How to score a backlog with RICE, step by step

Knowing the RICE formula and running it across a real backlog are different skills. Here's the workflow that keeps it honest and fast.

1. Gather every candidate first

Before you score anything, assemble the full list. The most common failure is ranking an incomplete backlog — you carefully score eight features and miss the one that mattered. Pull from everywhere: customer feedback themes, support ticket clusters, sales-blocking gaps, competitive holes, and your own ideas. Get them into one list with a one-line description each.

2. Set definitions before you score

Agree on what "a quarter" means for Reach, and what each Impact level looks like for your product, before you touch a single item. If you define impact mid-way through, the early items aren't comparable to the late ones.

3. Score in one sitting

Do the whole list in one focused pass. Scoring spread across days drifts — your sense of "high impact" on Monday isn't the same as Thursday. Go item by item: Reach (a number), Impact (the scale), Confidence (be ruthless — most things are 80% at best), Effort (ask engineering for a rough person-month figure rather than guessing).

4. Sort and sanity-check

Now sort by score and read the ranking against your instinct. The top three and bottom three should mostly feel right. Where the model and your gut disagree, that's the valuable signal — investigate it. Either your gut is carrying information the scores missed (add a factor, or fix an estimate), or the model just corrected a bias you didn't know you had.

5. Turn it into a plan, then revisit

A ranked list isn't a roadmap yet — it's the input. Take the top items into your phases (Now / Next / Later) and sequence them against capacity and dependencies. Then put a recurring hold on the calendar to re-score: reach changes, effort estimates sharpen, and last quarter's number-two is often this quarter's number-six.