The first week of every month
Your agreement with your partner bank asks for a monthly management information pack. It covers things like customer numbers, onboarding volumes and outcomes, transaction volumes and values, complaints, reviews and incidents. The template is a spreadsheet the partner sent a year ago, with a few tabs added since.
On the first working day of the month, someone in ops or risk starts pulling data. Onboarding numbers from the verification provider's dashboard, transaction totals from a database query an engineer runs, complaints from the helpdesk, review outcomes from the ops sheet, incidents from a Notion page. Each number is typed into the template. Then the numbers do not agree with each other, and the rest of the week goes on finding out why.
Why the pack is so painful
The pack is simple to read and hard to produce. Each figure has a definition, and each source system counts things slightly differently.
- Different tools use different date boundaries and time zones.
- Counts of customers depend on what counts as active, and each tool has its own idea.
- Some numbers need an engineer's query, which competes with sprint work.
- The template changes when the partner asks for something new, and last month's formulas break.
- The person who knows how each figure is built is often the only one who can do it.
Late, inconsistent or dependent on one person
A week of a senior person's time each month is a lot for a small company. Late packs are noticed. Figures that change between months without explanation raise questions from the partner that take more time to answer. And if the one person who knows the process leaves, the next pack is a guess.
What goes into the pack and how figures are defined is agreed between you and your partner. We make the agreed definitions repeatable.
An MI pack that builds itself for review
What we build treats each figure in the pack as a defined query that runs the same way every month.
- Each figure in the partner's template is written down with its definition, source and date boundary, agreed with your ops or risk lead.
- Data is pulled automatically from each source: your database or warehouse, your onboarding provider's API, your helpdesk and your ops tools.
- Figures are calculated in one place with the agreed definitions, so they are consistent across tabs.
- Checks run before anyone sees the draft: totals that should agree, figures that moved sharply from last month, missing data.
- The partner's template is filled automatically, so the format stays exactly as they expect.
- The reviewer gets the draft with the check results and a short list of things to explain, and signs it off.
| Pack section | Typical source | Check before review |
|---|---|---|
| Customer and onboarding numbers | Onboarding provider plus your database | Sign-ups equal approvals plus rejections plus pending |
| Transaction volumes and values | Your ledger or warehouse | Ties to the reconciled provider totals |
| Complaints | Complaints log or helpdesk | Opened, closed and open counts add up |
| Reviews and restrictions | Ops review tool | Counts by type against last month |
| Incidents | Incident log | Every incident has a start, end and summary |
If you already have a data warehouse and a tool such as Metabase or Looker Studio, we build on that. If not, a small reporting database in your cloud account is usually enough.
Month end with the pack automated
On the first working day, the draft is ready. The reviewer reads the check results, looks into the two figures that moved, adds commentary and sends it. When the partner asks why a number changed, the definition and the underlying data are one click away.
When the partner changes the template, the change is made once in the build, not relearned every month.
Is the monthly pack a problem for you?
- It takes days of a senior person's time each month.
- An engineer has to run queries for it.
- Figures from different tabs do not agree without manual fixing.
- Only one person knows how it is built.
- The partner has queried numbers that changed without explanation.