Weekly Recharge subscription health report in Slack

On a cron schedule, every Monday at 8am, build a deterministic weekly subscription health report for my Recharge store. This is a pure reporting workflow: no interpretation, no commentary, no AI summarization. Every figure is computed the same way every week so the numbers are directly comparable across runs.

Reporting window: the seven full days ending the day before the run, so a Monday run covers the previous Monday 00:00:00 through Sunday 23:59:59 in the store's timezone. Compute the window once at the start and reuse the exact same boundaries for every call, so subscriptions, charges, and orders all describe the same seven days.

Pull the data from Recharge with List Subscriptions, List Charges, and List Orders. Use List Subscriptions twice: once for status active filtered to subscriptions created inside the window (these are the week's new subscriptions), and once for status cancelled. Recharge does not offer a reliable cancelled_at filter, so for the cancelled pass request the window using updated_at_min and updated_at_max, then keep only records whose cancelled_at field actually falls inside the window. Use List Charges three times, once each for status success, error, and queued, scoped to the window. Use List Orders for the same window.

Pagination matters here and must be handled correctly or the numbers will be silently wrong. Recharge list responses are cursor based, returning next_cursor and previous_cursor. limit defaults to 50 and maxes at 250, so set limit to 250. Critically, a cursor cannot be combined with any other filter: send the filters only on the first request of each set, then send only cursor and limit on every following page. Keep following next_cursor until it comes back empty, and accumulate the records. The 2021-11 API version removed total record counts from list responses, so never look for a total field. Every count in this report is derived by counting the records you actually paginated through. Send the X-Recharge-Version: 2021-11 header on every request.

Compute the week's figures from the collected records. New subscriptions is the count of subscriptions created in the window. Cancellations is the count of subscriptions cancelled in the window. Net subscriber change is new subscriptions minus cancellations, and may be negative. Orders is the count of orders in the window. Successful charge revenue is the sum of the total price of every success charge. Failed charge value is the sum of the total price of every error charge. Failure rate is failed charge count divided by the count of charges actually attempted, meaning success plus error, expressed as a percentage with one decimal place. Queued charges are counted and reported separately but deliberately excluded from the failure rate, because they have not been attempted yet. If there were no attempted charges at all, report the failure rate as 0.0% rather than dividing by zero.

Recharge money fields are decimal strings in the store's currency, for example "12.99", not integer cents. Parse them as decimal numbers, sum them, and round the totals to two decimal places. Do not multiply into cents and do not assume any fixed currency symbol.

Before writing anything, get last week's figures for the comparison by reading the tracker spreadsheet with the Google Sheets Get Values operation, taking the last populated row. That row is the previous run's output, which makes the sheet the source of truth for the trend and avoids re-querying Recharge for an older window. If the sheet has only a header row, or no rows at all, treat this as the first run: skip the comparison entirely and note in the Slack post that there is no prior week yet.

Append one row per week to the Google Sheets tracker using Append Values, so the history accumulates rather than being overwritten. Columns in order: week start date, week end date, new subscriptions, cancellations, net change, orders, successful charge revenue, failed charge value, queued charges, failure rate. Use valueInputOption USER_ENTERED so dates and numbers land as real dates and numbers and stay chartable. If the sheet is empty, write the header row first, then the data row.

Then post the same figures to a Slack channel using the Slack Bot Send a Message operation. Keep it to one compact block, with each number followed by the prior week's value in parentheses and a direction indicator, up, down, or flat. Order the lines: new subscriptions, cancellations, net change, orders, successful charge revenue, failed charge value, and failure rate last so it reads as the health metric. For example: "New subscriptions: 142 (128) up" on its own line, ending with "Failure rate: 3.2% (4.1%) down". Lead with a one line header naming the date range, such as "Recharge weekly health, Mar 3 to Mar 9". No narrative, no analysis, no recommendations, just the labelled figures.

Respect the Recharge rate limit while paginating: a 40 request leaky bucket refilling at 2 requests per second on standard plans. Pace the requests so bursts stay inside the bucket, and on a 429 wait about two seconds and retry rather than failing the run. If Recharge cannot be reached at all, fail the run without appending a partial row to the tracker, so the sheet never contains a half computed week.