Case Study 1: Dana's Churn Memo, Three Ways

A composite, fictional-but-realistic example. The character (Dana Whitfield) and her stakeholder (Renée Okafor, VP of Marketing) first appeared in Chapter 2; the full three-version treatment is the payoff of the device introduced there.

The situation

Dana Whitfield, a data scientist at a subscription software company, spent two weeks finding out why new customers were canceling. Her finding, stated precisely for herself:

Quarterly churn rose from 5.1% to 9.3% over two quarters, concentrated almost entirely in accounts younger than 90 days. A random-forest model identifies time-to-first-value—the days before a new customer completes a meaningful action—as the dominant predictor. Accounts that reach first value within 7 days churn at 3%; accounts that take longer than 21 days churn at 22%. The onboarding flow, not the product or the price, is the leak.

Her reader, Renée Okafor (VP of Marketing), needs to decide one thing: whether to move retention budget upstream into onboarding. Renée skims on her phone between meetings, has high business context and low patience for method, and her output is a decision. Watch the same finding written three ways.

Version 1 — Method-First (the version that failed)

To: Renée Okafor (VP Marketing) Subject: Churn analysis — methodology and results

Renée, over the past two weeks I analyzed customer cancellation patterns to identify the drivers of our recent churn increase. I extracted all account cancellations over the trailing two quarters, yielding a cohort of 4,812 new accounts, and built a random-forest classifier (300 trees, max depth tuned via 5-fold cross-validation) to identify the most predictive features; the model achieved an ROC-AUC of 0.81 on held-out data. Feature-importance analysis identified time-to-first-value (TTFV) as the dominant predictor; partial-dependence was monotonic and steep past ~14 days. I controlled for plan tier (r≈0.3 with TTFV) and the effect held. Accounts reaching first value within 7 days churned at 3%; those taking longer than 21 days churned at 22%. Figure 2 (attached) shows churn by TTFV cohort. Notebook in the repo—happy to discuss.

What Renée actually did: skimmed it, saw "methodology," met a wall of "random-forest classifier" and "ROC-AUC," and replied: "Send me the bottom line when you have it." Dana had sent the bottom line—it was the second-to-last sentence. Renée never reached it.

The five defects:

1. Method-first ordering — the document follows the order Dana did the work, not the order Renée reads.
2. Credibility-subtracting jargon — "random forest," "ROC-AUC" signal this isn't written for you.
3. The finding is buried and in Dana's units — "22% vs. 3% by TTFV bucket," late and unframed.
4. No "so what?" — no recommendation; the most decision-ready analysis in the company asks Renée to draw her own conclusion.
5. The chart is dropped, not interpreted — "Figure 2 shows churn by TTFV cohort" labels; it doesn't interpret.

Version 2 — Findings-First (better)

The first fix is the big one: turn the document over.

To: Renée Okafor (VP Marketing) Subject: Why we're losing new customers — it's onboarding, not price

Renée, we're losing nearly 1 in 4 new customers, and it's getting worse—churn nearly doubled over two quarters. The cause isn't price and it isn't the product. It's onboarding. Customers who reach their first real "win" in the product within their first week churn at 3%; customers who take longer than three weeks churn at 22%—seven times higher. The leak is in week one, not at renewal.

How we know: this is based on every cancellation over the last two quarters and a predictive model that ranks what drives churn. Time-to-value came out on top by a wide margin, and the pattern holds even after accounting for the fact that bigger accounts onboard more slowly. I'm confident it's a real driver, not noise. Full methodology and the cohort breakdown are in the attached notebook.

Why it's better: the finding hits in the first two sentences, in Renée's units, with the decisive multiplier ("seven times higher") in plain language. The method is two sentences of reassurance, demoted below the finding. The test: could Renée stop after paragraph one and act correctly? Yes.

For many readers, this is the right place to stop. But Renée's job isn't to understand that onboarding drives churn—it's to decide whether to move budget. Version 2 tells her where the leak is; it doesn't tell her to pick up the wrench.

Version 3 — Recommendation-First (best)

The second fix: lead with the recommendation; let the finding support it.

To: Renée Okafor (VP Marketing) Subject: Recommend moving retention budget into week-one onboarding

Renée,

Recommendation: shift retention spend upstream—out of renewal-stage win-backs and into the first-week onboarding experience. I think this is our highest-return move on churn right now, and I'd like your approval to scope it. I can have a one-page plan to you by Friday.

Why: we're losing nearly 1 in 4 new customers, and it's accelerating. The cause isn't price or product; it's onboarding—customers who reach their first "win" within a week churn at 3%, those who take more than three weeks at 22%, seven times higher. The leak is in week one, not at renewal—exactly where our retention budget isn't.

How confident: based on every cancellation over two quarters and a model that ranks churn drivers; time-to-value is the top driver by a wide margin and holds after accounting for account size. Strong enough to act on.

The ask: approve a two-week scope of the week-one onboarding flow. Full analysis attached if useful.

Why it's best for Renée: in the first three lines she has the recommendation, a concrete next step (a one-page plan by Friday), and a dated ask. The finding still appears—it's the "why"—but it now serves the recommendation instead of waiting for Renée to act on it. The method is one line. The busiest possible reader, reading only the bold lead, can decide.

The three side by side

The lesson

The finding never changed across the three versions. Not one number moved. What changed was the order—and the order follows the reader's purpose. Renée's purpose is to act, so recommendation-first wins. Send the same finding to Dana's analyst peer, whose purpose is to audit and reuse the model, and Version 1's method-first ordering becomes correct—because that reader needs the apparatus first. There is no universally best order. There is only the order that matches what this reader has to produce. That is Chapter 2's lesson, paid off with money on the line: being right is not enough; you have to be read.

One more move sits inside the rebuild and is easy to miss. Version 1 ended with "happy to discuss"—a phrase that feels helpful but silently hands Renée the interpretive step that was Dana's to take. Running the "so what?" test on her own finding before writing is what gave Dana the recommendation that anchors Version 3. You cannot lead with a recommendation you never bothered to form.

Back to: Chapter 27 · Exercises · Case Study 2 · Key Takeaways