Hashtag research that is not superstition

The standard approach to hashtags is to search a term, note which related tags have the largest post counts, and use those. Sometimes a competitor's tag list gets copied wholesale on the reasoning that it seems to be working for them.

Both are guesses wearing the costume of research. The post count next to a hashtag tells you how many posts have used it. It says nothing about how many people browse it, follow it, or would care about yours if they found it. Those are different quantities, and only the first one is visible to you. A tag with millions of posts is a tag where a new post is buried in seconds by the next thousand; a tag with a few thousand posts might be followed closely by exactly the people a business wants.

Copying a competitor's list inherits their guess without inheriting whatever made it work, if anything did. Nobody outside that account can see whether those tags contributed reach or were simply present while other things worked.

It helps to be concrete about the mechanism, because most hashtag folklore comes from being vague about it.

A hashtag is a topic label. It puts a post into a browsable collection, and it can be followed, which means posts using it may surface in the feeds of people who follow the topic rather than the account. That second path is the one worth caring about, and it is why follower-ability matters more than post count: a tag nobody follows can only be reached by someone deliberately browsing it.

This also explains why a wall of loosely related tags underperforms. A post labelled with twenty-five topics is making twenty-five weak claims about what it is about. The platform is trying to work out which collection this post genuinely belongs in, and a scattergun list is less informative than a short accurate one. Meanwhile the tags themselves are visible to a human reader, and thirty of them below a caption reads as a broadcast rather than a post for anyone in particular.

The research that works starts from people rather than from volume, and it is less sophisticated than it sounds.

Open the accounts of people who are already customers, or who comment on posts in your category, and look at what they tag their own posts with and what topics they follow. Look at the tags used by accounts your audience follows that are not competitors — a local food account, a regional interest page, a community group. Then look at your own comments and DMs for the words people actually use for what you sell, which are frequently not the words the industry uses. A studio that describes itself as offering strength and conditioning may find its customers all write gym.

Write the candidates down in three groups: broad topic tags, specific niche tags, and location tags. The broad ones are lottery tickets, the niche ones are where a small account has a real chance of being seen, and the location ones matter disproportionately for any business that serves a particular city.

A tag list can only be evaluated if it is stable long enough to produce a comparison, and changed in small enough increments that a difference can be attributed to something.

The workable method is to fix a set of tags and use it across about three posts of similar type, then change one group and run three more. Comparing a single post against a single post tells you almost nothing, because the variance between two posts of the same kind is large and driven mostly by the content. Three gives a rough sense of a level rather than a point.

Changing everything between rounds is the mistake that makes the whole exercise unfalsifiable. If the caption style, the format, the posting day and the tags all changed, a difference in reach has four candidate explanations and no way to choose between them. Hold the content type roughly constant, change the tag group, and accept that this is a crude instrument. Crude and honest still beats confident and unfounded.

Instagram's own Insights, on a professional account, break down where a post's impressions came from, and hashtags appear as one of those sources. That figure is the only direct evidence available about whether tags did anything, and it is specific to your account rather than borrowed from a chart about someone else's.

Read it as a share, not a total. A post that reached many people with almost none of it attributed to hashtags did well for other reasons, and its tag list should not get the credit. A post with modest reach where a meaningful share came from hashtags is telling you the tags found people the feed would not have.

Two cautions. The figure moves with the content, so a run of posts is more informative than one. And an absent or tiny hashtag share is a real result, not a broken measurement — for plenty of accounts, tags contribute little, and knowing that is more useful than continuing to maintain a list on faith.

If a tag group produces a visible share of impressions across a few posts, keep it and stop rewriting it every time. Tag lists get churned constantly for no reason other than the feeling that something should be being optimised, and churn destroys the only comparison you had.

If hashtags turn out to contribute very little to your reach, the correct response is to spend less time on them, not more. This is the outcome the folklore never mentions, and it is common. The effort saved is better spent on the thing that actually determines reach on every format — whether the post is worth watching or reading — and on the parts of distribution that are within your control, such as posting consistently and answering the people who respond.

A scheduling tool such as Socie can keep a tested tag group attached to a post type so it stops being retyped from memory each time, which removes the drift. It cannot tell you which tags your audience follows; that comes out of the account's own Insights and a look at real customers.