Customer review software: five measurement mistakes to stop

What to take away

Mistake 1: Which source types should you keep separate?

Reviews arrive through several routes: on-site submissions, marketplace listings, third-party platforms and social comments. Each carries its own population and rules.

TikTok's community guidelines on misleading content show how platform policy shapes what can be published there, which is one reason social comments behave differently from verified purchase reviews.

Aggregate everything into one rating and you get a number nobody can defend. Split by source and show the total as a range.

Mistake 2: What does a missing denominator hide?

A count without a base tells you nothing about reach. Four hundred reviews sounds strong until you learn that 40,000 orders shipped that quarter.

Pair every count with the population it came from: orders, visitors, emails sent or invitations issued. The review rate then becomes a real metric.

Commerce reviews key metrics: data and sources sets out which denominators belong with which metric, so your dashboard can follow one consistent set.

Mistake 3: Are you picking the date window after seeing the result?

Choosing a start date once the numbers are in is the quickest way to mislead a board, even unintentionally. A strong month after a weak quarter looks like a turnaround if the chart begins in the right place.

Set windows in advance and publish them with the chart. If you change one, show both versions.

Mistake 4: Does one satisfaction score explain anything?

A blended score hides the difference between a delivery complaint and a product fault. Those need different owners and different fixes.

Break the score into at least two dimensions, such as product satisfaction and service satisfaction, and keep them separate.

Mistake 5: Do you treat unverified reviews the same as verified ones?

Verified purchase reviews and unverified ones carry different weight. A five-star review from a confirmed buyer is not the same evidence as an unverified comment. Label each review as verified or unverified, and report them on separate lines. If you must combine them, state the mix.

Compliance note: which data protection rules apply when you share review data?

Review data often contains personal information. Passing it to a software provider counts as sharing personal data, so you need a lawful basis and appropriate safeguards.

The ICO's Data Sharing Code of Practice explains how to approach that sharing. Apply it before you export a review file to a new supplier.

How should you sequence the fix?

The commerce reviews measurement and reporting guide 2027 shows how these steps fit a full reporting cycle, so treat this list as pre-work.

Common questions

Are five mistakes too many to fix at once?

No, because they are sequential. Source separation and denominators come first; the rest follow once the data model is clean.

Does this list apply outside England?

The data protection rules referenced here operate across the UK. Where a rule differs in Scotland, Wales or Northern Ireland, check that nation's regulator guidance.

How long before reporting improves?

For example, a team running six sources might need one full reporting cycle to rebuild the dashboard. Queries about conflicting figures should then fall.

Do I need new software to fix these mistakes?

Usually not. Most fixes are conventions: labels, denominators and fixed windows. Buy new tools only when a source cannot be separated in your current setup.

What is a review rate?

It is the number of reviews divided by the number of orders, visitors or invitations sent. Attach a denominator to every count to turn it into a real metric.

What is a denominator in review reporting?

The population a review count came from, such as orders, visitors or emails sent. A count without a denominator tells you nothing about reach.

How do I choose a date window?

Set it in advance and publish it with the chart. If you change one, show both versions so the reader can see the effect.

Why split product and service satisfaction?

A blended score hides the difference between a delivery complaint and a product fault. Those need different owners and different fixes.

What is a verified purchase review?

A review left by a confirmed buyer. Verified and unverified reviews carry different weight, so label them and report them on separate lines.

Which ICO guidance applies to sharing review data?

The ICO's Data Sharing Code of Practice explains how to approach sharing personal data. Apply it before you export a review file to a new supplier.

What is the first step in fixing measurement mistakes?

List every place reviews are collected or displayed, with an owner for each. Then assign each source a label and a separate line in the report.