ARCY AI
BETA
Platform

At-risk

Find the accounts most likely to churn, see the ARR exposed, and get a recovery game plan for each one on the At-risk page.

The At-risk page ranks accounts by churn likelihood, shows the revenue exposed, and gives each a concrete recovery game plan. It reads behavior ARCY already sees plus the planValue you pass to ARCYProvider. Nothing extra to connect.

What you need

At-risk works from behavioral signals ARCY collects on its own. To put a dollar figure on each account, pass plan and planValue on ARCYProvider, the same traits used for Metrics and Revenue Tracking. planValue is the current monthly value in USD for that account. Use organizationTraits for B2B products.

Without planValue, the page still ranks risk and shows recovery plans. The ARR figures show a dash instead of a number.

Who can see it

The At-risk page is limited to admins and managers, since it exposes account-level churn risk and revenue.

The two figures at the top

  • At-risk organizations is the count of accounts above your high-risk threshold.
  • Predicted churn (30d) is the count of accounts predicted to churn in the next 30 days, with the estimated ARR at risk shown underneath.

Both figures come from the nightly analysis, so they are stable within a day.

Reactivation

Next to those figures are two more that answer a different question: not who is slipping away, but who came back. Reactivation looks at accounts that had gone quiet, idle for 14 or more days, the same threshold that flags an account as at-risk, and reports how many resumed real activity in the last 30 days.

  • Reactivation rate (30d) is the share of previously-idle accounts that returned with a substantive session in the last 30 days. A substantive session means real usage, not a bare page load. Underneath it shows how many accounts were eligible to come back, so a small denominator is visible rather than hidden.
  • Reactivated revenue (30d) is the combined planValue of the accounts that came back, so you can put a dollar figure on recovered activity, not just a headcount.

An account counts once per 30-day window even if it goes quiet and returns more than once, so an unstable account does not inflate the numbers. The dollar figure is captured at the moment each account returned, so a later plan change does not rewrite past reporting.

Reactivation appears on both the Organizations and Users tabs, scoped to whichever you are viewing.

Until your app has a full 30 days of history, the reactivation figures show a "Collecting data" state instead of a rate, so you are not shown a misleadingly low number before there is enough history to compute one.

Reactivation describes accounts that returned. It does not claim ARCY caused the return.

The reactivated accounts list

Below the at-risk table, the Reactivated organizations (or Reactivated users) section lists the specific accounts that came back, so you can follow up individually. Each row shows the account identifier, the date it reactivated, and the plan value captured at that time. When no accounts have returned in the window yet, the section says so plainly.

The organizations table

Switch between the Organizations and Users views with the toggle at the top right of the page. B2C products, which have no organization dimension, see only the Users view and no toggle.

Each row is an account that currently meets the at-risk threshold, ordered by churn risk. The columns are:

  • Organization with its account identifier.
  • Plan for that account.
  • Users at risk out of the account's total users.
  • Adoption as a percentage.
  • Days idle since the account was last active.
  • Est. ARR exposed, from planValue.
  • Churn risk as a bar and percentage.
  • Signals that flagged the account, such as going idle or dropping adoption.

Use the search box to find an account by name, and the Plan filter to narrow the table to a single plan. The filter only lists plans that appear in the current data.

The account detail

Click any row to open a side panel for that account. It has four parts:

  1. Estimated ARR at risk for the account.
  2. Risk signals in plain language, the behavior that put the account on the list.
  3. Recovery game plan, a short ordered set of steps. Each step names an owner (customer success, account executive, an automated workflow, or product), when to act, and the outcome that step usually achieves.
  4. Expected outcome with a confidence level, so you know how strong the recommendation is.

The recovery plan is generated the first time you open an account and is reused after that, so opening the same account again is instant.

If a plan cannot be generated, the panel shows a short message with a retry button instead of leaving the section blank.

The users table

The Users tab lists individual at-risk users, ordered by churn risk. Before the nightly job has ever run for your app, this tab shows a "Collecting data" message. Once the first dataset is processed, the table appears. The columns are:

  • User with the display name and the underlying user identifier.
  • Plan the user is on, shown as a colored tag. Users with no plan show a dash.
  • Adoption as a bar and percentage.
  • Churn risk as a bar and percentage.
  • Stage in the activation lifecycle: signed up, activated, retained, or at risk.
  • Signals that flagged the user, such as going idle or dropping adoption. Up to three are shown, with a count for any extra.

Search finds a user by name or identifier within the current page. The Min churn risk buttons narrow the list to Any, above 40 percent (the default), or above 70 percent; changing it reloads the table from the start. Paged 25 users at a time.

The user detail

Click any row to open a side panel for that user. It shows the user's plan and stage, their adoption and churn risk, and the risk signals in plain language. If a user has no active signals, the panel says so.

When figures refresh

Risk scores, counts, and the table are recomputed by a nightly job, so the numbers are stable through the day. Recovery plans are generated on demand when you open an account.

On this page