A 1 to 5 score tells head office how a visit felt to the customer, in a form that can be averaged, tracked over time and compared between branches. It does not say why, so it cannot be fixed from. A message in the customer's own words tells head office what happened, who was involved and what the customer wanted instead, which is what a fix needs, but it cannot be averaged and it takes a reader. Head office needs both, in that order of importance: the words first, because they are what changes a branch, and the score as an optional signal alongside them, never as the thing the programme runs on. A programme built on the score alone produces a ranked list of branches and no idea what to do about it.
This article sets out what a score can do and what it cannot, what words can do and what they cost, the four things that go wrong when the score is the primary measure, how to use the two together, and what to do about the customers who only leave a number.
What a score can do
Scores are not useless, and an article that pretends they are will not be trusted by anyone who has run a branch network.
A score is fast for the customer. Tapping a number takes a second, and some customers who would never write a sentence will tap one.
A score is comparable, in a limited way. If the same question is asked the same way at every branch, the distribution of answers at branch 6 and branch 19 can be put side by side.
A score moves over time. A branch whose scores drift down across a quarter is telling head office something, even if it is not telling it what.
A score is easy to report upward. A board, an owner or a franchisor will look at a number where they will not read two hundred messages.
Those are real properties, and the limits below do not cancel them.
What a score cannot do
A score cannot say why. A 2 at a pharmacy counter could be a twenty-minute wait, a prescription that was not ready, a pharmacist who did not look up, or a price the customer thought was wrong. Each has a different owner and a different fix. The 2 is the same in all four cases.
A score cannot distinguish a branch problem from a bad day. One week of low scores at a branch could be a staffing gap, a broken system, or three difficult customers. Only the words say which.
A score cannot be compared across branches as cleanly as it looks. Branches serve different customers at different times for different things. A branch next to a hospital and a branch in a shopping centre will score differently with identical service, and a score comparison treats that as a service difference.
A score hides its own sample. Twelve scores at a branch averaging 4.1 is not the same as four hundred scores averaging 4.1, and an average displayed without its count invites decisions on noise.
A score, once it is the measure, changes behaviour at the counter. Staff who know the branch is judged on the number start caring about who is asked to give it. That is where selective asking begins, and selective asking is where a feedback programme stops being honest.
What the customer's words can do
Words say what happened. "I waited twenty minutes and then was told it would be ready tomorrow" is a case and an issue at once, with a cause in it.
Words say who. Not always a name, but a description that the manager can place: the person on the till, the pharmacist, the one who was on the phone.
Words say what the customer wanted. "I would have been fine if someone had told me when I arrived" is the fix, written by the customer.
Words carry praise that a score flattens. A 5 is a 5. "She rang the surgery for me to sort the prescription out" is something a manager can take to a person.
Words show patterns across branches when they are read the same way. "Not ready when promised" at four branches is one issue. Four low scores at four branches are four numbers.
Words resist being gamed. A staff member can nudge a customer toward a 5. It is much harder to nudge a customer into writing something specific and untrue.
What words cost
Words need a reader. Two hundred messages a month from forty branches is not something a head office reads in a spare hour, and reading them inconsistently (one person's "staff were rude" filed under service, another's under staff) destroys the comparison across branches that is the point.
Words take the customer longer. A sentence is more effort than a tap, and some customers will tap who would not type. Letting the customer speak or record instead of typing lowers that cost, but it does not remove it.
Words are harder to put on a slide. An owner who wants one number per branch will not get one from a pile of messages, and that is a real tension in any group where the board reads the dashboard.
Those costs are why most programmes default to the score. The score is cheap to collect and cheap to report. It is only expensive at the point of trying to fix anything.
Four things that go wrong when the score runs the programme
Branches get ranked
A score produces a league table almost automatically, and a league table tells a manager where they sit and nothing about what to do. Managers at the bottom feel judged for conditions head office set; managers at the top learn nothing about why.
The question becomes the target
Once a branch is measured on a number, the branch works on the number. Which customers are asked, when, and how, start to drift. The data gets better and the branches do not.
Routing starts to follow the number
The natural next step for a score-driven programme is to treat low scores differently from high ones: call the 1s and 2s, thank the 5s, invite the 5s to review. Selectively inviting positive reviewers is listed as rating manipulation in Google's Maps user contributed content policy, and it starts with the number being the thing the system looks at first. Every customer who sends a message should be offered the same choices regardless of what score, if any, they gave.
