Hashtags still work as a topic label and as a way of finding what your own community posts. What they barely do any more is hand out reach on their own, and no network returns the reach a given tag brought in, so any specific figure about them was invented somewhere. A few relevant ones, and move on.
What is known, and what nobody publishes
Start with the uncomfortable part. None of these platforms publishes how much weight a hashtag carries in distribution, and none of them returns, through its API, the reach a given tag brought in. Which means the guides handing you an exact number of hashtags, or a percentage of reach attributable to them, did not measure it: they inferred it from their own experience or copied it from another guide that did the same.
What can be said with some foundation is what the platforms themselves have said in their documentation and their creator channels, and what shows up in any account with history. Both point the same way: the weight has moved from the caption to the content of the piece. The audio, what is visible in the video, the on screen text and who you interact with describe your content better than five words with a hash in front of them.
And there is a third thing you can measure yourself, on your own account, with your own data, and it is the only part of this that will be true for you specifically. The last third of this guide sets that test up, because it is an easy one to run badly.
The three jobs they still do
They still do three jobs, and all three are real even though none is the one they were credited with five years ago.
The first is labelling: they tell the network, and the person reading, what the piece is about. In a narrow niche with its own vocabulary that still has value, because there are people who follow or search those exact terms. On a broad, crowded topic it has none, because your piece joins a list where it cannot be seen.
The second is collection, and it is the most underrated. A branded, campaign or event hashtag is how you find what other people publish, not how other people find you. If you work with creators, with clients who post on their own or with a community that shares, that tag is the only index you will ever have of all that material. It is the one category of hashtag that has become more useful over the years.
The third is social convention. In some communities, not using the tag everybody uses reads as not being from around there. It does not move reach, it moves belonging, and that counts too.
None of those three has anything to do with distribution, which is the quiet point of this whole section. Once you stop expecting tags to bring strangers in, choosing them takes about fifteen seconds and stops being a monthly argument with whoever writes the captions.
Where to put them, network by network
The perennial argument is caption or first comment. For reach it makes no difference, and for readability it makes a lot: the first two lines are all anybody sees without tapping more, and spending them on hashes spends the best space you own. The only practical objection to the first comment is that somebody has to post it, and that is only automated on four networks.
| Network | Automatic first comment | Caption characters | Audience measure |
|---|---|---|---|
| Yes | 2,200 | Reach and impressions | |
| Yes | 63,206 | Reach and impressions | |
| Yes | 3,000 | Reach and impressions | |
| YouTube | Yes | 5,000 | Plays |
| TikTok | No | 2,200 | Plays |
| Threads | No | 500 | Not measured here |
| No | 500 | Not measured here | |
| Bluesky | No | 300 | Not measured here |
On those four networks the first comment goes out by itself, alongside the post, with nobody having to come back to it. On the other ten a person writes it by hand after publishing, which is a small chore and a perfectly sustainable one at three posts a week and stops being one at fifteen. That is the whole practical difference between the two left hand columns.
The last column decides whether this argument concerns you at all. If you publish mostly on networks with no audience metrics, no hashtag test you run will give you a readable result, and the decision becomes purely editorial: use the ones that help somebody understand the piece and stop thinking about it.
The character column carries its own lesson. A block of 30 tags takes 300 to 500 characters, which is the entire caption on Bluesky, on Pinterest and on Threads. If you send the same piece to several networks from one place, that block is not decoration, it is what pushes your words out.
When a piece travels from one network to another under a standing rule, you can choose to leave the tags and the mentions behind. That is the sensible option when the destination has a short caption or a different vocabulary, and it saves rewriting by hand what you had already written.
How many, and the ones that do damage
Three to five relevant ones is what most people who actually measure settle on, not because there is a definitive study but because above that you start paying costs with no visible return. Blocks of thirty are not penalised and add nothing either, so what they do is clutter a caption people do read.
Four kinds do subtract, and those deserve a list.
- The huge generic ones. A tag with millions of posts will not give you visibility, and it does tell the network your piece belongs to a topic you cannot compete in.
- The ones that do not describe the piece. Using the trending tag to slip into it is the fastest way to put your content in front of people who did not want it, with the matching effect on engagement.
- Industry tags instead of customer tags. The words agencies use with each other are never searched by the people who buy.
- Blocks copied from another account. A set inherited from an account selling something else drags its audience along, and that audience is not yours.
One warning about reach, since it is where most of these conversations break down: if you test a change of tags and look at reach, remember that reach counts people rather than times and that it cannot be added across days. How to read it without tripping is in the reach guide.
Everything in this article happens in one place in GoFeed.
Try it freeHow to check it on your own account
The only test worth anything is yours, and it is built like this. Pick one format and one network, both fixed. Publish six pieces with your usual block and six with three relevant tags and nothing else, alternating them, not the first six one way and the last six the other, or you are comparing two fortnights rather than two treatments.
Compare impressions or plays, not reach, because those are the figures that do add up across pieces. And compare the median rather than the mean, since one runaway piece decides the whole result on its own.
The likeliest outcome is that you find no difference, and that is a perfectly useful result: it means you can stop spending time on this and spend it on what does move the needle, which is usually cadence and format. The first of those is covered in the posting frequency guide.
What to do with the time you save
Taking the hashtag ritual off the checklist frees a few minutes per piece, and there are three places where those minutes pay better.
The first is the opening line of the caption, which is what decides whether anybody taps more. The second is alt text, which almost nobody writes and which, besides being an accessibility matter, describes the image to the network. The third is replying, still the highest return activity there is and the first to fall away when things get busy.
If after all of this your account still is not growing, the problem was never the tags and was never going to be. The order to look for the cause in is in the stalled account guide, and how to get the caption, the first comment and the alt text settled in one piece that goes out to several networks is on the calendar and publishing page.