# Investigate a drop in checkout conversion

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

## Situation and outcome
Hennavale Market is a fictional online marketplace for independent home-goods sellers in the US and the UK. Customers check out on an iOS app, an Android app and the web. Checkout conversion, the share of started checkouts that end in an order, fell in the week of 21 September. The VP Commerce asks whether the new delivery estimate caused the drop and decides by 2 October whether to roll it back in the US and whether it joins the UK launch in October (R050).

Your outcome is a findings note for the VP Commerce. Each finding is tied to records, what is established is kept apart from what is still open, and any recommendation is a proposal. A well-argued "the data cannot settle this yet" is a valid finding. The note does not fix tracking, run a test, decide the rollback or the launch, handle an incident or authorise a refund.

The period is 2026-08-31 through 2026-09-28 inclusive. The weekly figures cover four Monday-to-Sunday weeks, 31 Aug to 27 Sep (W1 starts 31 Aug, W2 7 Sep, W3 14 Sep, W4 21 Sep); records dated 28 Sep were collected on the last day. Work with 58 records (funnel extracts, order counts, change logs, data-quality notes, support contacts and team messages) alongside three small data tables and an AI first analysis. The AI analysis is a draft, not an authority. Use the supplied record IDs and source context; do not contact the fictional teams or seek private data or live queries.

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

Own the investigation brief and the findings note to the VP Commerce. Do not state a measurement problem as fact until the data owner has confirmed it: name what you would ask them to confirm. Name payments as the owner of the incident and write down the handoff you propose; you do not contact anyone. Keep any recommendation a proposal. You may propose a test; the VP decides on rollback and the UK launch.

## About the team
A marketplace checkout team answering a decision owner's question about a fall in a key metric.

An analytics agent drafts a first memo when a metric alert fires; a person checks the numbers before a decision owner sees them.

Data owners confirm measurement problems; payments owns incidents and refunds; the VP decides any rollback or launch.

## Work to complete
1. Frame the question: what the VP Commerce must decide, over which period and which slices. Read sources/coverage-and-access.md before you make any claim about what the data covers.
2. Before you explain any change in behaviour, check that the data can be trusted: tracking, the data pipeline, the metric definition and the calendar are the usual suspects. Record a result for each check you run, even when it finds nothing.
3. Split the movement by market, device and funnel step. Line up the timing with the changes and releases in the records, and cross-check a figure against a second source wherever you can.
4. Audit the AI first analysis against the original records. Keep, correct or reject each claim and say why. Keep evidence that points the other way visible, and tell apart a measured figure, a person's claim and a gap.
5. Count records, slices and customers separately. Treat severity separately from frequency. Name the owner who can confirm or act on each item without deciding for them.
6. Apply the correction note (U001) in your download like the other sources. Carry each justified revision through every file and keep the earlier version visible. The revision log is your own account of how you worked.
7. Hand off clearly: findings, open questions with owners, and one proposed next step for the decision owner. A well-supported finding that the evidence cannot decide is a valid outcome.

## 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-confirmations-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 and say how you used the correction note (U001).
- findings.md: at most two pages equivalent (about 900 words). The data checks and their results, where the change sits, findings with what is established and what is open, the proposed next step, and the limits.
- evidence.csv: already has one row for every record, R001–R058, with the record ID and slice ID filled in. Do not retype them or remove rows. For each row, fill in your own hypothesis name, a status and, in one sentence, a rationale. A status such as supports, rejects, context, untestable or excluded is suggested and free text is accepted. If a record bears on more than one hypothesis, copy its row and change the hypothesis. If the correction note changes a record, revise its row in place, set evidence_version to U001 and keep the earlier reading visible in the rationale.
- revision-log.md: material changes, verification checks and remaining gaps. No private prompts or unrelated files.
- questions.md: a short list of open questions with suggested owners and decision boundaries.

Words are counted in the prose of each document you write; headings, tables, the evidence file and the README 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 sources and numbers
- Judges what the evidence supports
- Checks and repairs the analysis
- Hands off to the right owners
- Explains their choices
- Checks the data, then splits the change

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.
