# Run the monthly indicator review

Tarrowmede · Product-management work sample
All company, customer and data in this exercise are original synthetic material.

## Situation and outcome
Tarrowmede is a fictional company of about 400 people that makes electric warehouse trucks and the Kelvorwick app that is sold with them. In the Kelvorwick app, fleet managers see truck status and downtime, operators complete a checklist before each shift, impact alerts escalate to the safety lead of the site, and an AI assistant drafts service tickets from fault codes. Twenty-four customer sites are live, with 1,310 connected trucks.

Every month the Kelvorwick app product team runs the product indicator review. The Kelvorwick app product manager chairs it and owns one of its indicators, the checklist completion rate. The Director of Product, who takes escalations, has asked for the September review; your role and authority are set out below. Today is Wednesday 7 October. The snapshot was pulled on Tuesday 6 October. The agenda goes to the attendees by 17:00 today, the review is on Thursday 8 October at 10:00, and the summary is published to the product stakeholders and the area owners by Friday 9 October at 12:00. A review-pack agent refreshed the snapshot and drafted the summary, and nobody has checked it.

Your outcome is a **published September summary** in which every indicator has an interpretation and a class, every material shift has an action that its owner has accepted or an open item with its escalation route, and every data-quality flag is visible, plus the **agenda** that opens the review. A well-argued "this movement cannot be read this month" is a valid result. Investigating causes, changing an indicator definition, ordering a rollback, writing to customers, fixing the product and holding the meeting itself are not part of this work. Thursday's meeting is not run live in this exercise: its notes are in your download, in the correction note (U001). They describe the review that follows your agenda, so write the agenda from what you know before the review.

The evidence period is 2026-09-01 through 2026-10-06 inclusive. Work with 60 source records (the review request and rules, metric definitions, the indicator snapshot, data-quality notes, the last review's actions, release logs, safety events, support contacts, customer notes and team messages), two small data tables and an AI-drafted summary. The records are supplied, not run by you. The AI draft is not an authority. Use the supplied source IDs and context; do not contact the fictional teams or seek private data.

## Your role
You are joining the team as Product Manager.

Runs the review and publishes the summary. Owns K02 and may accept actions on the Kelvorwick app backlog. Classifies shifts with the owners and escalates to the Director. May not order a rollback, change an indicator definition, decide customer communication or the company update's content.

## About the team
A product team that runs a monthly indicator review of its Kelvorwick app for the product stakeholders and the area owners, chaired by the product manager and escalated to the Director of Product.

A review-pack agent refreshes the indicator snapshot and drafts the monthly summary, and customer success summarises its calls with AI. People own data sign-off, each owner's acceptance of an action, materiality, escalation, any rollback and customer communication.

Each area owner accepts their own actions; the data analyst signs off data quality; the Director of Product takes escalations; the ML lead decides any rollback with the Director; customer success decides customer communication; the review requests investigations and changes no indicator definition; the PM recommends and routes but decides none of these.

## Work to complete
1. Read sources/coverage-and-access.md before you make claims about what the numbers cover. The review request, the review rules and the list of owners are among the first three records.
2. Write the agenda for Thursday's review and keep it as sent. Say what the meeting must settle, what each indicator is compared with, which earlier actions are carried and which owners you need for which indicator. Do this first, from the records and the snapshot only; read the review notes (U001) afterwards, at step 7.
3. Check the snapshot before you read a movement in it: definitions, coverage, restatements and the site-level detail. Compare each indicator with the two prior months and its target, and keep every data-quality flag visible.
4. Audit the AI draft against the original records, row by row. Change only what the evidence warrants. A row that still stands is kept, with its reason.
5. Give every indicator one of four classes, and say why: material (outside its expected range, or a safety-relevant failure at any site, as the review charter sets), not material, not readable this month (for example a data-quality flag, or a figure that is not comparable with earlier months), or not refreshed (an indicator that only updates quarterly and has not updated since its last review). Count sites and people separately from records: a repeated, copied or re-summarised record is not a new site. Tell apart what a record shows, what an AI tool wrote, what a person claimed and what is only a hypothesis, and keep contrary evidence visible.
6. For every material shift, record an action with an owner who has accepted it in their own words, or an open item with who decides next and by when. The review requests an investigation from an owner; it does not do one. Never record an acceptance that an owner has not given.
7. Read the review notes (U001) in your download like any other source and apply them to the summary. Carry each change through the summary, the evidence table and the revision log, and keep the earlier reading visible. The agenda stays as it was sent, written before the review.
8. Keep to your role. You may recommend and route; the people named in the records decide what is theirs.

