When each branch has only a handful of reviews, you cannot compare branches on their star averages. A branch with a few reviews and one with many are not measured to the same standard, and a single angry or delighted customer can move a small branch's average more than anything the team did. The fix is to stop ranking branches on averages and instead compare what customers say, group branches with enough evidence, and mark the rest as unknown until they have more.
This guide explains why thin review data misleads, and gives a simple method for comparing branches anyway.
Why a star average on a few reviews is unreliable
A star average looks precise. A branch rated 4.3 looks better than one rated 4.1. With few reviews, the difference is mostly noise.
Three things cause it.
Small samples swing. One extra one-star review moves a branch with a few reviews far more than it moves one with many. The change says more about the sample than about the branch.
Reviewers are not typical customers. Public reviews tend to come from people who feel strongly, either way. Most customers never post, so the reviews you have are a skewed slice of what happened.
A star score says how someone felt, not what went wrong. Two branches with the same average can have completely different problems, and the number cannot tell you which.
What not to do with thin data
- Do not rank branches by average rating when review counts differ widely.
- Do not set targets or bonuses on the star average of a small branch.
- Do not conclude that a branch is fine because it has no negative reviews.
- Do not assume a branch with few reviews has few problems. It may have few customers who post.
- Do not compare branches in different markets or formats as if they were the same.
A better method
1. Compare topics, not scores
Read what customers say and sort it into topics such as wait time, staff attitude, cleanliness, order accuracy or price. Then compare how often each topic comes up as a complaint at each branch. This tells you what to fix, which a score cannot.
The sorting only works if it is the same at every branch. If different people read different branches, you are comparing the readers.
2. Set a minimum before you compare
Decide how many mentions of a topic a branch needs before you rank it against others. Below that, show it as having low evidence. Outhentik, for example, marks a location with fewer than five mentions of a topic as low evidence, shows it muted and does not rank it. Pick your own threshold and apply it the same way to every branch.
3. Count branches, not just mentions
If three separate branches each have a few complaints on the same topic, that is stronger evidence of a shared problem than ten complaints at one branch. Combine the thin evidence across branches instead of waiting for each branch to build up enough on its own.
4. Pool branches that are alike
Group branches by format or market and compare within the group. A small group can gather enough evidence together to be useful even when each member is thin.
5. Use more than reviews
Public reviews are only one source. Add anything that gives you more observations per branch: direct messages from customers, complaints taken at the counter, and operational data such as refunds or remakes. More sources means more observations and fewer swings.
6. Show silence as unknown
A branch with no reviews and no messages is a branch you know nothing about. Mark it as unknown. Treat it as a coverage gap to close, not as good news.
7. Look at change over time within a branch
A branch compared with its own history is a fairer comparison than a branch compared with a very different one. Look for a topic that has started appearing, not just where a branch sits in a ranking.
A simple decision rule
Write the rule down before you look at the data.
- If a branch has fewer than your minimum mentions on a topic, label it low evidence and do not rank it.
- If a topic appears at two or more branches with enough evidence, treat it as a possible shared issue and look at the messages.
- If a branch is above the others on a topic and has enough evidence, ask the manager what is different and visit if needed.
- If a branch has too little data of any kind, find a way to hear from its customers before judging it.
Where Outhentik fits
Outhentik gives head office a direct line from customers, so branches have more to compare than the handful of public reviews. A customer scans a code or opens a link and tells head office what went wrong in their own words. Every message is read the same way at every location, a problem raised at two or more branches becomes an issue on the daily brief, a location with too few mentions is marked as low evidence, and a branch with nothing shows as unknown.
It depends on customers choosing to send a message, and it does not remove the need for judgment. Whether enough customers send messages to make the branch view useful is the first thing every rollout finds out.
Frequently asked questions
Why can't I compare branches by star rating?
With few reviews, a single customer can move the average, and reviewers are not typical of all customers. The difference between two small branches' averages is often noise.
How many reviews do I need before a comparison is fair?
There is no number that fits every business. Set a minimum before you compare, apply it consistently, and treat branches below it as low evidence.
What should I compare instead of star averages?
Compare what customers say, sorted into topics in the same way at every branch, and compare how many separate branches raise each topic.
Can I combine branches to get more data?
Yes. Pooling branches of a similar format or market gives you enough observations to see patterns that no single branch can show.
Does a branch with no negative reviews have no problems?
Not necessarily. It may have few customers who post, or customers who have no easy route to tell you. Look at coverage before drawing a conclusion.
Are online reviews useless for comparing branches?
No. They are a useful signal, particularly for larger branches. They are just not enough on their own when volumes are thin.
How do I get more customer data from small branches?
Give customers a simple, direct way to tell head office, such as a code at the counter or a link sent after the visit, and decide who places it and who follows up. How much comes back is something you find out by running it.
Try Outhentik free for 7 days, no credit card required
Outhentik gives head office a direct line from customers at every branch: a case with an owner for each customer who asks to hear back, every message read the same way, and a daily view of which problems repeat and where. For a group comparing branches with thin review data, that means more to compare, an honest label for low evidence and silent branches shown as unknown.