Why did I lose Instagram followers overnight?

Your follower count is lower this morning than it was last night, and you did not do anything differently. In the overwhelming majority of cases that is Instagram removing accounts, not people leaving you. Overnight drops fall into five causes, and the largest by a wide margin is a platform-side sweep of inactive, bot and spam accounts, which happens to everyone in your niche at the same time and has nothing to do with your last post.

The distinction matters because the two situations call for opposite responses. A sweep needs nothing from you. Real people leaving needs a hard look at what you posted. This page tells you how to tell which one you are looking at, usually inside ten minutes, and what number to watch instead of the one that just moved. We sell followers, so read the section on paid batches as what it is. Everything above it works regardless of where your followers came from.

Start here: the five causes, and how to tell them apart

Every overnight drop anyone has ever described to us resolves into one of these. They look similar on the profile screen and they are easy to separate once you know what to compare.

The one test that separates a sweep from everything else

If a large number vanished inside a single window and other accounts in your niche are reporting the same thing on the same day, it is a sweep. That is the entire diagnostic: a big block, a short window, other people seeing it too. Nothing you can do to your own account produces a synchronized drop across thousands of unrelated profiles, so if the timing is shared, the cause is not yours.

If the drop is yours alone, the timing tells you the rest. Real unfollows follow content, so they cluster in the hours after a post rather than arriving as one overnight block. Removed bot accounts do the opposite: they go all at once, at an hour when nobody is awake to unfollow anyone, and the accounts that disappear turn out to have been empty when you look for them.

The May 2026 sweep, and why the reported numbers do not agree

Overnight on 6-7 May 2026, Instagram ran an AI-driven sweep that removed millions of bot, spam and inactive accounts in a single pass. Typical small and mid-sized creators lost between 2% and 5% of their count. Large accounts lost millions. The press named it the Great Purge of 2026, which is a press coinage rather than anything Meta has ever called it.

Two phrases in that statement are load-bearing. "Routine process" is Meta describing removal as maintenance rather than an event, which is a real change in posture and the reason to expect more of these. And "any restored suspended account will be included in the count again after verification" means part of what you lost can come back on its own, if the accounts behind it were caught wrongly and appealed successfully.

The celebrity figures contradict each other, and nobody has reconciled them

This is the part almost every page covering that night gets wrong, by repeating one number as though it were settled. The celebrity loss figures published in May do not agree across outlets, and the gaps are not rounding. Kylie Jenner's loss was reported as 7.24 million in some coverage and as over 15 million in others. Taylor Swift's was reported as roughly 1 million in some places and as 5.90 million in others. Those are not variations on a measurement. They are different measurements of the same night, presented with equal confidence.

The likely explanation is mundane, and it is worth understanding because it applies to your own numbers too. The sweep ran across hours rather than instantly, and no outlet had access to Instagram's own data. They were reading third-party follower trackers, which poll accounts on their own schedules and disagree with each other even on a quiet day. A newsroom that screenshotted a tracker at 2am and one that screenshotted at 9am were sampling different points on the same descending curve, against different baselines. That produces exactly the spread the coverage shows.

So we are not printing a celebrity figure here, because there is not one that can be verified. What can be verified is the shape of the event: a multi-hour, platform-wide removal of inactive and inauthentic accounts, announced afterward rather than before. The 2% to 5% band for ordinary accounts is the number that actually describes you, and it is the one worth remembering.

What a sweep removes, and what it leaves behind

The useful question is not how many accounts you lost. It is what those accounts were contributing before they went. Sorted that way, the answer is uncomfortable for anyone selling follower counts, ourselves included.

Read down the third column and the implication is hard to miss. Every account in the removed rows was contributing zero reach before it was removed, which means removing it cannot reduce your reach. What a sweep actually does is bring your follower count closer to the number of people who could plausibly see your posts. That is a correction to the measurement, not damage to the audience.

Losing 2% to 5% is a correction, not a punishment

Nothing about a sweep is aimed at you. Instagram did not review your account, decide it deserved fewer followers, and act. An automated system evaluated hundreds of millions of accounts against its own criteria and removed the ones that failed, and some of those happened to follow you. Your account was not the subject of the decision. It was downstream of it.

The number to watch afterward is reach. Open Insights, set the window so it spans the drop, and compare reach and views for the week before against the week after. If reach held roughly flat, the accounts you lost were not watching, and you have lost nothing but an inflated figure. If reach fell in step with the count, something else is going on and the follower drop was a symptom rather than the story.

One counterintuitive consequence: your engagement rate usually goes up after a sweep. Engagement rate divides engagement by follower count, and the denominator just got smaller while the numerator did not move. If you track that figure, name the denominator every time you quote it, because rates measured against followers and rates measured against reach differ by roughly an order of magnitude. Our guide to what a good engagement rate actually is works through both.

