Reach Isn't a Guess: Scoring a Roadmap with Real RICE Numbers

Marketing Academy · Field Work●Growth Marketing

All projects·From: ICE & RICE Prioritization

MiniForecast· 25 minutes

Robinhood

Objective: Score three competing roadmap features with RICE using real analytics-sourced Reach numbers, in person-weeks Effort, and forecast which one wins an engineering quarter.

Robinhood's product team has one engineering quarter and three candidate features: a power-user charting tool, an onboarding simplification, and a settings redesign. Reach must come from GA4, not memory.

Pull real 90-day Reach numbers for each surface, score all three with RICE, and forecast which feature the data supports shipping first.

Before you start

What you'll need

Free path (everything below is enough to finish)

Google Analytics 4

FreePull the real 90-day Reach number for each competing surface

Free, and the lesson explicitly warns against estimating Reach from memory

Google Sheets

FreeBuild the RICE table and compute the final scores

Transparent formulas for a roadmap review audience

Paid upgrades (optional, faster/deeper)

Mixpanel(optional)

FreemiumCross-check Reach with event-level data when GA4's page-path filtering is too coarse for a specific in-app surface

Useful when the three features live inside one screen and GA4 pageviews can't separate them

The process

2 steps

Step 01 of 02

RICE = (Reach x Impact x Confidence) / Effort

The most common RICE mistake is estimating Reach from memory; pulling the real 90-day session or user count from GA4 for the specific surface prevents the backlog from skewing toward features the team personally uses.

GA4 shows: the charting tool's power-user segment had 1,900 unique sessions in 90 days; the onboarding flow had 38,000; the settings page had 6,200. Charting scores Impact 3, Confidence 100%; onboarding scores Impact 1, Confidence 80%; settings scores Impact 0.5, Confidence 50%. What is each RICE numerator (Reach x Impact x Confidence), before Effort is applied?

Google Analytics 4— Filter GA4 to each surface's page path, set the date range to the trailing 90 days, and record unique users.

Procedure

Charting tool numerator: 1,900 x 3 x 1.0 = 5,700
Onboarding numerator: 38,000 x 1 x 0.8 = 30,400
Settings numerator: 6,200 x 0.5 x 0.5 = 1,550
Note that onboarding already leads by a wide margin before Effort is even applied

Sample output

GA4, trailing 90 days
  Charting tool: 1,900 sessions -> numerator 5,700
  Onboarding: 38,000 sessions -> numerator 30,400
  Settings: 6,200 sessions -> numerator 1,550

Healthy

Onboarding's real Reach number, pulled from GA4, dominates the ranking before Effort is even factored in, matching the lesson's point that Reach usually decides close calls.

Unhealthy

The team estimates 'the charting tool feels like it touches a lot of our best users' instead of pulling the actual 1,900-session number, and overrates it.

What this means

A 20x gap in real Reach (38,000 vs 1,900) is exactly the kind of gap ICE would have missed entirely, since ICE has no Reach term at all.

So what do I do about it?

Symptom	Action	Effort
A power-user feature keeps winning debates on 'feel' alone	Require a GA4 screenshot of the actual 90-day Reach number before any feature enters the RICE table	5 min

YouYou can do this yourself, no engineering access required.

Step 02 of 02

Effort must be scored in person-weeks, not days

Effort is person-weeks: one engineer for one week equals 1. Scoring in days instead flattens the denominator and makes every idea look artificially cheap.

Charting tool needs 1 engineer for 6 weeks (Effort 6); onboarding needs 2 engineers for 2 weeks each (Effort 4); settings needs 1 engineer for 1 week (Effort 1). Using the numerators from Step 1, what is the final RICE score for each, and which feature does the data support shipping this quarter?

Google Sheets— Add an Effort column in person-weeks next to the Step 1 numerators and divide.

Procedure

Charting tool: 5,700 / 6 = 950
Onboarding: 30,400 / 4 = 7,600
Settings: 1,550 / 1 = 1,550
Rank descending: onboarding wins by roughly 5x over the next-highest score

Sample output

Final RICE scores
  Onboarding: 30,400 / 4 = 7,600  <- ships this quarter
  Settings: 1,550 / 1 = 1,550
  Charting tool: 5,700 / 6 = 950

Healthy

The team commits the quarter to onboarding, with settings as a plausible next pick and the charting tool deferred, not killed.

Unhealthy

The team splits the quarter across all three features because each has a vocal internal advocate, diluting engineering time across all of them.

What this means

RICE's job in a roadmap review is to make the trade-off explicit: onboarding's Reach advantage outweighs even a 4x higher Effort cost than settings.

So what do I do about it?

Symptom	Action	Effort
Leadership wants to fund all three features 'a little bit' this quarter	Present the RICE-ranked table and recommend fully funding the top score before partially funding the rest	5 min

YouYou can do this yourself, no engineering access required.

Final deliverable

A RICE-scored roadmap table for all three features with real GA4 Reach numbers, and a one-line recommendation for the quarter.

See a reference example

Sample output

Snowflake data-platform roadmap, RICE pass (excerpt)

Query builder redesign: Reach 24,000, Impact 1, Confidence 80%, Effort 5 -> 3,840
Admin console polish: Reach 2,100, Impact 2, Confidence 100%, Effort 2 -> 2,100
Recommendation: fund the query builder redesign this quarter, its Reach advantage outweighs the admin console's higher Confidence

Success criteria

You're done when you can:

Uses real Reach numbers rather than estimates for all three features
Correctly scores Effort in person-weeks
Computes the final RICE score and states a single clear recommendation