Hashtag Set Builder

What hashtags do now, and what they stopped doing

Hashtags on Instagram have quietly changed jobs. For years the pitch was reach: tag a post, land in a hashtag feed, get found by strangers browsing it. Two things eroded that.

The first is that Instagram removed the ability to follow a hashtag, so a tag no longer pushes your post into anybody’s home feed as a subscription. The page still exists and can be browsed, but the passive delivery route is gone.

The second is that Instagram’s own leadership has publicly played down hashtags as a distribution mechanism, describing them as a way of saying what a post is about rather than a way of getting more of it seen. Instagram has never published how hashtags are weighted in ranking, so nobody outside the company can state their effect as fact — but the platform’s own messaging points away from hashtags as a growth tactic.

What replaced them, partly, is text: Instagram search reads keywords in captions and account information, not just tags. Hashtags are still worth setting deliberately — they are a compact topic label and they put you on a browsable page — but they are no longer the lever. Treat them as filing, not marketing.

Why thirty mega-tags is the worst set

• Visibility on those pages is measured in seconds. The most-used tags absorb an enormous, continuous stream of posts. A specific tag with a slower stream keeps you visible far longer.
• The audience is wrong. Someone browsing an enormous generic tag is browsing nothing in particular. Someone browsing a specific community tag is looking for exactly the thing you posted.
• The label is meaningless. If hashtags describe a post, thirty tags that describe everything describe nothing.

The buckets, and what each is for

Every tag in the bundled dataset carries a size bucket. These are editorial classifications from a curated dataset shipped with this page — not live Instagram data. This tool makes no network requests of any kind, so it cannot look anything up. That is exactly why no post count is displayed anywhere: Instagram publishes no feed of hashtag volumes, and a precise-looking number nobody can verify is worse than no number at all. What a bucket tells you is the rough character of a tag — how generic it is, and what job it should do in a set.

The five buckets collapse into three roles, and the tool labels every tag it gives you with its role rather than leaving you to guess.

Tags built from your own seed keywords are treated as community tags too, on the same reasoning: they are specific to you, so anyone finding them found you deliberately. The mix slider moves the balance between broad and niche, and the set is then filled by weight, so pushing it toward niche takes tags out of the top two rows and puts them in the bottom three.

Seed keywords and two-niche blending

You can add up to five seed keywords. Each is cleaned into a valid tag and expanded into a few variants — the word itself, the word with daily, with ofinstagram, and joined to your niche name. Seeds take up to about forty per cent of the set, so they never crowd out the curated tags. Anything that is only punctuation or only digits is skipped with a note saying why, and a seed that is itself engagement bait is refused outright, because everything built from it would smuggle the bait back in.

The second niche control is for accounts that genuinely sit across two categories — a food photographer, a running coach who also posts recipes. Set the blend anywhere from 10 to 90 per cent and the two pools are interleaved rather than stacked, so the set alternates instead of clumping. Tags that exist in both niches are counted once, and when blending is on each tag is labelled with which niche it came from. If you pick the same niche twice, the tool says the blend is off rather than pretending to do something.

The rotation planner

Ask for anywhere between two and seven sets and the planner builds them from the same settings, drawing from different points in the pool so they share as few tags as possible. Then it does the part most tools skip: it measures the overlap and shows you the number.

Overlap here is the share of the smaller set’s tags that also appear in the other set, averaged across every pair, with the worst pair named separately. Under 25 per cent is green, up to 55 per cent amber, above that red. It will not be zero for long, and the tool says so: a niche holds a finite number of curated tags, so once the sets outnumber the pool they have to repeat. Adding seed keywords or blending a second niche is what actually widens it — asking for more sets does not.

Why rotate at all? Because it stops you posting an identical block every time, and because varying your sets is the only way to find out which ones do anything for you. It is not a workaround for a rule. Nobody outside Instagram knows what repeating a block does, and anyone quoting you a precise number of days between reuses is guessing.

