Skip to main content
CRM Analytics turns your install and subscription history into the metrics a Shopify app business runs on (recurring revenue, churn, retention), computed from what Meridian already records for every store. Analytics is available for public apps, and reads your app’s Shopify connection (see Stores for what happens when that connection is missing or has stopped).

The overview

Four cards open the overview: MRR, with its line over the period, then paying stores, ARPU and LTV. Each card shows its value and its change against the previous period picked in the Comparison filter: a percentage for MRR, ARPU and LTV, a count for paying stores. Money keeps its cents ($11,588.58). A change is green when the move is good for the business, red when it is not, and grey when it rounds to zero. With Comparison set to none, or on all time, no change is shown. Nothing on the overview compares with a month earlier. Paying stores counts the stores whose live subscription is worth at least $0.01 a month, from its plan price or the usage it billed in the last 30 days. A store on a $0 plan with no usage is not paying, and neither is a store still in its free trial. Active subscriptions counts every live subscription, $0 plans included, and is the population ARPU divides MRR by. A Plan Builder free plan is not a subscription here: a store that moved to one is not counted. Below that: installs, MRR by pricing plan, and uninstall reasons where merchants gave one. MRR by pricing plan lists the plans that have subscribers at the end of the period, so it describes the same moment as the MRR card. Each row shows the plan, its price (Free for a $0 plan, Usage when its MRR is all usage, and “/mo · annual” for an annual plan), its MRR and its share of the total. The six largest plans are listed and the rest are grouped under Other. Installs counts first installs: a store counts once, in the period of its first install, wherever you read it (the home tile, the installs card, the Base metrics, the Report Builder). A store that removes the app and installs it again is a reinstall, counted apart and never added to installs. The installs card shows the installs of the period with their change against the previous period, and one bar per week, dated. It follows the selected period (in UTC), the segment, the plan and the test data filter. Reinstalls are not on it. A store that reopens after being closed or paused is a reactivation, neither of the two. Uninstall reasons counts one reason per uninstall: the first choice the merchant gave. A merchant who ticked several boxes counts once, under the first. An uninstall with only a written comment is listed as “No reason given”. Each reason shows its count and its share of the uninstalls listed; the six most frequent are listed and the rest are grouped under Other. The card follows the selected period (in UTC), the segment and the test data filter. Under a plan filter it counts only the uninstalls of stores that were on that plan when they uninstalled, the same rule as the Uninstalls figure on Churn, so the two always agree.

Filters

These controls apply to the whole screen:
  • Period: today, yesterday, last 7 / 30 / 90 days, last 12 months, month to date, year to date, all time, or a custom range. Ranges use UTC day boundaries. Charts switch grain with the window (day, week, or month) so a short range still has a readable series.
  • Comparison: against the previous period, previous year, or none. All-time has no previous period.
  • Plan: narrow to one plan.
  • Segment: narrow to a segment, which restricts every metric on the page to that audience.
  • Test data: include or exclude test stores and test charges, under the rule in Live and test stores. Excluded by default: a development store is out of installs, active installs and every revenue figure, and a test charge on a live store is out of its revenue.
  • MRR: what MRR is made of. By default it counts annual plans (at a twelfth of their price) and usage, and leaves out subscriptions still in their free trial. You can add trials, or leave annual plans or usage out. The choice applies to every figure built on MRR: MRR, ARR, ARPU, LTV, active subscriptions, the six movements, revenue and subscription churn, net and gross revenue retention, and revenue by plan. It stays set as you move between the report screens. The request parameter arpuPopulation (subscriptions by default, or paying) sets who ARPU, and LTV with it, is averaged over: all active subscriptions, or only the stores whose MRR is above zero, which leaves out a store on a $0 plan with no usage. It changes no other figure. Each app can store its own default for these choices and for arpuPopulation, which an Analytics request takes for any choice it does not set itself (the built-in default applies to a choice the app never set). The same control also picks who ARPU is averaged over (ARPU per paying customer or ARPU per active subscription). It reads Default when it matches the app’s default, and when it does not, Make it the app’s default saves the current choices as the default for everyone who opens the app’s Analytics. It is read and changed through GET and PUT crm/analytics-defaults, needs CRM: Analytics: View to read and CRM: Analytics: Edit (with View) to change, and each change is recorded in the audit log with the old and new values. Every Analytics response says which choices it applied. The default governs the Analytics figures only: the Report Builder’s MRR at end of range and the MRR on the stores list and a store’s page keep the built-in choices.
A segment that matches nothing is a legitimate answer: you get an empty report, not an error.

The metrics

Twenty-three metrics are computed:

LTV

LTV is the predicted lifetime value of a subscription: ARPU divided by the monthly customer churn rate. Customer churn is the stores that uninstalled the app over the 30 days before the end of the period, minus the stores that reinstalled it, out of the stores installed 30 days before. At 25 % churn and a $30 ARPU, LTV is $120. It follows the MRR control and the ARPU population (arpuPopulation), like ARPU: over active subscriptions by default, over paying stores with paying. With no customer churn in those 30 days there is no figure, and the card shows a dash: a business that lost nobody this month has no basis for a prediction, not an infinite LTV. LTV is a run-rate estimate. What a store has actually been billed is its lifetime revenue, on the Stores list and the store’s profile.

