Audits and results

How is the Quality score calculated?

The Quality score (0-100) measures data reliability: what goes into it, why checkpoints carry different weights, how to interpret the number and why it can change "on its own".

2 min read

The Quality score is the 0-100 number you see with every audit - a synthetic measure of your website's measurement health. It's worth understanding what this number expresses, and what it doesn't.

What goes into the score

The score is built from the results of all checkpoints covered by your audit:

Pass - works in the score's favor,
Error - lowers the score the most,
Warning - lowers the score more gently than an error,
N/A - fully neutral: checkpoints outside your business scope (e.g. e-commerce checks for a service site) neither help nor hurt.

Checkpoints are not equal - they carry weights reflecting their impact on data reliability. Missing transaction deduplication (which corrupts every revenue report) weighs more than, say, a missing coupon parameter. The heaviest problems are additionally flagged as critical - those are the ones that surface at the top of the Action plan. We don't publish the exact weights and thresholds: they're part of the audit methodology, and you don't need them to interpret the score.

How to interpret the number

The score measures data reliability, not "website quality". A site can sell brilliantly and score 40 - that only means you can't trust the numbers you base decisions on.
Compare a project's score against its own history, not against other projects. A store and a service site are computed from different checkpoint sets (How does the business type affect my results?), and two different websites have different risk profiles.
Direction matters more than the absolute value. A 68 rising from 51 is good news; an 85 falling from 92 is a regression signal to investigate.

Why the score can change "on its own"

Some checkpoints analyze the last 30 days of data (event continuity, channel conversions, the funnel). If something broke on the site between audits - a deployment overwrote the tag, the consent banner blocked measurement - the score will drop even though you "changed nothing" in GA4. That's not a quirk of the methodology; it's precisely the function recurring audits exist for: catching regressions before you notice them in your reports.

Related articles:

More in Audits and results

Didn't find your answer?

Write to us - we respond fast, implementation questions included.

Write to us