SkilsMVP
BootcampCapstone · Deliverable 6

Executive Readout & Recommendation

Builds on Topic 10.

What you'll produce

An answer-first executive readout — the one-page (or 3–5 slide) artifact you'd actually walk into FreshBox's quarterly planning meeting and present. It opens with a single-sentence headline conclusion, backs it with the two or three charts that prove it, quantifies the impact in dollars leadership already cares about, makes one concrete recommendation, states honestly what you can't yet prove, and banishes all methodology to an appendix. This is the deliverable that turns five artifacts of real work into a decision — and it's the one a hiring manager remembers, because it proves you can do the thing most analysts never learn: lead with the "so what," not the SQL. Everything from your brief, your data pull, your cleaning log, your findings memo, and your dashboard converges here into the only thing leadership ultimately buys — an answer they can act on before the meeting ends.

Instructions

  1. Write the headline first, in one sentence, before anything else. It must state the answer, not the topic. Not "An analysis of FreshBox retention" but "Revenue is flat because new Monthly-cadence customers in the EU are churning at nearly double the rate of everyone else in their first 90 days, and it traces to a delivery problem we can fix." If you can't say the answer in one sentence, you don't have a readout yet — go back to your findings memo (Deliverable 4). Carry the brief's exact vocabulary into the headline: name the billing cadence (Monthly) and the tenure cohort (first 90 days), not a vague "plan."
  2. Pull the 2–3 charts that prove the headline — no more. Reuse them straight from your dashboard (Deliverable 5): the MRR trend line that shows the flattening, the churn-by-segment bar chart that locates the bleed, and at most one chart that points at the cause. Every chart must earn its place by supporting the headline. Cut anything that's merely interesting. Keep axes zero-based and labeled (Topic 9) — an honest readout cannot rest on a misleading chart.
  3. Quantify the impact in leadership's terms — dollars and MRR, not percentages alone. "Churn rose from 4% to 9%" is an analyst's number; "this is costing roughly $48,000 in lost MRR per month and widening" is a number a Head of Operations can take to the CFO. Show the arithmetic in one line so it's checkable, then state the annualized figure.
  4. Make exactly one concrete recommendation. Not a menu of five options — one clear action leadership can approve in the room, scoped to the segment your data implicates, with a rough cost and the metric that will tell you it worked. A recommendation they can say "yes" or "no" to beats a buffet they'll table for "further discussion."
  5. Write the honest-uncertainty note as its own short line — don't bury it. Name the one thing your analysis correlates but hasn't proven caused the churn (Topic 8), and say what would confirm it (usually a small test or one more data pull). This is what makes leadership trust you: "I found the where and the likely why; here's how we'd confirm the why before betting big."
  6. Move all methodology to an appendix. Your metric definitions (brief), SQL (Deliverable 2), cleaning decisions (Deliverable 3), and the Simpson's-paradox check (Deliverable 4) belong after the recommendation, for the one person who asks. The body of the readout is answer → evidence → impact → recommendation → uncertainty. If methodology appears before the recommendation, you've buried the lede.
  7. Read it aloud as if presenting. It should take under three minutes to deliver the body. Cut every sentence that doesn't move toward the decision. Then sanity-check the thread: the headline, the charts, the dollar figure, and the recommendation must all name the same segment and the same story your earlier deliverables built.

Worked example

(FreshBox — the artifact you'd present in the Q3 planning meeting. Same company, same data, the investigation that began with "something feels off with retention.")

FreshBox — Why Revenue Is Flat, and the One Thing to Fix Prepared for the Q3 FY26 planning meeting · Analytics · Covers 2025-10-01 to 2026-03-31 (Q1–Q2 FY26)

Headline

Revenue is flat not because we're acquiring fewer customers, but because we're losing new Monthly-cadence subscribers in the EU — in their first 90 days — almost twice as fast as the rest of the book, and the cancellations cluster tightly around late or damaged deliveries, a problem we can fix without spending on more marketing.

The evidence (three charts)

1. MRR is flat while new customers keep arriving. Monthly recurring revenue held between $1.02M and $1.05M across all six months even though marketing added ~2,100 net new subscribers over the same period. New revenue is real; it's being cancelled out by revenue walking out the back door. (Line chart, MRR by month, zero-based axis.)

2. The bleed is concentrated, not company-wide. Overall monthly churn rose modestly from 4.2% to 5.1% — easy to shrug off. But broken out by segment, new Monthly-cadence subscribers in the EU (first 90 days) went from 4.8% to 9.3%, while tenured subscribers and Annual-cadence subscribers stayed flat near 3–4%. The company average hid the problem: the small, fast-churning segment is dragging a much larger book that's actually healthy. (Bar chart, churn by billing cadence × tenure × region, zero-based axis — this is the Simpson's-paradox catch from the findings memo, where cadence mix — not country — turned out to be the driver.)

3. The likely cause has a fingerprint. Among these EU new-Monthly cancellations, 61% had at least one support ticket tagged late delivery or damaged box in the 30 days before they cancelled, versus 17% of retained EU new-Monthly customers. The relationship is strong and consistent across both quarters. (Bar chart, % with a delivery ticket: churned vs. retained.)

