Case Study 1: Dana's Two-Audience Memo

A composite, fictional-but-realistic scenario. The company, names, and numbers are illustrative; the writing problem is universal.

The situation

Dana Whitfield, a data scientist at a subscription-software company, has finished a churn analysis. Her finding is solid: new customers who take more than three weeks to get value from the product cancel about seven times as often as those who get there within a week. The leak is in onboarding, not pricing or features.

Two people need this finding by Friday. Priya, a fellow data scientist, will review the work before it goes anywhere. Renée, the VP of Marketing, will decide whether to move retention budget upstream. Dana's first instinct — the instinct almost everyone has — is to write one memo and send it to both. Let's watch why that fails, and what she does instead.

The one-size-fits-all draft (what not to do)

Here is Dana's first attempt, written the way the analysis lives in her own head:

To: Renée, Priya
Subject: Churn analysis results

I ran a random forest (300 trees, 5-fold CV, ROC-AUC 0.81) on the
<90-day new-account cohort (n=4,812) to identify churn drivers. After
controlling for plan tier to rule out confounding, the dominant feature
by importance is time-to-first-value (TTFV). Partial-dependence is
monotonic and steep past ~14 days. Bucketed: <7d TTFV → 3% churn;
>21d → 22%. This suggests onboarding friction, not price or product,
is the primary churn driver. Happy to discuss methodology.

For Priya, this is nearly right — maybe even good. For Renée, it's a wall. Run K-R-A-C on Renée and every variable is mis-set. Knowledge: he doesn't know what a cohort, a random forest, partial dependence, or TTFV is — five undefined terms in the first three sentences, five slammed doors. Role/goal: he needs to decide where to spend money, and nothing here tells him what to do. Action: "happy to discuss methodology" is the opposite of the "approve the budget shift" that the document should drive toward. Context: he's reading this in thirty seconds on his phone between meetings, and the one number that matters (7×) is buried in the sixth line behind jargon he'll never decode.

The draft isn't wrong. It's wrong for Renée. And because it tried to serve both readers, it serves only one.

The fix: two documents, two readers

Dana deletes the combined memo and writes two.

To Priya (the peer): she keeps the methodology-first version above almost verbatim, and adds the thing a peer most wants — the honest caveat. "TTFV correlates with plan tier (r≈0.3); I controlled for it and the effect holds, but I'd value your eyes on the interaction." For Priya, leading with the method and surfacing the confound first is a sign of respect: it says I know your job is to find the hole, so here it is. Jargon stays. Nothing is softened.

To Renée (the decider): she throws out the structure entirely and rebuilds for a thirty-second skim.

To: Renée
Subject: Why we're losing new customers — and the fix (1 number)

Renée — short version, detail below if you want it.

The problem: we're losing nearly 1 in 4 new customers, and it's getting
worse (churn nearly doubled in two quarters). It isn't price and it isn't
the product. It's onboarding.

The number that matters: customers who reach their first real "win" in
the product within a week cancel at 3%. Those who take longer than three
weeks cancel at 22% — seven times higher.

The recommendation: move retention spend upstream into the first-week
experience. The leak is week one, not renewal. I'll put a one-page plan
in front of you by Friday.

Same finding. Almost no shared sentences. "Random forest," "cohort," "TTFV," "partial dependence," and "ROC-AUC" are all gone — not because Renée is dim, but because none of them can change his decision. What survives is the one vivid number (7×) and a clear action.

What Dana learned

The lesson isn't "write simpler for executives." It's that the finding is not the document. One result, depending on who reads it, becomes two artifacts that share a fact and almost nothing else. Trying to merge them produced a memo that lost Renée to protect a rigor Priya didn't even need spelled out for her benefit — Priya could read the technical version, but Renée could not read the merged one.

The takeaway for your own writing: When two readers genuinely differ, resist the urge to compromise into one document. Two short, correct documents beat one long, half-right one. This is the same Dana you'll meet again in Chapter 27, where we rebuild the Renée memo through three drafts — methodology-first (worst), findings-first (better), recommendation-first (best) — and watch the order of information decide whether the writing works at all.

Discussion questions

  1. Dana added a caveat for Priya and cut all caveats for Renée. Justify both choices using the "Action" variable — what does each reader do with a caveat?
  2. The merged draft was "not wrong, just wrong for Renée." Why is that distinction the entire point of the chapter?
  3. Suppose Renée forwards Dana's memo to the CFO unchanged. Does it survive? What in the memo makes it safe (or not) for that secondary audience?
  4. Rewrite the subject line of the merged draft ("Churn analysis results") for Renée specifically. What makes a subject line audience-adapted?