Free trials

A subscription in its free trial is not revenue yet, so by default it joins MRR, active subscriptions and ARPU on the day its trial ends, and that day it counts as new MRR. A cancellation inside the trial moves nothing, because nothing was in MRR yet, and neither does a switch to another plan still in its trial. A switch from a trial plan to a plan with no trial on record counts as new MRR at the switch, since the new plan is paid from then. A paying store that switches to a plan with a trial leaves MRR at the switch (churned MRR) and comes back as new MRR when that trial ends. The trial end comes from the subscription Meridian recorded (Plan Builder, the SDK’s subscription snapshot or Shopify’s subscription webhook). Without one, it is the first time Shopify charged the subscription, read from the billing transactions in the Partner API, once Meridian holds your app’s full transaction history. A first charge less than 24 hours after activation is billing delay, not a trial: Shopify trials last whole days. A subscription with neither counts from its activation, as if it had no trial. That leaves one gap on apps without recorded trials: a trial cancelled before Shopify ever charged it has no charge to read, so it counts from activation and its cancellation counts as churn. Meridian never guesses a trial from a plan name. When the trial end becomes known later (the first charge arrives with the daily transactions sync, for example), the days before it are corrected once, then stay put. MRR growth rate is month over month: MRR at the end of the period against MRR one month earlier (a month end against the previous month end). Each point of its chart compares that day, week or month end with the same day a month before, so the rate means the same thing whatever the period length. With no MRR a month earlier it reads 0%. No Analytics card or Home tile shows it: the change on the MRR card compares with the previous period you picked, like every other figure.

Frozen subscriptions

Shopify freezes a subscription when the store’s own Shopify bill goes unpaid. A frozen subscription is worth $0 in MRR until Shopify unfreezes it. A freeze is not churn: it is its own frozen movement, and the unfreeze is a reactivation. If the frozen subscription is then cancelled, its amount moves from the frozen line to churned MRR that day. A frozen subscription leaves the churn base, and the base of net and gross revenue retention, at its freeze, and comes back as a reactivation when Shopify unfreezes it. So its cancellation counts in subscription and revenue churn only when the freeze happened inside the window: a subscription frozen before the window began is not in its base, and its cancellation is not counted there. The freeze itself is not a loss in gross revenue retention either, and a store still frozen at the end of the window keeps its pre-freeze revenue in net revenue retention.

How movements and churn are counted

  • Plan changes: a cancellation followed by a new subscription within 60 seconds is one plan change, not churn plus a new subscription. When Shopify sends the new plan first, the old subscription’s cancellation is recognised whenever it arrives, even a full billing cycle later. A cancellation within 5 seconds after the same charge was activated is an ordering quirk of Shopify’s feed and is ignored, unless Shopify reports that subscription ended and nothing later shows it alive. A move to a Plan Builder free plan within 60 seconds of the cancellation is a downgrade to $0, not churn, unless the store uninstalls the app before those 60 seconds are over or the paid subscription was frozen (its revenue already left at the freeze, so the cancellation is the churn). That downgrade does not count in subscription churn: the store leaves the subscription count without a cancelled subscription. Meridian keeps only the latest activation of each free plan, so if a store moves to the same free plan again, a past month can show its earlier move as churn after the store’s history is rebuilt.
  • First 30 days: an upgrade or downgrade within 30 days of a store’s first payment counts as new MRR, not expansion or contraction, so a store still picking its plan does not inflate churn. With trials left out of MRR, the 30 days run from the trial end. With trials included, they run from the first activation. One exception: when the store was already subscribed just before the window, its downgrade is a loss of that starting revenue in revenue churn and in gross revenue retention, as net revenue retention already shows it.
  • Win-backs: a store that subscribes again after its subscription ended (cancelled, expired or uninstalled) counts as reactivation, not new MRR. With trials left out of MRR, a win-back on a plan with a free trial is the exception: it enters MRR when the trial ends, and counts as new MRR that day.
  • Usage: the usage in MRR is what each store billed in the last 30 full days, counted only while the store has a live subscription. A store that cancels, uninstalls or is frozen takes its usage out of MRR on that day, and a store billing usage with no subscription is not counted. The movements carry usage too: more usage from a subscribed store is expansion, less is contraction, and a store’s usage arrives and leaves with the store, so the six movements add up to the change in MRR.
  • Churn rates: each trend point covers the 30 days up to it, and each figure is the average monthly churn over the period, the mean of those points (the rate over the last 30 days on a period of 30 days or less). Each rate is measured against the base just before its window: the subscriptions, MRR (usage included when the MRR filter includes it) or installed stores you had when it began. A subscription that starts during the window is not in the base, so a young app can churn more than 100% of a small starting base.
  • Install churn subtracts reinstalls from uninstalls, so it goes below 0% when more stores come back than leave. A store that is closed or paused is not counted as an uninstall. If it was closed or paused before the period began, it is not in the base either, so its uninstall during the period is not counted.
They’re reconstructed from each store’s lifecycle and subscription history rather than stored as running totals, so a figure for a past month is the figure as of that month, and it stays consistent when you change the window.

The report screens

For anything else (your own dataset, grouping, filters and chart) use the Report Builder.
Last modified on November 10, 2026