The block checker

The paste box takes any hashtag block — one of your old posts, a competitor’s, one you found in a spreadsheet — and tells you what you are actually holding. Nothing is uploaded; the check runs in your browser like everything else here.

Every token is validated first: the hash and any punctuation are stripped, letters, numbers and underscores are kept, and anything left empty, made only of digits, or longer than 40 characters is reported as unreadable rather than silently dropped. What survives is then sorted:

• Unique tags and how many of your thirty slots remain — or how many to remove if you are over.
• Duplicates, named. A repeat is not counted twice; it is simply not counted, so the slot is wasted.
• Engagement bait, listed in red, with the reason it should go.
• Recognised tags, each with its bucket, its role and which niches it appears in.
• Tags not in the bundled dataset — which is a statement about the dataset, not about the tags.

That last group matters: the dataset covers 32 niches and 857 tags, so brand names, place names, your own coinages and anything in another language are not in it. Being unrecognised is not a verdict and certainly not a ban. One extra warning fires when more than 60 per cent of the recognised tags sit in the two broadest buckets, because a block of only huge tags is the most common way a set does nothing. There is also a copy the cleaned block button: your tags with the duplicates and the bait removed, trimmed to thirty.

About “banned hashtag” lists

The rules a tag has to follow

• Letters, numbers and underscores only. A space, hyphen, apostrophe or full stop ends the tag at that character.
• A tag cannot be numbers alone, and the builder rejects anything over 40 characters as unusable.
• Accented and non-Latin characters work; substituted “font” characters do not, which is one more reason to keep the font generator away from anything functional.
• Tags are case-insensitive, so #StreetPhotography and #streetphotography are the same tag — but capitalising each word helps humans read it and helps screen readers pronounce it as words rather than one run-on string.
• More than 30 tags and Instagram counts only the first thirty. The builder caps at 30 and tells you when it has.

Saving, reusing and exporting sets

Named sets are kept in local storage in this one browser, newest first, up to forty, and can be loaded, copied, duplicated or deleted. Loading one pins it in place of a freshly generated set until you change a control or press shuffle, so you are never confused about which set you are looking at.

Local storage means what it says: nothing syncs between devices, no account is involved, and clearing your site data removes every saved set. That is why the JSON export exists — it writes every saved set to a file with your own browser, the only copy that survives a cleared cache or a new phone. It is never uploaded, because this page cannot upload anything.

Using a generated set well

• Read every tag before you post it. A generated set is a shortlist, not an answer. Delete anything that does not honestly describe the post.
• Do not reuse one set forever. Saved sets are for speed, not for identical repetition. The rotation planner exists for exactly this.
• Fewer is fine. There is nothing magic about hitting 30. Ten accurate tags is a perfectly good post.
• Include a couple of tags you own. A branded or community tag used consistently builds a browsable archive of your own work — the one hashtag use case that has not degraded.
• Put the effort into the caption instead. Since search reads caption text, one clear sentence naming the subject is likely worth more than the tags below it; the caption counter handles that side.
• Never use a tool that wants your login to “research” tags for you. Nothing on this page needs an account, and the Instagram safety guide covers what those requests usually are.

The honest downside

The buckets are editorial judgement, not measurement — a reasonable sense of how crowded a tag tends to be, bundled with the page and going gently out of date from the day they were written. Check a few in the app before you commit to a set.

The bigger caveat is expectations. Hashtag research is comforting because it feels like a lever you control and it produces tidy output. It is also, on current evidence, one of the lowest-yield things you can spend an hour on. If a post underperforms, the tags are rarely the reason. Build a sensible set in thirty seconds and put the rest of the time into the thing you posted.

And no tool can tell you what is working for your account better than your own Insights. Instagram reports how much of a post’s reach came from hashtags; check it on your own posts for a month and you will have better information than any generator, including this one. The analytics guide covers how to read those numbers, and the captions, bios and hashtags guide puts the tags back in context with everything else you write.

Questions people actually ask