The dollar impact

The EU new-Monthly segment is ~6,400 subscribers at an average $39/month. The churn increase from 4.8% to 9.3% is +4.5 percentage points, i.e. ~290 extra cancellations per month:

290 subscribers × $39 = ~$11,300 in lost MRR per month, and compounding. Held over a year on the current trajectory, that's ~$135K+ in annualized recurring revenue — and because we keep paying to acquire into a leaking segment, the marketing spend behind it is partly wasted.

This is the number worth putting next to the cost of fixing delivery: the problem is small enough to feel manageable and large enough to be worth a meeting.

The recommendation (one action)

Audit and fix the EU delivery lane for new Monthly-cadence subscribers before spending another dollar acquiring into it. Specifically: pull the carrier-level on-time and damage rates for EU shipments, and run a 6-week pilot switching the worst-performing EU region to the backup carrier we already use in the US. Rough cost: minimal — it's a routing change and one analyst's time, not new headcount. Success metric: the EU new-Monthly delivery-ticket rate falls below 25%, and that cohort's churn returns toward the 4–5% company norm within two billing cycles.

What I can't yet prove (honest note)

The delivery link is a strong correlation, not proven causation. 61% vs. 17% is a clear fingerprint, but it's possible a third factor (e.g., a specific EU region or a pricing change) drives both the bad deliveries and the cancellations. The 6-week carrier pilot is itself the cleanest test — if churn drops when delivery improves, we'll have moved from "likely why" to "proven why" at almost no cost. I'd treat this readout as a strong, actionable signal, not a closed case.

Appendix — methodology (details on demand)

  • Definitions (from the analysis brief): active subscriber = a subscription with status active on the first of the month; churn rate = cancellations in month ÷ active subscribers at month start; MRR = sum of active subscriptions' monthly-equivalent prices (Annual cadence ÷ 12). The brief locked two separate plan dimensions — plan tier (Basic/Family/Premium) and billing cadence (Monthly/Annual); the headline segment is cut on billing cadence × tenure cohort × region, never a merged "plan."
  • Data pull (Deliverable 2): JOIN of customers × subscriptions, GROUP BY billing cadence × tenure × region for churn, monthly MRR trend from subscriptions, support-ticket flag joined from the cleaned CSV. Row counts sanity-checked against the warehouse total (~40,000 active).
  • Cleaning (Deliverable 3): de-duplicated 312 repeated ticket rows, standardized country, plan_tier, and billing_cadence formatting (each into its canonical set), imputed 47 missing cancellation dates to the subscription's last paid date (flagged, not deleted), excluded 3 impossible negative order values. No silent deletions.
  • Why median where used (Deliverable 4): order values are right-skewed by a few bulk orders, so segment summaries use the median; churn rates are unaffected.
  • Simpson's-paradox check: the 4.2%→5.1% company average masks the 4.8%→9.3% rise in the EU new-Monthly cohort; the apparent "US problem" was really cadence mix, not country (per Deliverable 4). Reported the segmented number as the real story.

Questions or the full query workbook on request.

Why this works: a busy Head of Operations gets the answer in the first sentence, the proof in three charts, the cost in a number they can repeat to finance, and a decision they can approve in the room — with an honest flag that keeps you credible for the next analysis. The methodology that took 80% of the effort sits politely in the back, exactly where executives want it.

Rubric

The app's AI scores the learner's submission against these criteria and gives feedback. Levels: Needs work (1) / Solid (2) / Excellent (3). Passing = every criterion at Solid or above.

  • Answer-first headline — 1: leads with topic, process, or methodology; the conclusion is buried or vague · 2: opens with a one-sentence conclusion that states an answer · 3: a single, sharp, decision-ready sentence that names the segment and the likely cause, readable in one breath.
  • Evidence that proves the headline — 1: too many charts, irrelevant charts, or misleading/non-zero axes · 2: 2–3 relevant, honestly-labeled charts that support the conclusion · 3: a tight 2–3 charts that each earn their place — including one that locates the cause — reused coherently from the dashboard.
  • Impact quantified in leadership's terms — 1: percentages only, or no impact figure · 2: a dollar/MRR figure tied to the finding · 3: a checkable dollar impact (arithmetic shown) annualized and framed against the cost of the fix.
  • One concrete, actionable recommendation — 1: no recommendation, or a vague menu of options · 2: one clear action scoped to the implicated segment · 3: one action with rough cost and a named success metric leadership could approve in the room.
  • Honest treatment of uncertainty — 1: overclaims causation or omits any caveat · 2: flags that a key link is correlation, not proven cause · 3: names the specific unproven link and the cheap test that would confirm it, without undermining the recommendation.
  • Coherence and structure — 1: methodology precedes the recommendation, or the thread breaks across artifacts · 2: answer → evidence → impact → recommendation → uncertainty, methodology in an appendix · 3: every element names the same segment and story traced cleanly from brief → data → memo → dashboard → readout.