Your company: Fieldly — a B2B SaaS platform that helps field service teams manage scheduling, dispatch, and invoicing. Customers range from one-person HVAC contractors to mid-size property maintenance firms.
Your role: Data Analyst. You report to the VP of Operations and work closely with the Customer Success and Product teams.
Maya Chen, Head of Customer Success, has just come off a difficult conversation with the CEO. Monthly churn has risen from 2.1% to 3.4% over the past six months — a 62% increase — and the board wants answers at the quarterly review in five days.
Maya stops by your desk with a simple ask:
"I need to know who is churning, why they're churning, and what we should do about it. I have to present this to the board on Thursday. Can you dig into the data?"
You have access to three datasets (provided as CSV files): Download datasets
| FILE | WHAT IT CONTAINS | KEY COLUMNS |
|---|---|---|
| customers.csv | 500 customer accounts — subscription and revenue data | customer_id, industry, company_size, plan_tier, mrr_usd, signup_date, churned (Y/N), churn_date |
| usage.csv | Product engagement for each account over the past 6 months | customer_id, avg_monthly_logins, feature_scheduling_uses_6mo, feature_dispatch_uses_6mo, feature_invoicing_uses_6mo, features_used_count, last_active_date |
| support.csv | Support ticket history per account (last 6 months) | customer_id, tickets_submitted_6mo, avg_resolution_days, primary_ticket_category |
Your job is to diagnose the churn increase, surface the most important insight, and give Maya something she can take to the board.
These constraints reflect real conditions you would navigate in this role. Honor all of them in your submission.
| # | CONSTRAINT |
|---|---|
| 1 | Work only with the provided datasets. You cannot collect additional data, run customer surveys, or access systems not described in the data dictionary. Your analysis must be defensible from what you have. |
| 2 | Your analysis must be reproducible. Document your methodology clearly enough that another analyst on the team could replicate every number in your findings without asking you a question. |
| 3 | Your recommendation to Maya must be actionable within 30 days. No six-month roadmaps, no proposals that require new tooling or budget approval. What can the team do with what they have, right now? |
| 4 | The board presentation context means your key finding must be expressible in two sentences. If it takes a paragraph to state the finding, it is not the finding — it is the methodology. Lead with the conclusion. |
| 5 | You own this analysis. If a number is wrong in your submission, that is your number — regardless of how it was generated. Verify before you publish. |
Diagnose a business problem using messy, multi-source data — join and analyze three separate datasets to identify the primary drivers behind a meaningful change in a core metric.
Translate analysis into a board-ready narrative — distill complex findings into a two-sentence insight and a clear recommendation that a non-technical audience can act on immediately.
Build a reproducible analytical workflow — document methodology with enough clarity that another analyst can replicate every number without a handover conversation.
Recommend under real constraints — propose interventions that are actionable within 30 days using only the resources the team already has, rather than defaulting to long-term roadmaps or new tooling.
On this page
