasely connector, your assistant breaks an MRR drop down into new subscriptions, churn, refunds and unconverted trials, by plan, platform and country.
  • With the Google Play and App Store Connect connectors, it then searches the reviews for what departing users say.
  • Product, marketing or leadership ask with no Purchasely account and no spreadsheet to maintain.

MRR dropped 6% in October. Where does the drop come from?

MRR goes from €48,200 to €45,300. Paid exits rise to 1,150 against 820 in September, almost all on the monthly plan on Android (+290); annual plans are stable and so are refunds (41 against 38). New subscriptions do not make up for it: 980 against 1,060, because trial conversion falls from 31% to 24%. On the review side, 27 Google Play reviews in October mention the price, against 6 in September, for example “€9.99 a month is too much for what it is”.

Sources consulted

  • Purchasely · revenue by month
  • Purchasely · subscription trends by platform and plan
  • Google Play reviews · search
Example exchange, fictional data.

The problem: MRR is down, and the board wants a reason

The Purchasely dashboard shows the curve. To say why it drops, you open churn by plan, then by platform, then by country, check refunds, see whether trials convert less, and finally read store reviews to understand what departing users say. An hour of clicks and screenshots, to redo before the next board meeting.

An AI assistant walks that path in one conversation, as long as it can read Purchasely and the reviews.

What the assistant does, step by step

With Persyst’s Purchasely connector, the question “Where does the MRR drop come from?” runs this reasoning:

  1. Measure the gap (get_purchasely_revenue, by month): MRR, subscription and one-time purchase revenue, churn and refunds, for the month and the previous one.
  2. Break down the movement (get_purchasely_subscription_trends): new subscriptions, reactivations, paid entries and exits, trials and introductory offers, then the same split by platform and by plan.
  3. Isolate the segment that moves by reading again with the filter found: the monthly plan on Android, or one country.
  4. Look for the cause in the reviews, when Google Play or App Store Connect are connected (search_play_reviews, search_app_store_reviews): “price”, “subscription”, “cancel”, with several phrasings.

The answer separates what comes from exits, entries and refunds, and quotes the reviews that illustrate it.

Asking the right question

  • Name the period and the reference: “October, against September” or “against the same month last year”.
  • Ask for the breakdown: “exits, entries, refunds and trials, separately”.
  • Name the app when several are open: the assistant does not add them up.

Three questions to try

  • “Are annual plan exits rising too, or only the monthly plan?”
  • “What share of trials started in September converted to paid, by platform?”
  • “Do October’s refunds come from one country in particular?”

Setup

  1. An administrator of the Purchasely account creates a Management API key with viewing scopes only, and the Persyst administrator plugs it into Data sources: the Purchasely guide covers the steps.
  2. They pick the open apps, then open them to product, marketing or leadership. For the review side, the Google Play and App Store Connect guides come on top.
  3. Each member adds Persyst to their assistant (getting started) and asks their question.

Limits to know

  • Read-only: the assistant analyzes, it changes no plan, paywall or campaign.
  • Purchasely limits the number of calls per key, especially on dashboards: over a long period, the assistant may have to wait or narrow the requested range.
  • The figures are Purchasely’s: a subscription sold outside its SDK is not in them.
  • Reviews only explain part of the churn: a user who cancels without writing anything only shows up in the figures.

Questions

Frequently asked questions

Does each member need a Purchasely account?

No. An administrator plugs a read-only Management API key into Persyst and opens the apps to the relevant teams. Members ask about the figures with no account in the Purchasely console.

Are these the figures from the Purchasely console?

Yes. Persyst queries Purchasely on every question and passes its figures through, with no recalculation on its side, apart from a cache of a few minutes. For the same period and filters, the MRR is the console’s.

Does the assistant see who unsubscribed?

Not by default. The tools that read a user’s record or the list of purchases stay disabled until an administrator enables them. Analyzing an MRR drop does not need them: it works on aggregated figures.

Can we compare several apps?

Each open app is read separately and compared in the answer. The assistant never adds up the figures of several apps, and only sees the apps open to the person asking.

Connect one tool. The whole team asks away.

Connect a first tool and try Persyst for free with your team for 14 days.

You use your existing AI assistant subscription.