Skip to content
AppDues
Log in
Contact salesSign up

Charts v3: real-time subscription metrics and what changed

Charts v3 is a rebuilt charts experience for subscription apps. Instead of waiting on batched refreshes, most charts now update as events arrive, and the shared data model behind them makes new charts and dimensions faster to ship. This guide explains what changed, why some numbers may look different from legacy charts, and how to interpret the new views without misreading a counting change as a business change.

What real-time means in practice

Legacy charts refreshed in batches every 2–12 hours depending on the dataset. In Charts v3, nearly all charts refresh as events come in. That gives you immediate feedback on launches, paywall tests, and promos, better intra-day monitoring for spikes, dips, or anomalies, and more consistent behavior across charts because they share one live pipeline.

Real-time charts currently cover App Store, Play Store, Stripe, and Subscription Platform Web Billing apps; other supported stores are planned. If a store you rely on is not in that list, expect the older batch behavior there rather than assuming every chart is live.

Why some numbers look different after switching

Charts v3 also revised data definitions, so some values will not match legacy charts exactly. Three changes account for most of the difference: consistent treatment of subscriptions across stores, refunds that no longer rewrite history, and a revised cohorting method. Before treating a shift as a performance problem, check whether it lines up with one of these definitional changes.

Consistent treatment across stores

Rather than handling each store's behavior slightly differently, Charts v3 applies the same rules everywhere. That lets the model separate a product change (moving from plan A to plan B) from a plain renewal, and separate a resubscription after a lapse from a continuous renewal.

In the Active Subscriptions Movement chart, resubscriptions after a lapse now appear as their own positive line item. Previously, Apple App Store resubscriptions were shown as negative churn. Subscription Retention and Trial Conversion now align more closely with how you actually think about subscriber behavior — a product switch is not the same event as a normal renewal.

The change most people will notice is Trial Conversion Rate, which will probably go down slightly. It now counts only direct trial-to-paid conversions; before, someone who resubscribed after finishing a trial also counted as a trial conversion.

Refunds no longer rewrite history

Previously a refund could change metrics in already-completed periods, so historic reports could shift days or weeks after the fact. Charts v3 keeps historical data stable:

  • Revenue is added on the day the purchase happens.
  • A later refund is recorded on the date it occurs, so it never reaches back to alter the original purchase day.

Completed periods therefore stay fixed, and you will see negative revenue entries on refund dates. The same rule applies to conversions and LTV: the original purchase contributes to the relevant timeframes, and a refund removes those contributions only in the periods where the refund falls.

Worked example: a purchase on March 3 adds revenue to March. If it is refunded on April 12, March stays as it was and April shows a negative entry. In legacy charts, that April refund could have reduced March's figure instead.

Cohorting now follows each customer's start

Cohort Explorer and Prediction Explorer use a revised cohorting method. Instead of defining a large cohort (such as a whole month) by its first and last calendar days, each customer's lifecycle is calculated relative to when they actually started, and the results are then aggregated.

Two practical effects: late-joining customers no longer have their early revenue pushed into the next period, and cohort metrics such as 0–30 day LTV are more consistent and easier to compare over time.

Expected metric shifts at a glance

The quick-reference table below summarizes the expected direction of change and the reason for each. Treat it as a guide to what to expect, not a guarantee that every account will move by the same amount.

Period-over-period comparisons

A single toggle in the Compare dropdown lets you compare the selected date range against the previous period. When enabled, charts plot the current and comparison periods as separate lines, show total values and percentage change in the summary, and let you hover any point to see the percentage change versus the previous period at that moment.

This is available on key charts including Active Subscriptions, MRR, ARR, Churn, New Customers, New Trials, Conversion to Paying, and Refund Rate.

New and redefined dimensions

The segment picker now groups dimensions and includes search. New dimensions include Custom attributes (segment by attributes you set on customers), App version (the first app version a customer used), and Attribution (filter or segment by Source, Campaign, Keyword, and similar fields).

Two existing dimensions were tightened:

  • Platform now means the customer's first-seen platform, not the last one they touched, so the segment is more stable over time.
  • Country now prioritizes the app store storefront over IP-based location, which aligns charts more closely with how revenue and subscribers are actually distributed across markets.

One gap to know about: customers created through the API have no platform, country, or SDK version attached, so they cannot be segmented or filtered by those dimensions.

New charts worth knowing

Active Customers plots every user seen in the measured period, whether or not they purchased, which makes it useful for tracking daily active users. It does not yet correct for aliased customers, so when the same person appears under multiple linked app user IDs, counts may be overstated.

Trial Conversion Rate is cohorted by trial start date and focuses purely on the trial-to-paid step, making it easier to see how acquisition, pricing, or paywall changes affect payment behavior over time.

Active Subscriptions Movement has been refreshed into a state bar format that shows how subscribers move between states — new, renewed, churned, product changes, resubscriptions, and so on.

Paywall Charts give paywall performance its own place: conversion, drop-off, and downstream value, segmented by the same dimensions used elsewhere. Other charts can also be filtered and segmented by Paywalls, so conversion and revenue outcomes can be tied back to specific paywall experiences and experiments.

A checklist before you act on a metric change

When a number moves after switching to Charts v3, work through these questions in order:

  1. Is this metric on the expected-change list? Trial Conversion Rate, Historical Revenue, Active Subscriptions, and MRR/ARR each have a documented reason for shifting.
  2. Did a refund land in this period? Check for negative revenue entries on refund dates before assuming a sales drop.
  3. Is this a counting change or a behavior change? Active Subscriptions and MRR/ARR should change minimally because the counting method changed, not the actual subscriber base.
  4. Did the cohort definition change the comparison? Late-joining customers are now counted from their own start date, so older cohort figures may not line up exactly.
  5. Are you looking at a segment that excludes API-created customers? Those customers lack platform, country, and SDK version, so they drop out of those filters.
  6. Is the Active Customers figure inflated by aliasing? If the same person has multiple linked app user IDs, expect overstatement until aliasing is handled.

Getting started

Charts v3 is available in the dashboard for apps using the supported stores. If your metrics differ after switching, the Charts v3 documentation provides a chart-by-chart breakdown of what changed and why.

Expected metric changes in Charts v3
MetricExpected changeReason for change
Trial Conversion RateLikely decreaseOnly direct trial-to-paid conversions counted
Historical RevenueMay shift between periodsRefunds impact the refund date, not purchase date
Active SubscriptionsMinimal changeCounting method changed, not actual subscribers
MRR/ARRMinimal changeBased on same active subscription count

Expected directions as described at launch of Charts v3; actual movement depends on your data and refund/cohort mix.

Frequently asked questions

Why did my Trial Conversion Rate go down after switching to Charts v3?

Charts v3 counts only direct trial-to-paid conversions. Previously, someone who resubscribed after completing a trial also counted as a trial conversion, so the new figure is typically lower.

Will a refund change last month's revenue?

No. Revenue is recorded on the purchase date, and a refund is subtracted on the refund date. Completed periods stay stable, and you will see a negative revenue entry on the day of the refund.

Why do cohort numbers differ from before?

Cohorting now calculates each customer's lifecycle from when they actually started, then aggregates, rather than defining a cohort by the first and last calendar days of a period. Late-joining customers no longer have early revenue pushed into the next period.

Why can't I segment some customers by platform or country?

Customers created through the API do not have a platform, country, or SDK version attached, so they cannot be segmented or filtered by those dimensions.

Why might Active Customers look higher than expected?

The chart does not yet correct for aliased customers. If the same person appears under multiple linked app user IDs, active customer counts may be overstated.