Turn raw weekly exports into a clean KPI workbook with fact-checked insights
Starts every Monday morning or on demand. Ends with a clean KPI workbook, fact-checked notes and a digest in your team chat.
Bring the step you are stuck on: 30 minutes, a straight answer and the next step. If you would rather have it built, a reply within 5 working days.
The route
7 steps: 4 rule steps, 1 AI step and no person step
Copy, share, discuss
- Copy
Available once this workflow passes its checks.
- Share
todkett.com/automations/operations/weekly-kpi-report/
- Discuss
Opens when the first workflow passes its checks. No GitHub account? Ask about this by form.
How the library, the tracks and the talk work: the community page.
At a glance
- Starts when
- Every Monday morning, or on demand.
- Input
- The newest export files in a Drive folder.
- Output
- A clean KPI workbook, fact-checked notes and a digest in your team chat.
- AI
- AI in one step only: writing the weekly notes
- Rule checks
- 5 checks around the AI
- Tools
- Google Drive, Google Sheets, Slack, OpenRouter
- Last stress test
- No recorded run yet
01Steps
The steps
7 steps: 4 rule steps, 1 AI step and no person step. Every Monday someone rebuilds the numbers for the team or the investors: downloading exports, pasting them into a workbook, fixing broken rows and writing the commentary by hand.
01Trigger
New exports in Drive
The newest export files in a Drive folder.
02Rule step
Check the settings
Check the settings and read the reference tables; stop early if a setting is missing.
03Rule step
Rebuild the tables by key
Build clean usage, access, purchase and reconciliation tables, and write them back by key so history is kept.
04Rule step
Compute the findings
Read the data-quality checks and build a packet of findings computed in code.
05AI step
AI words the notes
AI picks and words 3 to 5 notes from the computed findings.
06Rule step
Fact-check every note
Code fact-checks every note; a failing note is replaced by its template.
07Output
Workbook and digest
The workbook, a chat digest and a run log.
KPI workbook
02Where AI is used
Where AI is used, and what stays with you
AI reads, judges or drafts. Fixed rules check, decide and route. A person approves anything that goes out.
AI step
Step 05Writing the weekly notes
Plain-English commentary. Every number is computed and checked in code before and after.
What the model reads
- A packet of findings that code computed from the week's data, as aggregates only.
What it never sees or decides
- Personal data: emails are hashed when they are read.
- A number it was not given: a note that quotes one is replaced by its template.
- The tables: code builds every one and computes every number.
Model calls: Not counted from the workflow yet. It is counted before this page goes live.
Fixed rules
They check, decide and route. 4 rule steps.
- 02Check the settings
- 03Rebuild the tables by key
- 04Compute the findings
- 06Fact-check every note
A person
No step waits for a person. Human sign-off (ST7) is not needed: stays with you.
03Checks
Checks around the AI
What runs before the model, and what checks its answer after, in plain words. Build these into your version too.
- 01The run stops early if a setting is missing.
- 02Emails are hashed on read; only aggregates go to the AI step.
- 03The data-quality checks are read before any notes are written.
- 04Each note must cite a known finding, use only its numbers, state no cause and hold no personal data.
- 05A failing note is replaced by a template; if the AI step is down, every note is a template.
04The 8 checks
Stress test
The 8 checks we run on every workflow, and you can run too. A check marked Not needed says why.
- Status
- Not yet0 of 6 that apply have a recorded pass
- Last run
- No recorded run yet
- Model review
- Still open
Checks marked Not yet are still open. One of them was seen in the workflow design on 23 Oct 2026; a check shows Passed only once its test run is recorded.
ST8Rules around the AI
Status: Not yet
Seen in the workflow design on 23 Oct 2026. The recorded test run is still open.
0 passed · 2 not needed · 6 not yet
What we testWe show what we tested, never the test data. The example below is made up.
05Test replay
How to test it
Before you trust a workflow, try to break it. The sample record runs the route step by step. Each break test below gets its own replay once its recorded run exists.
- Record
- WEEK-43
- Data
- Sample run, made-up data
01New exports in Drive
One week of Fernway usage, access and purchase files
Done02Check the settings
Every setting present: the run continues
Done03Rebuild the tables by key
Tables rebuilt by key; history kept
Done04Compute the findings
Data-quality checks read; findings computed in code
Done05AI words the notes
Four notes written from the findings
Done06Fact-check every note
Three pass; one quoted a number not in the data and became its template
Done07Workbook and digest
Workbook updated; digest in the team channel
KPI workbookArrived
WEEK-43 reached the last station: KPI workbook. Workbook updated; digest in the team channel.
Sample run on made-up data. No stress-test run is recorded yet, so this plays the example, not a test. Nothing runs live.
The break tests for this workflow
- ST3Not yet
Same input sent twice
No recorded run to replay yet.
- ST2Not needed
Instructions hidden in the input
AI sees only numbers our code computed.
- ST4Not yet
A connected service fails
No recorded run to replay yet.
- ST8Not yet
The AI returns a broken answer
Seen in the workflow design. No recorded run to replay yet.
06Example
The example, in and out
Made-up data, written to show the steps; no recorded run stands behind it yet. Fernway is a made-up small software company, and its address ends in .example, so it cannot belong to a real business.
InputExample
Example on made-up data: one week of Fernway exports, usage, access and purchase files, in a Drive folder.
OutputExample
Tables rebuilt by key and the data-quality checks read. Three notes written from computed findings; a fourth that quoted a number not in the data was replaced by its template. Digest posted to the team channel.
07Limits
When not to automate this
- It describes the numbers. It does not explain why they moved.
- It needs your exports in one place, in a format that stays the same week to week.
08What you need
What you need to run it
| Tool | Access |
|---|---|
| Google Drive | Read one export folder |
| Google Sheets | Read and write the KPI workbook |
| Slack | Post the digest to one channel |
| OpenRouter | One API key with a credit limit, for the notes step |
- Where it runs
- In n8n, self-hosted or n8n cloud.
- Model calls
- Not counted from the workflow yet. It is counted before this page goes live.
- Running costs
- Yours, paid to each service directly. Scope in writing first. A quote before any build. No prices on the site yet.
How to adapt it
Changes we have made or tested on this workflow.
- 01Run it on demand as well as on its Monday schedule.
What changed
- ContentFirst version, written from the 23 Oct 2026 catalog.
09FAQ
Questions people ask about it
Does AI calculate the numbers?
No. Code builds every table and computes every number. AI only picks and words 3 to 5 findings that code already calculated.
What stops a wrong number in the notes?
Code checks each note: it must cite a known finding, use only that finding's numbers, state no cause and hold no personal data. A note that fails is replaced by its template.
What happens if the AI step is down?
Every note falls back to its template, and the report still goes out.
Does personal data reach the model?
No. Emails are hashed when they are read, and only aggregates go to the AI step.
10Related
Where to go next
Use cases
7 steps
7 steps
Can you automate it?
Stuck on your version of this workflow?
Bring the step that does not work. In 30 minutes you get a straight answer and the next step. If you would rather have it built, tell us your tools and the rules you already follow. Scope in writing first. A quote before any build. No prices on the site yet.