Write the section of a strategy doc that argues for a data investment with no immediate feature payoff.

Picture the strategy doc before anyone has argued for this line item. Wren Relief Network runs AskWren, a text line that matches people in crisis with nearby food banks and shelters. Marisol Adeyemi is Head of Product, and the doc she's writing has to convince a skeptical board that the next sprint should go to a data pipeline nobody will ever see, instead of the Spanish-language rollout everyone is already asking for.

How to answer this, stage by stage

Nobody is scoring your writing style. They're scoring whether you can make an invisible cost feel as real as a missed launch date.

Let's learn

Here is what happens when the number on your dashboard and the truth on the ground quietly stop being the same thing.

AskWren launched two years ago handling about 120 texts a day: someone in crisis writes in, the model matches them to a nearby shelter or food bank, and a reply goes out with directions and a phone number. Today it handles about 900 texts a day. Back at 120, Renata Sokol, the crisis-line coordinator, personally called back a sample of harder cases within 48 hours, just to see if the resource had actually helped. At 900 a day, she can't. Somewhere around month fourteen, she quietly stopped.

None of these four parts exist yet. That's the whole gap the doc is arguing to close.

Here's the turn: "message matched" kept reading 98 percent the entire time. Nobody had a number for "message that actually helped," because nobody had ever built the pipeline that would tell them the difference.

At its worst, an organization can look flawless on paper for years while a fifth of the people it's supposed to help are quietly being sent to a door that won't open, and the only thing standing between "looks fine" and "front-page story" is which week a reporter happens to call.

What I would leave alone: I wouldn't delay the Spanish-language rollout by more than a sprint. That's a real, visible cost to real people waiting for it, and PICK only wins here because the hidden cost is worse, not because visible costs never matter.

The lesson: a metric that only counts what your system did, never what happened to the person after, will always look better than it deserves to. The gap doesn't show up on a dashboard. It shows up in a news story, months late.

Now here is the same thing as a story

The short version above is what you'd write into the doc itself. Read this one for how a sensible promise quietly turned into a blind spot over a year.

Her name is Renata Sokol. She has worked crisis intake for six years, four of them at Wren. Ask her what a caller needs and she'll usually know before they finish the first sentence.

When AskWren launched, Renata treated it as a second pair of hands, not a replacement for her judgment. Every evening, she'd pull ten of the day's harder matches and call the person back: did the shelter have a bed? Was the food bank actually open? For the first year, at 120 texts a day, that was easy to keep up with.

For most of that year, the good months were genuinely good. Then AskWren's reach grew, word spread, and daily volume climbed past 900. Renata kept trying to call back her ten hardest cases every night. Then it was five. Then, without ever deciding to, she stopped.

One box is small and easy to defend in a meeting. The other one is the whole reason this doc exists.

Nobody at Wren decided, on any single day, that referral quality no longer needed a human check. It just quietly stopped being anyone's job once the one person who used to do it for free ran out of hours in a day.

The gap widened for eleven months before a local reporter, working on a piece about a shelter that had closed two months earlier, found that AskWren was still texting people its address. The story ran the same week the board was reviewing next quarter's roadmap.

A three-week delay is a scale that balances back out. A missing quarter of ground truth never does.

Two years earlier, when a leader first told the board "every roadmap item ships something a user can see," it sounded like the right kind of discipline for a young nonprofit trying to prove it could deliver. Nobody meant it to become a rule against ever funding something invisible.

Three sentences, and the whole doc section is really just these, said plainly.

Rerun the same year with the pipeline built from the start: a 48-hour automated text checks in on every referral, a lightweight partner API confirms real-time capacity, and Renata reviews a flagged sample instead of trying to cover all of it herself. The reporter's story never runs, because the closed shelter gets flagged and pulled from matching within two days of closing, not two months.

What I'd tell myself, reading that reporter's first email to our press line: the promise to always ship something visible was never wrong to make. It was wrong to leave standing long after the org had outgrown the one person quietly making it true for free.

PICK, the sentence that goes in the docNot a lecture on why data matters. PICK is what turns "we should invest in data" into an actual, defensible line item.

The recap, one line per letter: position is invest in verification, delay the feature by three weeks; impact is Spanish-speaking users waiting versus people sent to closed doors with nobody measuring it; cost asymmetry is a bounded delay against an unrecoverable truth gap; and kill criteria is 90 percent sampled success held for a quarter, which ends the investment on purpose instead of letting it run forever.

And if you want to be sure it really works, try it somewhere elseSame four letters, a funeral home's scheduling tool instead of a crisis line. Different flip family entirely, the same quiet promise breaking.

Thornwell Funeral Home uses a small AI tool that drafts service timelines and staffing plans from a family's intake form. Mapped onto PICK: position is invest a few weeks in tracking whether families' actual requests changed mid-process, instead of shipping a requested online payment feature next. Impact is families waiting slightly longer for online payment, against grieving families whose last-minute changes never make it back into the model because nobody logs them. Cost asymmetry is a payment feature delay that ends the day it ships, against silently repeating the same scheduling mistakes on every family whose needs shift, because the system never learns what changed and why. The flip here is delegation, not abandonment: Wendell Ashgrove, the senior director, used to let junior staff run the intake tool unsupervised once it proved reliable. When it started missing late changes to services, he started re-reviewing every single intake himself, and two people were now doing the work one tool was supposed to do alone.

The same three branches work for a funeral home's intake changes as they do for a crisis-line referral.

Swap the trigger and it still runs.
Speed: an interviewer caps you at sixty seconds. Say "pick the cost that's hidden and can't be undone over the one that's visible and ends on its own," and stop.
Cost: no sprint to spare for either. Say so honestly, and ship the smallest possible version, an automated 48-hour text with no human review yet, rather than nothing at all.
The feature turns out cheap to delay for real: if the visible feature can slip two weeks with zero real-world cost, that's a legitimate reason to invest in the hidden gap first, not a shortcut you're taking.

Where people run it wrong.
They treat "ship something visible every sprint" as an unbreakable rule instead of a promise that made sense at one size and not another.
They let a proxy metric stand in for the real outcome forever, instead of setting a date to go check whether the proxy is still telling the truth.
They invest in a data pipeline with no kill criterion, turning a smart bet into a standing tax nobody ever revisits.

How to use it live. The moment someone asks you to defend an invisible investment, ask yourself: which of these two costs can I actually undo later, and which one compounds quietly while I wait? Name the one that compounds, and the doc practically writes itself.

Flashcards (tap any card to flip it)

Check yourself Score: 0 / 0