RICE scoring,
and what actually happens to it on Monday.

The formula, briefly

RICE was published by the Intercom product team in 2016. The acronym is Reach × Impact × Confidence ÷ Effort, and the output is a single number you can sort a backlog by.

The number itself isn’t very meaningful. The ordering it produces is, and the standardized process it forces while you assign each input is the actual product of the exercise. If you take nothing else from this page, take that.

Defining the four inputs (without lying to yourself)

A worked example

Two initiatives on the table, real numbers from a planning session I sat in last year (rounded for the writeup):

Onboarding rebuild. 4,200 new accounts next quarter, Impact 1.0, Confidence 80% (we ran a smaller test on the welcome email and got the lift we expected), Effort 3 person-months.

Pricing-page rewrite. 12,000 unique visitors, Impact 0.5, Confidence 60% (no prior test, but the page is genuinely confusing), Effort 1 person-month.

Pricing wins, roughly three to one, at a third of the cost. That outcome surprised the room, which is the point. We did the pricing rewrite first, shipped it in three weeks, and the conversion lift paid for the next two quarters of onboarding work.

Where it falls apart

I’ve never seen a RICE implementation fail because the math was wrong. They fail in three predictable ways:

Confidence drifts upward. The person who proposed the initiative is the person who scores it, and over a few weeks “60%” quietly becomes “80%” without any new evidence.

Effort is scored by the optimist. Whoever is excited about the idea estimates the work and under-counts the technical debt.

The list goes stale between reviews. Most teams score everything in a quarterly off-site and never touch the numbers again, which turns the ranking into historical fiction.

Where AI actually helps

AI shouldn’t write your strategy, but it should audit your inputs. Three places that pay off immediately:

• Reach auditing. Cross-reference proposed Reach against historical cohorts and flag numbers that are statistically implausible.
• Effort benchmarking. Compare the estimate to what similar work has actually taken, so optimism gets challenged by evidence, not opinion.
• Confidence integrity. Score the written justification behind a Confidence rating and demand more evidence before the number rises.

What I’d skip

RICE isn’t a replacement for a core vision. It’s a tiebreaker between things that already cleared a strategic bar. If you find yourself RICE-scoring a list that contains “rebuild the billing system” next to “change the color of the Save button,” the problem is upstream of the formula and no amount of arithmetic will save the meeting.

I also don’t recommend RICE for early-stage zero-to-one work where Reach is one customer and Confidence is a coin flip. It produces numbers that look rigorous and aren’t. Use it when you have a portfolio of comparable bets.

How Vantage handles it

Every RICE setup I had run before Vantage decayed the same way: the spreadsheet got out of sync with the work. In Vantage, Reach, Impact, Confidence, and Effort are first-class fields on every initiative, the score recomputes when any input changes, and the ranking on the prioritization board reorders in real time. Every edit is versioned with author and timestamp, so a rank change is always traceable to a specific decision.