Que sont les followers fantômes sur Instagram, et comment les éviter ?

A ghost follower is any account that counts toward your total and does nothing else. It never likes, never replies, never watches a story. The word describes a behaviour, which is why the term is so slippery: it says nothing about where the account came from, and the three places they come from need three different responses.

The three kinds, and why only one of them is a bot

A bot is an account made to inflate a number. It was never a person and never will be. That is the kind everyone pictures, and on most profiles it is the minority.

An inactive human is someone who followed you, meant it at the time, and stopped opening the app. Their account is real, their interest was real, and there is nothing to clean up because nothing went wrong.

A disinterested human is the awkward one. They followed for a giveaway, or from a post that travelled a long way from your usual audience, and they have no reason to look at anything you post next. Real person, real account, zero relevance to you, and no software can tell them apart from the second group.

That last point is the reason to be careful with checkers. A tool looking at your followers from outside sees what any visitor sees: posts, profile photo, follower and following counts, whether the account is private. From that it can guess at bots, and it guesses reasonably well at the crude ones. It has no way to see whether a quiet account is a lapsed reader or a person who followed for a prize, and it cannot see whether they opened your last story. So a checker's number is an estimate of one of the three groups, presented as a verdict on all three.

The reason the follower count is the wrong number to look at

Here is a comparison a marketplace can make and a general guide cannot. This site lists Instagram growth services from eleven independent supplier catalogues, all priced per thousand units, which makes it possible to ask what each kind of number costs to manufacture.

Measured across the live catalogue on 31 August 2026, and expressed as a ratio to that same seller's own follower rate, so that no seller's pricing can dominate the answer:

What is being bought	Cost against the same seller's followers	Sellers where it holds
Followers	The reference point	All nine measured
Likes	Cheaper, from roughly seven tenths down to under a tenth	Nine of nine
Views	Cheaper by a factor of tens to thousands	Nine of nine
Comments	Dearer, in places several times over	Six of eight dearer, one level, one below

Ratios within each seller's own catalogue, 31 August 2026. Only sellers with at least five services in a category were compared.

Two of those three orderings are unanimous. Every single seller prices likes below their own followers and views below that again. That is not a quirk of one cheap catalogue; it is what the whole market believes about relative cost.

The consequence is uncomfortable if you have ever judged an account by its numbers. The two figures a visitor sees first, the follower count and the likes under a post, are the two that are cheapest to produce, and likes are cheaper than the followers themselves. An account with a ghost audience can be made to look alive for a small fraction of what the audience cost.

The one number that resists this

Comments run the other way. On six of the eight sellers with enough of both to compare they cost more per thousand than that seller's own followers, in one case by an order of magnitude, and only one seller prices them below.

The reason is not mysterious. A follower is a state, and a view is an event. A comment is a piece of writing that has to be relevant to the specific post it sits under, which is why the listings that command the highest rates advertise custom or written-to-order text rather than a generic pool. Effort has a price and the market has set it.

So when you are working out whether an audience is real, yours or someone else's, weight what you read over what you count. Replies that respond to the actual content of the post, questions in the messages, people who show up on consecutive posts. Those are the signals that cost real money to fake, which is exactly why they are worth reading.

How ghosts accumulate

Nobody sets out to collect them. They arrive through four routes, and three of the four feel like growth at the time.

• Bought batches. The obvious one, and the only one that is entirely bots. If a previous owner or agency did this, you inherited it.
• Generic giveaways. A prize anyone would want attracts everyone, and everyone is not your audience. Tie the prize to the thing you actually post about and the entrants who stay are the ones who wanted that thing.
• Follow-for-follow and engagement pods. These produce a reciprocal obligation rather than an interest. The follow is real, the attention is performative, and it stops the moment the arrangement does.
• A post that travelled. The one nobody warns about, because it looks like a win. Something reached an audience far outside your niche, several hundred of them followed, and none of them will read what you write next. No money changed hands and the effect on your composition is the same.

Only the first of those is worth removing wholesale. The rest are ordinary consequences of running an account, and the fix is upstream: be specific enough that the people who arrive know what they arrived for.

What to measure instead

There is no published engagement rate you should be hitting. Any specific figure you have been quoted was decided by whoever quoted it, because the number moves enormously with account size, subject and country. The comparison that works is against yourself: the same account, the same kind of post, this month against last, and whether your interaction is moving in the same direction as your follower count. Divergence is the signal, and it needs no benchmark to read.

If you want to see who your audience actually is rather than how many of them there are, the real-versus-fake guide walks through the readings in your own account that a supplier cannot influence. If the audit says a chunk of yours is bots and you want them gone, the removal guide covers what the control does and why there is no bulk version of it. And if the numbers fell off a cliff rather than drifting, the sudden-drop guide is a better place to start than a cleanup.

If you do decide to buy a push at some point, the thing worth comparing is not the price. The Instagram listings here come from several independent suppliers and print the delivery terms per service, and the other platforms follow the same layout, so the terms are on the page rather than in the pitch.