Algeria AgencyDigital marketing
The person who keeps coming back: when it stops being a crisis
An incident ends; somebody who returns every month does not. What changes, and why the crisis method becomes harmful.
The article accompanying this one describes these surfaces: you are talked about where you are not present, the complaint arrives as a video, and the mention vanishes in twenty-four hours. It describes a landscape.
This page is about one case within that landscape, the hardest and least documented: somebody keeps coming back. Not once, not for three days, but every month, sometimes for a year, with the same criticism or a new one each time.
The difficulty is not what is being said, it is that the method learned for incidents becomes actively harmful here. An incident is handled by replying fast and then stopping on the third day; applied to somebody who returns, that method produces twelve replies a year, each giving the next one a starting point.
So this article is about what repetition changes: how to recognise that the problem has changed, counting before judging, telling apart the three kinds of person who return, deciding once whether to engage, and living with what cannot be made to disappear.
What changes when it repeats
An incident and a recurring critic look like the same thing close up and have almost nothing in common. The difference is who is reading, and it decides everything else.
During an incident the audience forms a first impression: they discover the complaint and your reply at the same time, and the quality of the reply weighs heavily. Facing somebody who returns, the audience has already seen the previous episodes, and what they are assessing is no longer the substance but the relationship between the two of you.
That shift inverts the calculation completely. Replying to an incident serves you; replying for the twelfth time to the same person writes you into a serial, and the reader arriving at episode twelve is not looking for who is right — they observe that there is a history and file both parties together.
There is a second effect, less obvious and more expensive. Every reply supplies material: a sentence to quote, a screenshot to comment on, an occasion to start again. Somebody who returns needs you in order to continue, and the only resource they lack is your participation.
The practical consequence is that the first decision is not what to reply but whether to reply, and that decision is taken once rather than at each occurrence. A decision retaken each time is a decision taken in irritation, twelve times.
Count before judging
Before any decision you have to know what you are dealing with, and internal impression is even less reliable here than during an incident, because repetition amplifies memory. Somebody who has written four times in a year is experienced as constant harassment by those who read it.
So keep one line per occurrence, in a simple document: the date, the surface, what was said in five words, and whether there is a verifiable fact in it. Nothing else. The register takes two minutes an episode and answers questions nobody can answer from memory.
The first is the real frequency. Many cases experienced as relentless turn out to be five occurrences in eighteen months, which is an unpleasant fact rather than a situation to manage. Others reveal an acceleration, which is the information that actually matters.
The second is the trigger. Setting the dates beside your own calendar often reveals that the episodes follow something — an advertising campaign, a post, a reply you made. That link is invisible without the register and it changes the decision.
The third settles it: is there a verifiable fact? Somebody repeating a precise and accurate grievance is describing a problem you have not fixed. Somebody whose complaints change each time is describing something else, and the two cases are handled nothing alike.
Three different people
Under the shared appearance — somebody who returns — sit three situations with neither the same causes nor the same outcomes, and confusing them is the costliest error.
The first is the customer whose problem was never fixed. They return because nothing happened, and they will stop the day something does. It is the commonest and easiest case, provided you accept that the solution is not a reply but an act: a refund, a replacement, a visit.
The second is a competitor, or somebody acting for one. The distinguishing sign is not the tone but the content: precise criticisms of commercial points — your prices, your lead times against others, the mention of an alternative — and a total absence of verifiable detail about a personal experience.
The third is a person for whom you have become a subject. The original grievance may have been real; it is no longer the engine. The complaints change, widen, and move onto the way you reply. It is the hardest situation because nothing you do commercially will change it.
The sorting is done with the register rather than by instinct, because instinct files everybody in the third category. The question that separates them is simple: does an act of yours exist that would end this? If yes, do it and stop communicating. If no, the rest of this article applies.
The two-reply rule
Where no act settles the matter, you need a written rule decided in calm conditions, and the only one that holds over time is a ceiling: two public replies per person, never more.
The first reply is complete and serious. It addresses the substance, it is short, it gives a direct means of contact. It addresses the reader as much as the person, and it is written assuming it will be quoted out of context, because it will be.
The second, if there is one, is shorter and it closes: we answered here, our offer stands, this is how to reach us. It refutes nothing new. Its function is not to convince the person but to leave the next reader visible evidence that you replied twice.
After that, nothing in public, whatever the provocation and however unjust what is being said. That is the hard part, and it is hard for a reason worth naming: the urge to answer the thirteenth attack is proportional to its injustice, and that is exactly the mechanism feeding the serial.
Write the rule down and give it to everybody with account access. A rule held in the head is not a rule: the day somebody breaches the ceiling it will be an indignant employee at ten in the evening, and they will remember no decision that was not written.
