What are the product implications of not owning your own data?

Bramwell Apparel is a clothing brand that sells almost all of its volume through one large marketplace platform. Delphine Marrow runs growth and marketing there. She kept a private spreadsheet, built from scraps of order data, that was the only thing standing between Bramwell and total blindness about its own repeat customers. Then she went on medical leave for three weeks.

How to answer this, stage by stage

Nobody is scoring whether you can define "data ownership." They're scoring whether you can name the exact lever someone else is holding over your product.

Let's learn

Here is what happens when your only channel to your own customers runs through someone else's system, and that someone owes you nothing.

Bramwell Apparel sells clothing almost entirely through one large marketplace platform. About 78 percent of its revenue flows through that single channel. Of the customers who buy there, only 4 percent are independently identifiable to Bramwell itself, meaning Bramwell has an email, a name, anything it could use to reach them directly, without the platform's permission.

The five letters, held up as one page. Ability to contest is the step this question is really testing.

Delphine's own private workaround had been the closest thing Bramwell had to visibility: a spreadsheet cross-referencing order-confirmation emails, which occasionally showed a customer's address in a shipping notice, against ad spend, to guess at who the repeat buyers actually were.

Here's the turn: the real problem was never Delphine's spreadsheet being imperfect. The real problem is what happens the day the one person who maintains it isn't available, because at that point Bramwell doesn't have an imperfect view of its customers. It has no view at all.

At its worst, a product Bramwell built its whole growth motion around turns out to depend entirely on one person's improvised, undocumented workaround, and the platform hosting all of it has no obligation to notice, explain, or wait.

What I would leave alone: I wouldn't try to build a full alternative sales channel to replace the marketplace platform entirely. Seventy-eight percent of revenue is a real dependency to reduce over time, not one to abandon overnight by walking away from the channel that built the business.

The lesson: a private workaround can feel like it's solving the data-ownership problem. It's actually just hiding how bad the problem is, right up until the one person running it disappears for three weeks.

Now here is the same thing as a story

The short version above is what you'd say defending an investment in a first-party channel to a skeptical finance team. Read this one for how a private spreadsheet quietly became a company's only real asset, with nobody deciding that on purpose.

Delphine built her workaround eighteen months ago, mostly out of frustration. The marketplace platform's dashboard showed total sales, total returns, total reviews, and nothing else. No way to see who bought twice. No way to email a customer directly. No way to know, when a listing's ranking dropped, whether it was the algorithm, a competitor, or bad luck.

Two people looking at the same outcome. Only one of them can actually reach the lever that controls it.

So she started cross-referencing whatever scraps she could find: a shipping notification that occasionally leaked a customer's email, an ad-spend report, a returns log. It wasn't much. It was more than anyone else at Bramwell had, and over eighteen months it quietly became the company's entire picture of its own repeat customers.

The third step is where every ranking question at Bramwell has always dead-ended, long before Delphine ever went on leave.

Then a rival brand ran an aggressive review campaign against Bramwell's top-selling product, and the platform's ranking algorithm, reacting to a sudden shift in review volume and sentiment, dropped that listing from its usual third position toward the middle of the second page within two weeks. Nobody at Bramwell could ask the platform why, exactly, or get a straight answer about what would fix it.

That same week, Delphine went on unplanned medical leave for three weeks. Her manager knew the spreadsheet existed. Nobody else had ever needed to open it.

Three weeks where the one document that could have triggered a win-back campaign sat on a laptop nobody else could open.

The marketing team could see the ranking dropping in real time. They could see the top product's sales sliding along with it. What they couldn't do was reach the specific repeat customers who might have bought again anyway, out of loyalty, because the only record of who those customers were lived in a file only Delphine had ever opened.

The two things that actually mattered most for growth sat in the one corner Bramwell never controlled.

When the spreadsheet was first built, someone said, "Delphine's just keeping her own notes, that's fine, whatever helps her do her job," and it sounded completely reasonable, since at the time it really was just a personal habit, not a company system.

Rerun the same three weeks with a required, incentivized warranty-registration flow already in place for two quarters: the marketing team pulls the list of repeat customers for that product in minutes, runs a win-back email offer the same week the ranking drops, and the two extra weeks of lost sales never happen, because reaching those customers never depended on any one person being at her desk.

What I'd tell myself, watching a company's entire customer relationship live inside one person's private file: the workaround was never the problem. It was evidence of a real gap that the company had quietly agreed to let one employee's goodwill paper over, instead of fixing it directly.

GUARD, the imbalance that would have been visible on day oneNot a warning about "vendor risk" in general. GUARD is what forces you to name who holds the lever and who's just hoping it doesn't move.

The recap, one line per letter: groups is the platform against Bramwell, unequal is Bramwell carrying all the risk with none of the control, ability to contest is having no explanation, no export, and no real exit, reduce is a required first-party registration flow, and detect is watching independent-identification share and the platform's own terms on a fixed schedule.

And if you want to be sure it really works, try it somewhere elseSame five letters, a school district instead of a retail brand. Different flip family entirely, the same missing lever.

Radovan Petrusek leads data for Ashgrove Unified School District, which relies entirely on a third-party ed-tech platform for student assessment data, with no independent copy kept anywhere in the district. Mapped onto GUARD: groups is the vendor, which controls the export format, the retention period, and the contract terms, against the district, which has no fallback if any of those change. Unequal is that a vendor outage or policy change costs the vendor nothing while it can cost the district a full semester of assessment history. Ability to contest is that neither the district nor, further downstream, individual teachers and students, ever agreed directly to the vendor's data terms, and none of them can appeal a silent processing error. Reduce is a scheduled, mandatory local export of raw assessment data, independent of the vendor's own dashboard. Detect is running a manual spot-check on a sample of exports every month, comparing vendor totals against the district's own attendance and roster records.

The flip here is verification, not workaround. District IT staff used to spot-check that automated weekly exports had actually completed correctly, until a full year of clean, error-free exports passed and the checking quietly stopped, replaced by blind trust that the pipeline simply worked. A silent vendor-side bug then corrupted three months of assessment exports with no error message at all, and nobody at the district noticed until a state audit flagged the gap, precisely because the one habit built to catch a rare, quiet failure had been retired right when it was needed most.

The vendor held every lever that mattered. The district held a dashboard it had stopped double-checking.

Swap the trigger and it still runs.
Speed: an interviewer caps you at sixty seconds. Say "build a required first-party channel, and track how much of your revenue depends on someone else's platform," and stop.
Cost: no budget to build a full alternative sales channel right now. Say so honestly, and start with the cheapest possible independent contact point, even a simple registration card, rather than nothing at all.
The platform actually treats sellers well, for real: even a fair, well-run platform can change ownership, get acquired, or shift strategy. A good relationship today doesn't change who holds the lever tomorrow.

Where people run it wrong.
They treat a platform's good track record as a reason not to build any independent channel at all.
They let one employee's personal workaround quietly become the company's only real system, with nobody officially responsible for it.
They wait for a suspension notice or an outage to discover how dependent they'd become, instead of tracking the dependency as it grows.

How to use it live. The moment an interviewer asks about not owning your data, ask yourself: who actually holds the lever over my customer relationships, and what happens to my product the day they pull it without warning? Name both sides, and the rest of the answer follows.

Flashcards (tap any card to flip it)

Check yourself Score: 0 / 0