Hashtag research with AI is a Skill job you finish in one sitting, about an hour for the first sets and twenty minutes each month after that. You give ChatGPT or Claude your services, your town, your customers and the platforms you post on, and it returns a short, grouped set of candidate tags for each platform. Then you search every tag in the app itself and keep only the ones where posts like yours already live. The AI builds the sets. The app tells you which tags are real, and you decide which ones stay.
- Shape
- Skill
- Sitting
- one sitting
The shape: Skill
A Skill job is a brief you write once and reuse: one prompt, one answer, and your judgment on top. You connect nothing to the AI and build no system.
Hashtag research fits this shape because most of the work is sorting, and sorting is what the AI does well. A small business owner who sits down to pick tags usually types the first ones that come to mind: the trade, the city, maybe a word like "smallbusiness". The AI can lay out the whole field at once, in groups: what you do, where you do it, who you do it for, and the communities you belong to. Seeing the groups side by side shows you gaps, such as a neighborhood name you forgot or a service you never tag.
The limit of the shape is just as important. The AI has no live view of any platform. It cannot see how many posts sit under a tag today, whether the tag is active, or whether it has filled up with spam. Some of the tags it suggests will sound right and be empty, and a few will mean something you did not expect. That is why the job splits in two: the AI drafts the sets, and a quick search in each app checks them.
What this means for you: treat the AI's list as candidates, never as a finished set. The part that makes the sets good is the ten seconds you spend looking at each tag in the app.
The job also has a rhythm: per platform, per niche, refreshed monthly. You build one set for each platform you post on, for the kind of work you do, and you look at it again once a month.
What you give the AI
The AI can only suggest tags as specific as what you tell it. Give it these, by name, in one message:
- Your services, in the words customers use. "Drain cleaning, water heater repair, sewer camera inspection" beats "plumbing services".
- Your town, plus the neighborhoods and nearby towns you serve. List the ones you would actually drive to.
- Who your customers are, in one sentence: "homeowners in older houses who want one plumber they trust".
- The platforms you post on, and only those. A set for a platform you never open is wasted work.
- The communities you belong to: a local business group, a trade association, a market you sell at, an event you sponsor.
- What is coming up this month: a seasonal service, a holiday, a local event, a new offer.
- Tags you already use, if any, so the AI can sort them into keep, check and drop.
- Tags you never want, such as anything political, anything a competitor owns, or words your industry rules forbid.
If you post in a regulated field, such as health, finance or law, say so in one line. The AI should then avoid tags that suggest results or claims you cannot make, but check its list for them anyway.
What comes back
You get one set per platform, each split into groups, with a short note on each tag saying what it describes. You also get a short list of tags to check first, and a list of broad tags the AI left out on purpose, with the reason.
Paste this brief into ChatGPT or Claude, with your inputs below it:
Build hashtag sets for my business, one set for each platform I list below.
Rules:
- Group each set: what I do (service), where I do it (place), who it is for (customer), and communities I belong to.
- Every tag must describe my business or a post I would really make. No tags for things I do not offer.
- Prefer specific tags over broad ones. Leave out one-word general tags like #business, #love or #viral, and list the ones you left out with one line on why.
- Give each tag a short note: what it describes and which kind of post it fits.
- Mark any tag you are unsure exists or is in active use as "check first". You cannot see the platforms, so say so where it matters.
- Never use a tag from my "never use" list. No tags that make a claim about results.
- Keep each platform's set short enough to use, and match the habits of that platform. Do not copy the same list across platforms.
- Add a "this month" group with tags for the dates and events I list.
- After the sets, give me a three-column table I can paste into a spreadsheet: tag, platform, status (all "unchecked").
My services, in customers' words:
[list]
My town, neighborhoods and nearby towns:
[list]
Who my customers are:
[one sentence]
Platforms I post on:
[list]
Communities I belong to:
[list, or "none"]
Coming up this month:
[dates, events, seasonal services]
Tags I already use:
[paste, or "none"]
Tags I never use:
[list]Paste the table into a sheet, then open each app and search each tag. Change its status to keep, drop or watch as you go. If a whole group comes back thin, ask once more with a single direction, such as "ten more place tags, using neighborhood names only". Asking for round after round tends to drift into broader and broader tags. Next month, paste the same brief with your kept and watch tags under "Tags I already use" and the new month's events, and the AI sorts them into keep, check and drop.