Private, once
The private channel is the only place anything can actually be settled, and it works on one condition: go there once, early, with a concrete proposal.
That message is short and contains an offer, not an explanation. A refund, a replacement, an appointment, a gesture. An explanation in private is perceived as the debate continued by another channel, and it is often republished as a screenshot, which is the outcome to avoid.
Write it knowing it will be published. That is the single rule of this channel: anything you send privately may appear publicly ten minutes later, so nothing that could not be read by anybody belongs in it — no judgement of the person, no allusion to other customers, no irony.
If the proposal is refused or ignored, do not go back. A second and third private attempt are experienced as pressure, and they supply precisely the material for the next public episode: "they message me privately to shut me up".
Record the proposal and the date in the register. It serves twice: it reminds you that you did what was needed, which helps hold the two-reply rule; and it is the only useful element should the situation ever take a legal turn.
What can be reported, and what cannot
The standing temptation is to have it removed, and it does more damage than it prevents because it rests on a confusion: unfair criticism is not unlawful content, and platforms remove only the second.
What is usefully reported is a short list: abuse, threats, publication of personal data — your address, a number, a photograph taken without consent in a private setting — impersonation, and content claiming to come from your business. Those succeed often, and they succeed because they fall under written rules.
What is not reported: a harsh opinion, a low rating, an account of an experience you consider inaccurate, an unfavourable comparison. Reporting those does not remove them and produces two negative effects — the report fails, and the person learns they are getting to you, which is information better not given.
One practical rule avoids most errors: report what would be forbidden whoever wrote it and whoever it targeted. If the content would only be a problem because it is aimed at you, that is not a report, it is a disagreement.
And never announce publicly that you have reported something. It is the act that turns criticism into a censorship story, and it makes a defensible position look like an attempt to silence.
The video you cannot make disappear
On these surfaces the dominant form is video, and it has two properties text does not: it is hard to contradict in its own format, and it stays findable long after the conversation has died.
The first temptation is to answer with a video, and it is almost always a mistake. A video reply doubles the audience for the subject, gives the person material to comment on, and puts you on ground — editing, pacing, tone — where they are better than you, because it is what they do.
The useful reply, where there is one, goes in a comment under the video, in two sentences, without aggression. It is not to convince the author: it is because the comment is what arriving readers see, and criticism with no visible reply reads as criticism accepted.
What matters more is what somebody searching your name finds three months later. An isolated video is the only result if there is nothing else; the same video among recent ordinary content about your business weighs far less. That is the only real lever, and it is substantive work rather than a reaction.
Accept finally that it stays. That is not a defeat: attention is short-lived even when content is not, and a video seen by four thousand people in a week is seen by nobody the following month — unless somebody revives it, and the only one who can is you.
Your employees do not reply
This is the most important rule in the article and the most frequently broken, because it asks loyal people not to defend their business.
An employee replying from their personal account turns a dispute between a business and a person into a conflict between two individuals. The conversation stops being arbitrable, becomes personal, and continues after the business has stopped participating — because it is no longer the business participating.
The rule has to be set in calm conditions, explained, and justified by protecting the employee rather than the business, because that is the truth: the exposed party in this kind of exchange is the one with a name, a face and a personal account.
It covers the gestures that look harmless: not commenting, not reacting, not reporting collectively. Mass reporting by a team is detected by platforms and produces exactly the opposite of the intended effect.
What replaces it is an internal channel: where to flag what you saw, to whom, and what to say to colleagues who ask. Without that, the ban is experienced as abandonment, and it is broken the day somebody has had enough.
When the person is right
This case has to be planned for explicitly, because it is more frequent than admitted and because an organisation six months into a defensive posture stops seeing it.
The sign is in the register: a grievance that does not change, phrased the same way, about facts nobody in the business has actually verified. Criticism repeated identically is more often the sign of an unfixed problem than of an obsession.
Verification is done with operations and against other complaints from the same period, not with the person internally accused. That is the delicate point: when a customer has been complaining for months, the person concerned has already given their version ten times, and that version has become internal truth by repetition.
If the grievance is founded, the exit is public and brief: what was true, what has been done, and nothing else. No justification of the delay, no explanation of the misunderstanding. A late but clean acknowledgement is received far better than feared, and it ends the situation immediately, which no reply strategy would have done.
And the register stays useful afterwards: it shows how long acknowledgement took, which is the only information that improves the handling of the next case.
What it actually costs
A situation of this kind costs little commercially and a great deal internally, and inverting those two costs is what produces bad decisions.
Commercially the measurable effect is small in most cases. People who come across old criticism among other content weight it naturally; those following the serial are few and rarely customers. Look at your incoming enquiries over the period rather than at how it feels.
