You Don't Have a Prioritization Problem. You Have a Memory Problem.

Every few months a team decides its prioritization is broken and adopts a new framework. RICE. MoSCoW. Cost of delay. Weighted scoring. Each one promises that this time the roadmap will be rational.

It never is, and the framework is never the reason. You can run flawless arithmetic and still get the wrong answer when the numbers going in are made up.

The frameworks are fine. The inputs are fiction.

Look at what a RICE score is made of:

• Reach: how many customers it affects.
• Impact: how much it matters to them.
• Confidence: how sure you are about the first two.
• Effort: what it costs to build.

Effort, engineering can estimate. The other three are dressed-up guesses about customers.

"Reach: this affects most of our enterprise accounts." Says who? "Impact: high." Based on what? "Confidence: 80%." Eighty percent of what evidence, exactly?

In most teams those numbers come from the same place "the customer said" comes from: someone's memory of some calls, smoothed into a confident input, multiplied by someone else's memory of some other calls. The framework does precise math on the vibes. The output looks rigorous. It is rigor laid over guesswork, which is worse than obvious guesswork, because it hides.

You didn't make the decision more rational. You made it look more rational.

The loudest opinion in the room didn't leave. It learned to fill in the Confidence column.

The real meeting runs on recall

Watch a prioritization meeting closely and it doesn't run on the scores. It runs on who remembers what.

Someone says "customers keep asking for this." Someone else says "really? I haven't heard that." Now it's two memories against each other, and the tiebreaker isn't evidence, it's seniority or volume. The most confident person wins, and everyone calls it data-driven because a spreadsheet was open.

The spreadsheet drove nothing. Recall did. And recall is biased toward the calls you personally sat in, the customers who sounded like you, and the feedback that confirmed what you already wanted to build.

Fix the inputs, not the formula

A prioritization framework is only as good as its inputs, and the inputs are only as good as the evidence behind them. So fix the evidence:

"This affects most enterprise accounts" should be a count, not a feeling. How many of your enterprise accounts raised it, in their own words, in the last ninety days? That is knowable from your recorded conversations. "The biggest pain is onboarding" should come back as fourteen quotes from nine companies, or it should come back as "you've heard that twice and assumed it was everywhere."

When the inputs are grounded, the framework finally does what it promised. The argument stops being whose memory is louder and becomes "given the evidence, what do we build." Same RICE score, completely different meeting, because the Confidence column is now a number you can click into instead of a number you defended.

The cheap part and the expensive part

Adopting a new framework is the cheap, satisfying move. It feels like progress and changes nothing, because you re-pointed the same memories at a new template.

The expensive part, the part that moves outcomes, is making sure every claim feeding the framework traces to real customers saying real things. That used to be impossible at the speed of a planning cycle. It isn't anymore.

---

We built Fragment so the inputs to your roadmap are receipts, not recall, and the framework you argue over is finally doing math on something real.