> ## Documentation Index > Fetch the complete documentation index at: https://docs.atllasx.com/llms.txt > Use this file to discover all available pages before exploring further. # Recovery Analytics > Revenue recovered, the recovery funnel, why members leave, and the proof behind every save. **Analytics** is Revenue Recovery's home page and its report. Before you're connected it walks you through how the product works and what's left to set up; once recovery is running, it tells the whole story: * **Revenue recovered**, in two tiers — **verified** (money that was actually charged: a checkout completed or a renewal/redemption charge collected) and **pending** (the member said yes — or redeemed their code — and their charge hasn't landed yet; shown with the date it collects, and it graduates to verified automatically when the charge happens). A secondary **"≈ \$X/mo ongoing"** line shows the monthly run rate of the saves that are still active — kept separate from the dollar totals, never added into them. * **A members-recovered tile** split by who gets the credit — said yes on a call, came back after our text, or came back on their own — with a note when earlier wins re-cancelled (those are not counted). * **A plain-word recovery funnel** of three stages: members cancelled → we reached them, by an answered call or our offer text → came back after we reached them. A member with no touch at all still counts in the members tile, just not in the funnel. On apps showing the [Money won back](#money-won-back) report below, a **Revenue recovered over time** chart takes this slot instead, because the funnel's numbers are already on the cards above it. * **Why members leave** — drawn from what members actually say on calls. Click any reason to see the members and call recordings behind it. Each reason's percentage is its share of the members who actually told you something, not of every cancellation, so the reasons stay readable however many people left without saying why. * **The members recovered list**, with the proof behind each save. A date-window selector (This week / Last week / Month to date / Last 30 days / All time — the one you pick also shows the exact dates it covers) scopes everything. When an A/B test is running or recently closed, an **Offer performance** card compares the variants here too — see [A/B Testing](/docs/revenue-recovery/analytics/ab-testing). The cancellation reasons are built from what your customers actually said on their recovery calls. There is no fixed list of reasons: the categories are named from your own customers' words, so a dating club sees something like "No matches nearby" while a fitness app sees something else entirely. They are refreshed automatically — new reasons are filed into your existing categories as they come in, and the full set is reconsidered weekly, so categories can merge or split as the picture becomes clearer. Customers who never gave a reason get no bar of their own, and are never grouped under "unknown" or "other" — only what was actually said is counted as a reason, so the totals here can be smaller than your total cancellations. They are not hidden either: the line above the bars states both numbers ("8 of 60 cancellations shared a reason"), and the count underneath it ("52 never told us why") clicks through to exactly those members. ## Money won back Analytics leads with a **Money won back** report computed directly from your billing provider's own records rather than our internal tracking — so the headline is what your store actually charged, verified purchase by purchase. It covers **both billing rails**: subscriptions billed through RevenueCat (App Store / Play Store / RevenueCat-managed web billing) and subscriptions billed directly through your connected Stripe account. If you use both, the headline is one combined total with a per-rail split underneath, and the member table gains a **Billing** column — never two competing numbers. A member counts as won back only when **both** of these are true: 1. **We actually reached out** — they got a call or a text from your win-back workflow. Members we never contacted are never counted (or billed), no matter what they later bought. 2. **They came back and paid** — a paid subscription that started **after** they entered recovery, within the attribution window (21 days by default). Purchases from *before* the cancellation never count, so a member's original purchase can never be claimed as a recovery. For Stripe-billed members whose cancellation we reversed (they kept their existing subscription), the recovery is counted from the moment the cancellation was reversed, and only the money paid **after** that moment counts — earlier payments on the same subscription are never claimed. Recovered money is split by what the member came back **on** — two groups that never overlap and always add up to the total: * **Win-back plan** — they came back on your designated win-back plan (the plan only offered through win-back outreach). These are proven recoveries. * **Other plan** — they came back on any other paid plan inside the window. Real recoveries — the plan itself just doesn't prove it was the outreach. The report leads with four numbers: **Revenue recovered**, **Members reached** (the members we actually contacted — an offer or pre-call text, or at least one dial; being enrolled is not being contacted), **Won back**, and your **Recovery rate** (won back divided by reached). Hover any of those titles for its definition; on the **This week** window they also show how they moved against last week, whenever there's a real prior week to compare with. Under them, a **Revenue recovered over time** chart plots the same money by the day it came back, rolling up to weeks once the range gets long. Atllas's fee is 20% of what's recovered. Every number is auditable: the table below lists each recovered member — what they bought, when, and for how much — with its own search box and an **Export** button that saves exactly the rows you're looking at, and clicking a row opens their full story (calls, texts, recording, transcript). The report is computed from a snapshot of your billing records (the stamp under the headline says exactly when); **Refresh** recomputes it on demand. If one of your billing connections is down at compute time, the report says which side isn't measured rather than presenting a partial number as complete. If your win-back sends the **one-tap apply link** (texted when a member cancels while their plan is still active and the recovery call goes unanswered), Analytics adds a separate **One-tap apply link** funnel: links sent, then tapped "Keep my plan", then kept their plan, with the tap and save rates. It counts only real button taps and the saves that follow, never link opens. A page view is not counted as a person, because a link can be fetched by a preview or security scanner before anyone taps it. The section appears only once you've actually sent apply links. ## Needs attention When something genuinely needs you — for example, your App Store offer-code pool runs empty so members who say yes can't get their discount — ONE small banner appears at the top of Analytics saying what happened, how many members it affects, and what to do. It's dismissible, and a NEW outage always reappears even if you dismissed an earlier one. Calls that would waste a member's yes pause automatically until you add codes, and resume on their own the moment you upload more. Members whose save needs a manual step (for example, they asked to speak to a person) appear as flagged rows on the [Activity log](/docs/revenue-recovery/analytics/activity-log). ## Where you're losing people Before any win-back campaign is running, the main **Dashboard** (home) estimates **where subscribers leave** — never paid, free trial never converted, and paid then left — how many of those people have a phone we can call or text, and links you into win-back setup. Those are high-level estimates from your billing connections; campaign performance and recovered dollars live here, on Recovery Analytics. See [Dashboard](/docs/ai-calling/managing-campaigns/dashboard).