Internally the cost is real and badly distributed. One person — often whoever runs the accounts — reads everything, absorbs everything, and ends up defending a business in front of a hostile audience without anybody having decided they should. That is the part to deal with first.
Three measures suffice and cost little. Nobody reads alone: episodes are logged by one person and read by two. Nobody reads outside working hours. And the decision not to reply is taken by somebody other than the person receiving the messages.
The last is the most important. Asking the exposed person to decide not to defend themselves is a burden no written rule offsets — the decision has to come from above, and be repeated, so that silence is an instruction rather than a humiliation.
When it goes beyond communications
A small proportion of these situations leaves the field of reputation, and you need to know when, because continuing to treat as communications what is something else is what lets serious cases rot.
Three thresholds, and they are factual rather than emotional. An explicit threat against a person. The publication of personal data — an address, a telephone number, a photograph of an employee. And impersonation: an account passing itself off as your business or one of its owners.
In those three the order of actions changes completely. Preserve the evidence first — dated screenshots, links, account identifiers — report to the platform on the precise basis, and take legal advice. Public communication stops meanwhile, without announcement.
Note that preserving evidence is what is almost always missing, because content disappears: these surfaces are ephemeral by construction, and content deleted by its author after a report is content that no longer exists if nobody recorded it.
The neighbouring article on brand protection covers impersonation in detail, including what can be demanded and from whom. What belongs to this page stops at the moment the subject stops being what people think of you.
What we do, and what we refuse
What we do is bounded. We keep the register of occurrences, we sort between the three situations in section 3, we write the two public replies and the single private message, and we establish what constitutes a founded report.
We refuse to reply more than twice in public to the same person, including when asked and including when what is being said is false. It is the rule that protects best, and it holds precisely because it was decided before the episode that makes you want to break it.
We also refuse to report criticism that is merely unfair. An unfounded report fails, teaches the person they are getting to you, and turns a disagreement into a censorship story if they tell it. We report what would be forbidden whoever wrote it and whoever it targeted, and that is a clear line.
And we do not promise to make content disappear. Nothing lawfully published is removed for being unfair. What we commit to is the register, the rule, and the substantive work that stops an old video being the only result when somebody searches your name. There is also a decision only you can take, and it is worth more than all of the above: tell your employees they are not required to defend you.
Frequently asked questions
Somebody criticises us every month. Should we keep replying?
No. Two public replies per person, never more, and the rule is decided in calm conditions because the urge to reply is proportional to how unjust the criticism is. After the second episode the arriving reader no longer assesses the substance but the relationship between you, and every reply supplies the material for the next.
How do we know whether it is serious?
Keep one line per occurrence: date, surface, what was said in five words, and whether there is a verifiable fact. Many cases experienced as relentless turn out to be five episodes in eighteen months. The register also answers two questions memory cannot: is it accelerating, and do the episodes follow something you do.
Can we get the video taken down?
Only if it is unlawful: abuse, threats, personal data, impersonation. Unfair criticism, a low rating or an inaccurate account are not removed. The rule that avoids errors: report what would be forbidden whoever wrote it and whoever it targeted. And never announce a report — that is what turns criticism into a censorship story.
Should we reply with a video?
Almost never. A video reply doubles the audience for the subject, supplies material to comment on, and puts you on ground where the other party is better. Put two sentences in a comment under the video — for arriving readers, not for the author — and put the effort elsewhere: what is found when somebody searches your name three months later is the only real lever.
Our employees want to defend us. Should we let them?
No, and explain that the rule protects them. An employee replying from a personal account turns a dispute with the business into a conflict between two individuals, and the exposed party is the one with a name and a face. Give them an internal channel in exchange: where to flag, to whom, and what to say to colleagues who ask.
What if the person is right?
It is more frequent than admitted, and the sign is in the register: a grievance that does not change, about facts nobody in the business has verified. Check with operations, not with the person internally accused, whose version has become truth by repetition. Then acknowledge publicly, briefly, without justifying the delay.
Where we come in
Somebody returning every month is no longer an incident, and the crisis method turns harmful. What is missing then is a register, not a reply.
- We note each return with its date, its platform and its real audience.
- We decide together each time whether to answer or to note it and move on.
- We tell you when the frequency changes, because that is the only useful signal.
If the person returns fewer than three times a year, build nothing: a register for two occurrences costs more than it prevents.
Read next
Other networks: here, the complaint is a video
On these platforms what gets said about you is not text: it is a film, impossible to search by your name and far more convincing.Online reputation: five places, and you watch one
What gets written about you is spread across five surfaces. Most businesses monitor one, and it is almost never the one that decides.The week of an incident: what to do hour by hour
An incident is handled in time, not by channel. The first hour, the holding line, the third day, and what actually changes afterwards.
Let us talk about your project
A free audit, no commitment: we look at your online presence and tell you what is holding it back.