Nobody reads the comments
When the score is the measure, the optional comment box becomes an afterthought. It is collected, stored and skipped, because the report is already written from the number. The one part of the data that said why is the part nobody opens.
Using both
The words are primary. Every message is read, the same way at every branch, by one reader, with the original kept behind every finding. Cases are opened from the words. Issues are found from the words. Praise is passed down from the words.
The score is optional and secondary. The customer can give one or not. It is shown alongside the words, never instead of them, and never used to decide who is contacted, who is thanked, or who is invited to do anything.
Comparison between branches is done on the issues found in the words, not on the average score. "Not ready when promised appears at branches 3, 6 and 19" is a comparison head office can act on. "Branch 6 averages 3.4 and branch 19 averages 4.0" is not.
Trends over time are read from both. A branch whose scores drift down and whose messages start saying "wait" is a branch with a staffing or process problem and the evidence to say which.
Outhentik is built this way: the customer types, speaks or records in their own words, the rating is optional, every message is read by the same reader at every branch, and nothing about who is contacted or what is offered depends on the score.
The customers who only leave a number
Some will. A 2 with no words is a signal that a visit went badly and no explanation of why.
Do not call them because the score was low. If the customer did not ask to be contacted, a low score is not a request. A call that opens with "we saw you gave us a 2" tells the customer the business is watching the number, which is exactly what should not be true.
Do count them, alongside the words. A branch producing many low scores with no words and few messages with words has a channel problem, not just a service problem: the customers are tapping and leaving, which suggests the ask is for a score and the words feel optional. The fix is at the counter and in the first screen, not in the reporting.
Do not build the comparison on them. A branch's unexplained low scores are a reason to read that branch's words more closely, not a line on a league table.
What not to do
Report the average score without the count. It invites decisions on four responses.
Rank branches on score. It tells nobody what to change.
Let staff see the branch's live score. It changes who they ask.
Route anything on the score: who is contacted, who is thanked, who is invited to review. Every customer who sends a message is offered the same choices.
Collect comments and not read them. If the words are not going to be read, do not ask for them; the customer who wrote a paragraph and heard nothing has been taught the channel is decorative.
Treat a score drift as a finding. It is a reason to read the words from that branch.
What to expect
If the first screen asks for words first and makes the score optional, the balance will tilt toward words. How far is not knowable in advance, and it will differ by branch and by what the ask at the counter says.
Reading takes time, or it takes a reader that does it the same way at every branch. Groups that try to read by hand across many branches usually stop within a quarter; groups that let the branch manager read their own branch lose the comparison. Either the reading is one person's full-time job or it is done by something that reads consistently.
Owners and boards will still ask for a number. The honest answer is the count of open issues, the oldest open case, and the share of branches with coverage, none of which is a satisfaction score and all of which say more.
Customers who leave a score or a message can still post a review anywhere they like. The words give head office the account and a chance to act. They do not decide what the customer does elsewhere.
Frequently asked questions
Is a 1 to 5 rating useful customer feedback?
It is useful as a signal that a visit went well or badly and as a trend over time. It is not useful for fixing anything, because it does not say why, and it is misleading for comparing branches, because branches serve different customers in different conditions. It should sit alongside the customer's words, not replace them.
What does open-ended customer feedback tell you that a score does not?
What happened, who was involved, what the customer wanted instead, and whether the same thing is happening at other branches. It also carries praise about specific people that a 5 flattens. Those are the things a fix needs.
Should branches be compared on their average feedback score?
No. Branches serve different customers at different times, so identical service scores differently. Compare branches on the issues found in the customer's words ("not ready when promised at three branches") and on coverage, not on the average.
Should a business call customers who leave a low score?
Only if the customer asked to be contacted. A low score is not a request, and a call that opens with the score tells the customer the business is watching the number. Every customer who sends a message is offered the same choices, whatever they scored.
Why do most feedback programmes rely on scores?
Because a score is cheap to collect, easy to average and easy to put on a slide. It is only expensive when someone tries to act on it. The words are the opposite: harder to collect and report, and the only part that says what to fix.
How do you read hundreds of customer messages across many branches?
With one reader that sorts every message the same way at every branch, so "not ready when promised" at branch 3 is counted like the same thing at branch 19, with the original message kept behind every finding. Reading by hand, branch by branch, loses the comparison and usually stops within a few months.
Try Outhentik free for 7 days, no credit card required
Outhentik gives head office a direct line from customers at every branch: every message read the same way, a case with an owner for each customer who asks to hear back, and a daily view of which problems repeat and where. The customer writes, speaks or records in their own words; the rating is optional and decides nothing.