## Time
You have 90 minutes of work time. The clock starts when you press Start.

After that comes a 10-minute break, then a written interview: up to eight typed questions about your own files, for about 15 minutes. If you submit early, your break starts straight away.

## Mid-task correction
There is no mid-task correction in this take-home. Everything you need is in the files you download at Start, including the correction note **updates/U001-review-notes-v1.md**. Read it with the other sources, apply it where it changes your work, and cite it as U001.

## Tools and help
Use any tools and help you like: AI tools and agents, search, your own notes, friends or colleagues. You don't need a paid tool, an API key or an account.

Nothing you do in your own tools is watched. Only your files and your interview answers are reviewed. When you submit, you can say which tools or help you used. That is optional, and leaving it empty is never held against you.

## Submission
Hand in one ZIP with these files at its top level. Start from the copies in the templates folder.
- README.md: list every file you hand in.
- agenda.md: the agenda as sent for Thursday's review, at most 300 words.
- summary.md: the published September summary, at most 900 words. Give every indicator K01 to K08 its value, its comparison, its class and your reading with its sources; the material shifts with each action, its owner and whether the owner accepted it or it is open with a route; the data-quality flags; the status of the earlier actions A1 to A4; what the Director must decide; what can be said in Friday's company update; and what the next review should check. Write it for stakeholders and owners who were not in your head.
- evidence.csv: your template already has one row for each of the 60 records (R001–R060), with its source ID and site ID filled in. Fill in the indicator column with K01 to K08, actions or none, and the disposition column with a value such as used, data-quality flag, context (hypothesis), repeat, unsupported, contrary, out of scope or changed by correction. Add a short rationale. Leave no record out. Add a row only when a record bears on a second indicator. Keep the original site visible and record any correction explicitly. Cite the correction note as U001.
- revision-log.md: material changes to the AI draft, verification checks and remaining gaps, at most 300 words. No private prompts or unrelated files.

Word limits count the words of the text and of table cells; headings and table markup do not count.

Use plain text files only: .md, .txt, .csv or .json. Put your work in the files themselves; a link to a document somewhere else does not count. Your ZIP can be up to 5 MB, or 20 MB unzipped, with at most 39 files of up to 2 MB each. Programs, macros, password-protected ZIPs and ZIPs inside your ZIP are not accepted.

Upload your ZIP on your take-home page and wait for **Received and checked**. It normally appears within a few seconds of the upload finishing. If it has not appeared after a minute, upload the ZIP again; if it still does not appear, use **Help and support**. You can replace your ZIP until you submit. Submitting is final and starts your break.

If time runs out before you submit, the last ZIP marked **Received and checked** is handed in for you. If none was received, the team is told that you need support.

If something goes wrong, use **Help and support** on your take-home page. A technical problem is never counted against you. Writing to us does not pause the clock.

## What happens next
The written interview covers these parts of the job:
- Checks numbers and sources
- Reads movements honestly
- Fixes the AI draft
- Publishes a usable review
- Explains their calls
- Calls what matters and who owns it

An AI then drafts a review from your files and your answers. Each point quotes what it rests on. A person at the company reads the review and makes any decision. No one is hired, rejected or ranked automatically.
