Your clients have already told you. In writing.

Four channels, four silos

An unhappy client does not fill in a complaint form. They leave an online review on a Sunday evening, they slip a sentence into the end of an email to their salesperson, they tick “other” on a contact form, or they open a ticket that support closes without comment.

Four channels, four recipients, four tools. The salesperson saw their email. Support saw their ticket. Marketing saw the review. Nobody saw that it was the same complaint, made four times by four different people.

That is the underlying problem: the signal is not absent, it is simply spread out. And a spread-out signal reads as noise, until the day a client leaves saying it plainly, and everyone remembers having heard it somewhere already.

Why the survey arrives too late

The satisfaction questionnaire has two flaws. The first is timing: you send it once or twice a year, so you learn about a problem months after it started. The second is phrasing: you ask what you thought to ask, and you miss what you did not think of.

Inbound channels have neither flaw. They are continuous, and they are open-ended. The client writes what annoys them, in their own words, at the moment it annoys them. That is exactly the material you are looking for, and it is already yours.

The survey keeps its use for measuring change over time. It is poor at discovering a subject. Those are two different jobs.

Reading the four together

Rereading all of it by hand is possible, once. Doing it every quarter is not: it is several days of reading, and the sorting is subjective. That is the work an assistant absorbs well, on three conditions.

• It groups by theme, not by channel. The call-back issue must come up once, with feedback from every channel underneath, not four times in four reports.
• It quotes, it does not summarise. Every irritant comes back with the clients’ exact sentences, their date and their source. A summary gets argued with in a meeting; a dated verbatim does not.
• It counts what it has read, and nothing more. A theme is flagged as recurring only if it appears in several pieces of feedback, and the assistant tells you which. No percentage out of nowhere, no extrapolation over what it has not seen.

The third condition is the one people forget. A tool that announces “32 % of your clients are unhappy with the lead time” without being able to say which feedback it means puts you in danger in a board meeting, the day someone asks where the figure comes from.

What you do with it in a board meeting

The useful output is not a forty-page report. It is three lines you can put on the table.

• A named irritant, in the clients’ words, not in yours.
• A dated quote that carries it better than any rewording.
• A stated scope: which channels the reading covered, and which were not connected.

That last point matters as much as the other two. If support tickets are not connected, the assistant must say so, and the board must know before deciding. A partial view presented as partial stays usable. A partial view presented as complete leads to bad decisions.

And once the decision is taken, the same reading, run again a quarter later, says whether the subject has fallen away. That is the only indicator worth having: not declared satisfaction, but the disappearance of the unprompted complaint.