FREE TOOL

Score your A/B test ideas before you run them

Most experiment backlogs are a mix of high-signal ideas and noise. This scorecard separates them — using a weighted formula across impact, confidence, reach, effort, and learning value.

SCORE AN EXPERIMENT

Rate each dimension 1–10. The weighted score out of 10 tells you whether to run it, consider it, or skip it.

Experiment details

How much will this move the target metric if it works?
5
NegligibleTransformative
How strong is your evidence that this will work? (data, research, analogues)
5
Pure guessNear-certain
What percentage of your users will this experiment affect?
5
<5% of usersAll users
How much engineering and design time does this require? Rate low effort as high.
5
Weeks of workHours of work
If this experiment fails, how much will you learn about your users?
5
Nothing newFundamental insight
5.0 / 10

Weighted Score

Consider it

Impact

5

×3

Confidence

5

×2

Reach

5

×2

Effort

5

×1.5

Learning

5

×1.5

Compare and rank multiple experiments

EXPERIMENT BACKLOG

Experiments are automatically sorted by score. The top two are highlighted.

No experiments added yet. Score one above and click "Add to list".

Save your backlog

Get a PDF of your scored experiments with a prioritisation note.

HOW THE SCORE IS CALCULATED

A weighted formula — not an equal average

Not all dimensions matter equally. Impact has more weight than effort because a high-impact experiment is worth running even if it's hard. Learning value prevents you from discarding experiments that will teach you something even if they fail.

Score = (Impact × 3 + Confidence × 2 + Reach × 2 + Effort × 1.5 + Learning × 1.5) ÷ 10

Score > 7.0 Run it
Score 5.0 – 7.0 Consider it
Score < 5.0 Skip it

NEED HELP WITH YOUR EXPERIMENT PROGRAMME?

We design and run experiment programmes for B2B SaaS teams

From backlog prioritisation to statistical rigour to learning velocity — we help teams build experimentation into how they make product decisions.

Talk to us about your experiment programme →