What to do in the next hour

1. 01Wait 24 hours before concluding anything. Profile counts and Insights refresh on different schedules, and a partial or wrong number is common while a sweep is still running.
2. 02Compare reach across the drop. Seven days before against seven days after, in Insights. This single comparison is the diagnosis: flat reach means you lost accounts that were never watching.
3. 03Open Settings, then Account Status. It tells you whether anything on your account has been removed or made ineligible for recommendation. A sweep does not appear there. A real restriction does, and that is a different problem entirely, covered in our shadowban check.
4. 04Look for other people reporting the same thing. If your niche is talking about it on the same day, you have your answer and there is nothing to fix.
5. 05Do not panic-post. The instinct after a drop is to post three times before lunch to win the number back. That is the one response that can genuinely cost you, because rushed posts are what produce the real unfollows you were worried about in the first place.
6. 06If the drop landed on a batch you paid for, file the refill claim. That is what a refill guarantee exists for, and using it costs nothing.

If the drop was on followers you bought

Purchased followers can be swept like anything else, and how many go depends entirely on what was sourced. Shell accounts created in a batch last month share creation timestamps, IP ranges and behavioral patterns, which is precisely the clustering these systems look for. Aged accounts with real posting history do not cluster, because they were never created as a group. That difference is most of what separates a cheap batch from an expensive one, and our breakdown of real versus fake followers shows how to audit which one you received.

Our own position is a 60-day refill guarantee: if your count falls below what you ordered within 60 days of delivery, we top it back up free. That covers a platform sweep the same way it covers ordinary drop-off, which we walk through in detail in the guide to why bought followers drop. Orders themselves run across hours at a steady rate rather than landing in one burst, and every package on the followers page completes within 1 to 24 hours depending on size.

The part we would rather not write, and which is true anyway: a refill replaces the number, not the reach. If a batch was swept, topping it back up restores the count and changes nothing about how many people see your posts, because those accounts were not seeing them beforehand either. Buy the count when the count is what you need. Do not buy it expecting reach to follow.

The slower version: bleeding followers day after day

This is a different problem wearing the same clothes, and people mistake it for a purge constantly. If you are down two here and five there, most days, with no single overnight block, no sweep is involved. Real people are leaving, one at a time, for reasons that have accumulated.

Being blunt about it: that is a content problem and it cannot be fixed by adding followers. Buying more while real ones walk out is filling a bucket with a hole in it. The count goes up, the leak continues, and the gap between your follower number and your actual reach gets wider rather than narrower. We sell followers and that is still the honest read.

The usual causes of a steady bleed are ordinary and fixable:

• A change in what you post. People followed a specific thing. When the account pivots, the audience that came for the old thing leaves, and that shows up as weeks of small losses rather than one drop.
• A cadence collapse followed by a burst. Going quiet for a month and returning with six posts in two days reminds a lot of people that they follow you, which is not always the outcome you want.
• Selling harder than you were when they followed. A feed that turns into a storefront sheds the people who came for the content.
• Follow-for-follow accounts unwinding. Accounts that followed you to be followed back will unfollow on a schedule, usually days later, and they are the least worth chasing.

Diagnose it with the same comparison as before. If reach per post is falling alongside the follower count, the content is the problem and the follower number is just reporting it. If reach is holding while the count slides, you are shedding accounts that were not watching, which is the version you want. Our guide to Reels that stop getting views covers the reach side in detail.

What changed in 2025 and 2026

Sweeps used to be occasional and newsworthy. They are now continuous, automated, and described by Meta as ordinary maintenance. Four shifts are worth knowing before you interpret your next drop.

• Removal is AI-driven and routine. The May 2026 pass was large enough to be noticed, but Meta framed it as a "routine process to remove inactive accounts" rather than an operation with a start and an end. Expect the next one without warning.
• The scale is enormous. Roughly 4% of Meta's 3.54 billion monthly users are fake accounts, which is more than 140 million profiles, with enforcement precision on Instagram above 87% as of Q3 2025. A pass at that scale reaches every niche on the same night.
• Meta does not publish Instagram fake-account removals separately. Its enforcement reporting breaks fake accounts out for Facebook only. That means any specific "Instagram removed X million accounts" headline you see cannot be checked against a primary source, including the ones that circulated in May.
• Views replaced impressions and plays as the primary metric, announced in August 2024 with the API cutover on 21 April 2025, and views counts repeats. Any before-and-after comparison that spans that boundary is not comparing the same quantity.

One more piece of context that reframes a flat count. Follower growth slowed sharply across the platform between 2024 and 2025: accounts in the 1,000 to 5,000 band grew 38% annually in 2024 and 22% in 2025, and every larger band fell further, from 27% down to 11.25% for accounts between 100,000 and a million. Standing still is closer to normal than it used to be. A one-night correction of 2% to 5% is noise against that trend, and our timeline for reaching 10,000 followers puts the arithmetic in front of you.