Closed-loop feedback is easiest to understand by following a single piece of feedback from the moment it arrives to the moment it changes something. So let us take one. The company is an invented broadband provider, the customer is Rohit, and the survey is a relationship survey that asks the standard Net Promoter question: how likely are you to recommend us, from 0 to 10.

Step 1: the score arrives

Rohit gives a 3. In NPS terms, scores of 0 to 6 are detractors, 7 and 8 are passives, 9 and 10 are promoters. He adds a comment: “Technician came, said fixed, internet dropped again the same night. Called twice.” Without a loop, this goes into a monthly report as one of many detractors, and the company learns that its score moved by a fraction of a point. With a loop, the survey tool flags it the same day because the score is below the trigger and a comment is present.

Two things are needed before this can work at all. The survey must capture identity or a reference to the ticket, and the customer must have consented to being contacted again. Surveys that are anonymous can still feed the second loop below, but cannot support a personal callback.

Step 2: route it to a person with authority

The alert goes to a resolution desk, not to the agent who took Rohit’s earlier calls. This is a design choice worth noticing. If the original agent receives it, the callback becomes a defence of that agent’s work. A small team of experienced people who can raise a technician visit, approve a credit and escalate a network fault is better placed. They need to be able to fix things, because a callback that only says “sorry” is worse than none.

A common target is to contact detractors within one to two working days. The exact window is a company choice. What matters is that it is short enough that the customer still remembers the issue, and published internally so that it is tracked.

Step 3: the callback

The caller opens by saying who they are, thanks Rohit for the feedback, and asks him to describe what happened in his own words. Then they listen. The aim is not to argue the score up. Customers who feel they are being lobbied to change their rating dig in, and it also corrupts the data. The caller finds out that the first technician visit closed the ticket after replacing a cable, but the drops were coming from the router. The caller books a router swap for the next morning and notes that the earlier ticket was closed as resolved when it was not.

Observe what happened: this is the inner loop. One customer, one problem, fixed, with a human acknowledgement. Often that alone turns a detractor into a customer who stays. But if we stop here, we have learnt nothing that helps the next Rohit.

Step 4: tag the root cause

After the call the resolution desk records a cause code from a short, agreed list, for example: closed without confirmation, repeat contact, wrong information, long wait, billing error, technician no-show. Choosing the code is a discipline. If the list has forty items, people choose “other”. If it has six, the data becomes usable. Here the code is “ticket closed without customer confirmation”.

Step 5: patterns, the outer loop

Suppose that in a month there are 30 detractor comments with callbacks completed, and 12 carry the same code. That is 40 per cent. It is no longer a story about Rohit, it is a defect in a process. A weekly review, attended by someone from operations and someone from the field team, looks at the top two codes and assigns an owner. The owner finds that tickets auto-close after 24 hours if the customer has not replied to an SMS. Many customers do not read the SMS. The proposed fix is to require a confirmation call or an in-app tap for outages tagged as repeat, and to measure repeat tickets within seven days as the check.

The outer loop is the one that is hardest to sustain. Inner loops satisfy people. Outer loops cost money and sometimes require a client or another department to change something. Many programmes quietly stop at the inner loop and call themselves closed-loop.

Step 6: tell people what changed

Two audiences need to hear. The team that handles customers should hear that the auto-close rule changed because customers like Rohit reported it. That is the most effective way to show agents that feedback is taken seriously. And, where practical, customers who complained can be told that their feedback led to a change. This is sometimes called “you said, we did”. It costs little and builds the trust that makes people reply to future surveys.

What can go wrong

  • The loop is closed on paper: calls are logged as “attempted” and counted as done. Track completed conversations separately from attempts.
  • Callers are measured on whether the score improves. That pushes them to pressure customers. Measure speed of contact and resolution instead.
  • Only detractors get attention. Promoters also leave useful comments about what is working, and passives are the quiet middle that can slide either way.
  • Privacy is ignored. Respect consent and the data protection rules that apply, and keep the customer’s comment accessible only to people who need it.

Look back at Rohit. The score was a 3; the real information was in the sentence below it. A closed loop is mostly a habit of reading that sentence and doing something about it twice: once for him, and once for everyone who will be in his position next month.

Figures and names in the examples are illustrative.

Leave a Reply

Trending