Should You Buy Bluesky Followers? Everyone Can See That You Did
You can buy them — $10 to $95 per thousand across the vendors I priced in August 2026 — and you shouldn't, for a reason that is specific to this network rather than general internet moralising. Bluesky's follow graph is public and unauthenticated, so anyone who wonders about your followers can audit them with one HTTP request that needs no login, no API key and no approval. On X or Instagram that check needs paid API access, which is why bought followers there mostly go unnoticed. Here it costs nothing and takes a minute, and the accounts being audited are the least durable objects on the platform.
Search "buy Bluesky followers" and every result on page one is someone selling them. Not one is a disinterested answer. So let me do the arithmetic they'd rather skip — including the part where the cheap end of the market undercuts my own product.
How Much Does It Cost to Buy Bluesky Followers?
Less than you'd think, and the spread is the story. Prices read off vendor order pages on August 19, 2026:
| Vendor | Advertised price | Per follower | Per thousand |
|---|---|---|---|
| SocialPlug | "Starting from $0.095 / Follower" | $0.095 | ~$95 |
| Viplikes | $55.49 for 1,000 | $0.055 | $55 |
| GTR Socials | from $2.00, 100 minimum | $0.02 | ~$20 |
| SMM Laboratory | $1.00 per 100 | $0.01 | ~$10 |
The same nominal product — a thousand Bluesky followers — swings tenfold depending on where you shop, and the cheap end is cheaper per head than my own tool. Nothing real behaves like that. What varies isn't quality, it's how disposable the accounts are. (SocialPlug's page can't decide either: the headline says "From $0.95" while the pricing block underneath says $0.095. I've used the smaller number, which is the one on the pricing block itself.)
For scale on the honest alternative, our own growth study puts the median Agent Sky account at roughly 280 net new followers a month — a median across the 20 accounts with a full 30-day window, from a study of 58,334 follows across 37 accounts, with the middle 50% landing anywhere between +92 and +920. I'm keeping that out of the table above, because a table would imply these are one kind of thing and they are not. One is a subscription that puts you in front of people; the other is a list of names.
So cost per head is a wash, and anyone who tells you otherwise is selling something (possibly me). The interesting denominator is cost per follower who still exists.
Why Do Bought Followers Disappear on Bluesky?
Because the population they're drawn from is the one Bluesky spends its year deleting. Per the 2025 Transparency Report, published January 29, 2026, the company removed 2,446,334 items covering both accounts and content during the year; account takedowns alone "grew just 104% (from 1.02M to 2.08M)". Of 9.97M user reports, "misleading" was 43.73%, with spam accounting for 2.49M of those.
There is also a documented batch. In April 2025 the researcher Conspirador Norteño found — and wrote up that December — a network of over twelve thousand Bluesky accounts that "popped into existence in the span of a few hours," each running one bio template — "Iran based, interested in music," "New Zealand based, interested in technology" — and each empty, following nobody. The purpose only became clear when screenshots on a follower-sales site turned up displaying accounts with those identical bios. His reading, hedged exactly this far: "almost certainly a batch of fake followers intended to be sold via Morelikes." By the time he published, it was "the now-suspended network."
That's the lifecycle in one citation, but it's one batch, so I wanted the background rate to go with it. Every Bluesky account ever created leaves a public record in the plc.directory identity log, timestamped, readable by anyone, no key required. I sampled four windows a month for 22 months, kept the records Bluesky's own servers issued, and re-checked every account on August 19, 2026:
| Signup month | Accounts sampled | Still live today | Of those, never posted |
|---|---|---|---|
| November 2024 | 400 | 93.5% | 57.0% |
| April 2025 | 171 | 77.2% | 79.5% |
| October 2025 | 313 | 67.1% | 83.8% |
| March 2026 | 400 | 70.0% | 81.1% |
| August 2026 | 113 | 76.1% | 77.9% |
| All 22 months | 6,727 | 78.3% | 78.6% |
Roughly one in five ordinary Bluesky accounts is gone within a couple of years — 1,457 of the 6,727 I looked up returned nothing at all. That is the floor, the rate for accounts that mostly belong to real if inattentive people. Purchased accounts are the tail of that distribution, not a different distribution: bulk-registered, templated, reported early, and removed first. The November 2024 cohort, which arrived in the post-election rush and is the oldest in the table, is also the healthiest at 93.5% — attrition here isn't about age.