How to judge it
Judge the AI's list first, then judge each tag in the app. You can reject a bad set in a minute once you know the signs.
A bad set looks like this:
- It is mostly broad tags. A list full of words like marketing, smallbusiness or instagood puts your post in a pile so large it disappears.
- It invents a community. A tag that sounds like a local group but has no posts under it does nothing. The AI often builds these by joining a town and a trade into one word.
- It describes things you do not do. A tag for a service you do not offer brings the wrong people, and they leave.
- It is the same on every platform. One list pasted across Instagram, TikTok and LinkedIn means the AI ignored how each one is used.
- It promises results. Tags that claim the best, the cheapest or a cure are claims you would have to stand behind.
Then check each tag in the app. A good tag passes three tests when you search it:
- There are recent posts under it. Open a few of the posts the app shows and look at their dates. Recent means this month, not two years ago.
- The posts look like yours. Other local businesses, real customers or people in your field, not spam, giveaways or unrelated content.
- You would be glad to sit next to them. If the top posts are something you would not want your business shown beside, drop the tag, however well it fits on paper.
A tag that passes all three goes in the keep column. A tag that is real but quiet goes in watch, and you look again next month.
What stays with you
The AI can sort a long list into a short one. It cannot see the platforms, and it cannot post for you.
These parts stay with you:
- The check in each app. Only a search inside the app shows whether a tag is alive and what sits under it. No prompt replaces it.
- The current limit. Each platform sets its own rules for how many tags a post can carry, and those rules change. The live composer is the source of truth, so check it rather than trusting any number from the AI or from an old article.
- Fit with the post. A set is a menu, not a stamp. Pick the few tags that describe the post in front of you, and leave the rest.
- The meaning of a tag. Some words mean one thing in your trade and another thing online. You are the one who can spot a tag that would embarrass the business.
- The monthly refresh. Once a month, search the kept tags again, drop the ones that went quiet or filled with spam, and add the month's events. The sheet from your first sitting makes this a twenty-minute job.
- The post itself. Tags help people find a post. They cannot make a weak post worth stopping for. The caption and the image do that, as our Instagram caption best practices explain.
An example (invented business)
This is an editorial example with a made-up business, to show the shape of the work.
Say you run a one-van mobile bike repair business in a mid-sized town, and you post on Instagram and TikTok. You list your services as flat repair, tune-ups and e-bike brake service, your area as the town and three nearby suburbs, and your customers as commuters and families who do not want to haul a bike to a shop. You add the Saturday farmers market where you set up a stand, and the spring ride the town holds each year.
Part of the Instagram set might come back like this:
Service: #mobilebikerepair #ebikerepair #biketuneup
Place: #[town]cycling #[town]bikes #[suburb]local
Customer: #bikecommuter #familybikeride
Community: #[town]farmersmarket
This month: #[town]springride (check first)
Left out: #cycling, #bike (too broad), #bikelife (mostly unrelated posts)You open Instagram and search each one. The service tags have steady, recent posts from repair shops and riders, so they stay. One of the town tags is lively, but the other has only a handful of posts from years ago, so it goes to watch. The farmers market tag is full of posts from other stallholders, which is exactly the company you want, so it stays. The spring ride tag the AI made up does not exist, but the event's own page uses a slightly different spelling, so you keep that one instead.
The TikTok set comes back shorter and leans on how-to and repair words, because that is what people search there. You check it the same way, in the TikTok app.
One set or a set per platform
There is one real choice here, and how many platforms you post on mostly decides it.
| Your situation | What to do | Why |
|---|---|---|
| You post on one platform | Build one set, check it, refresh it monthly | All your time goes where your customers are |
| You post on two or three platforms | Build a short set for each, with place tags shared where they check out | Each platform's search works differently |
| You post the same content everywhere through a scheduler | Keep one place group, then add a few platform tags per post | A shared base saves time, the platform tags keep it honest |
Whichever you choose, keep the sheet. It is the input for next month, and it turns this from a fresh search into a short review.
Where the same sheet goes next
The tags you kept are also a list of the conversations your customers are in. Searching them now and then is a simple start on social listening, and the tags nearby businesses use are worth a look during a competitor analysis.
Of every job on this list, social media is the one that comes back every week, and Boomp is software that does that one for you. See ten posts made from your website, free