One mechanical footnote, because people notice it and assume something is broken: when those accounts go, they drop out of your followers list without necessarily moving your follower count. You end up with a number nobody can click through to.
Can Anyone Tell If You Bought Bluesky Followers?
Yes, and this is the part the vendors will never mention: auditing someone's followers is a free, unauthenticated HTTP request. No login, no API key, no developer approval, no paid tier. Here's the whole method — I ran it while writing this:
- Call the public AppView:
https://public.api.bsky.app/xrpc/app.bsky.graph.getFollowers?actor=<handle>&limit=100 - Read the JSON. Every follower comes back with
handle,displayName, and acreatedAtaccount-creation timestamp. - Sort by
createdAtand look for clusters. Three hundred followers registered in the same week is not a coincidence.
Then apply the correction almost everyone skips, which is the one that stops you libelling a stranger. An empty profile proves nothing. Of the surviving accounts in the census above, 78.6% have never posted once, and the median survivor has two followers. Bluesky is full of real people who signed up, looked around and never came back; "no posts, no bio, no followers" describes them exactly as well as it describes inventory. The signal that separates the two is the clustering — creation timestamps seconds apart, handles built from one word list — not the emptiness of any single account.
While I was building that census I ran into a second reason to distrust anything you haven't resolved yourself, and it's worth a paragraph because it surprised me. Of 8,700 identity records claiming to have been created on a Bluesky-run server, 1,973 were never issued by Bluesky at all. Anyone can write a record into the public log naming any server they like, and somebody is doing it in bursts: 22.7% of my sample, absent entirely for five straight months, then two thirds of December 2025's sample and 97% of a single window this August. They're easy to spot once you know the tell — Bluesky's own servers stamp every account they create on a *.host.bsky.network host with one shared pair of rotation keys, and these carry their own; nearly all the handles contain underscores, which atproto handle syntax forbids; each one pads nine junk strings into a field that normally holds a single name. Not one of the 1,973 resolves to an account. They aren't dead accounts, they're records of accounts that never existed. Whatever they're for, they're a decent demonstration of how cheap it is to manufacture the appearance of a Bluesky user — which is the same trick a follower vendor is selling you one layer up.
None of this is a hole in the directory, incidentally — the PLC spec says outright that it "does not cross-validate alsoKnownAs or service entries in operations," and that the data "should not be trusted without bi-directionally verification." These records are validly signed. They just lie about where they live.
The census is 22 runs of ./scripts/bsky_stats.py signups <YYYY-MM>, one per month. Here is a single one-minute window, to show the shape of the forged half:
$ ./scripts/bsky_stats.py signups 2026-08-19T15:14:00Z 1
=== Bluesky account records: 2026-08-19T15:14:00Z ===
genesis records naming a Bluesky PDS 190
issued by Bluesky (rotation keys) 16 (8.4%)
forged (keys are not Bluesky's) 174 (91.6%)
of the 16 real accounts:
still live today 16 (100.0%)
…
of the survivors, never posted 13 (81.2%)
of the 174 forged records: 0 resolve to a profile (lookup failures: 0)
sample:
2026-08-19T15:14:00.282Z fabian_ut.bsky.social alsoKnownAs=10
2026-08-19T15:14:00.632Z raegan_et.bsky.social alsoKnownAs=10
…
(One minute of the log, elided in two places. Those sixteen accounts were minutes old when I checked, so 100% live is arithmetic rather than a finding — this window is here for the forged half. fabian_ut.bsky.social is not a handle Bluesky would issue — the underscore is illegal — which is the giveaway before you check a single key.)
I'm not claiming to have invented the follower audit. Conspirador Norteño has run a far more sophisticated version publicly for over a year, and free tools like bskycheck.com will bucket your own followers by their follow ratios for you, though you still have to read the clustering yourself. That's rather the point: the technique is published, partly automated, and available to anyone curious about your account. Here the follow graph is public infrastructure — the same openness that makes custom feeds and third-party clients work also let me do all of the above from a laptop, with no credentials.
Bluesky maintainer bnewbold framed the design intent in an AT Protocol discussion: "an original philosophy of the protocol is that folks should be able to independently compute and verify these numbers." He was talking about verifying counts network-wide, which he notes is genuinely heavy lifting. Auditing one profile's followers is the easy case, and it's the case that decides whether a sponsor believes you.
Is It Safe to Buy Bluesky Followers, or Against Bluesky's Rules?
It's against the rules explicitly — follower counts are named by name — and the likeliest damage isn't a ban. From Bluesky's Community Guidelines, which took effect October 15, 2025:
"Do not artificially manipulate features or social signals to gain unearned reach or mislead users, including engagement metrics, follower counts, or other measures of community interest."
Being precise about the risk matters, because scare-mongering is cheap. The likeliest outcome is not your account getting nuked. It's the purchased accounts getting removed on the schedule in the table above while you keep a hollow number and a public timestamp cluster that says how you got it. But you are handing a moderation system a reason to look at you, and if you also mass-follow to "activate" your new audience, you've stacked a second behavioural signal on the first — that failure mode is why accounts get marked as spam.
Where the line actually sits: automation isn't the problem, fabrication is. Bluesky publishes a bot tutorial in its own developer docs and its guidelines regulate what automation does rather than banning it, which is the full picture on Bluesky's bot rules. Software that follows relevant humans at a human pace is fine. Paying for accounts that impersonate an audience is the thing being prohibited.
How Do You Get Bluesky Followers for Free Instead?
The organic playbook is genuinely free and works about as fast as the honest paid version — it just costs attention instead of money. The four levers, in the order they pay off:
- Get into starter packs in your niche, and make one worth being in.
- Reply before you post. Replies put you in front of an existing audience; standalone posts don't.
- Follow relevant people and let follow-backs accumulate rather than following everyone.
- Post when your people are awake, which is a smaller effect than the timing blogs claim but a free one.
The full version, with the parts I've left out here, is in the free playbook. No tool required, mine included.
What Should You Do Instead of Buying Bluesky Followers?
Obvious conflict of interest: I sell a growth tool, so "don't buy followers, earn them" is a convenient argument for me to make. Here's why I think it survives that.
A purchased follower is worthless because it will never reply, repost, or carry your post into anyone's feed — and, on the numbers above, because there's a decent chance it won't be there. A real follow-back is worth something because a person looked at your profile and decided you were relevant to them. You can't buy the deciding part. You can only put yourself in front of people likely to decide yes, at a pace that doesn't look robotic.
That's the job Agent Sky does, and the mechanics are deliberately boring: by default it never fires two follows less than three minutes apart, jittering every gap out to ten, it stays inside the hours you set in your own timezone, it works the audiences of accounts you nominate and ranks the candidates by measured likelihood of following back, and it checks that a follow-back is real before it drops anybody — deferring the unfollow entirely when it can't tell. It's free to start and $9 a month after that. My bet is that followers earned this way are still there in six months. I can't prove that yet; I can prove the alternative isn't, which is a weaker claim and the only one this post is entitled to make.
The vendors are selling you a number. Bluesky hands everyone else the receipts.
Quick FAQ
Can you buy Bluesky followers? Yes. A dozen SMM panels sell them and the transaction generally works — a few hundred accounts follow you within a day or two. What arrives is bulk-registered accounts with formulaic bios and no posts. Bluesky's Community Guidelines, which took effect October 15, 2025, explicitly prohibit artificially manipulating "engagement metrics, follower counts, or other measures of community interest."
How much does it cost to buy Bluesky followers? Prices read off vendor order pages on August 19, 2026: SocialPlug advertises "Starting from $0.095 / Follower" (about $95 per thousand), Viplikes lists 1,000 followers at $55.49, GTR Socials starts at $2.00 for its 100-follower minimum ($20 per thousand), and SMM Laboratory lists $1.00 per 100 ($10 per thousand). That is a tenfold spread for a supposedly identical product, which tells you there is no product — just accounts of varying disposability.
Is it safe to buy Bluesky followers? The likeliest damage isn't a ban, it's a hollow number. Bluesky removed 2.08 million accounts in 2025, up 104% from 1.02 million in 2024, per its 2025 Transparency Report — and in my own sample of 6,727 accounts Bluesky itself created over 22 months, 21.7% were already gone when I re-checked them on August 19, 2026. Purchased accounts are drawn from the population that removal rate falls hardest on, so you keep the number and lose the followers behind it.
Can people tell if you bought followers on Bluesky?
More easily than on any other network. Bluesky's follow graph is public and unauthenticated — anyone can call app.bsky.graph.getFollowers with no login, no API key and no approval, and every follower comes back with a createdAt account-creation timestamp. Three hundred followers registered in the same week is visible to any sponsor or rival in about a minute. One caveat before you accuse anyone: an empty profile proves nothing on its own, because 78.6% of the real, surviving accounts in my sample have never posted either.