AI content money guide

The Hashtag & Caption Playbook

Every platform pretends its algorithm is a mystery. It isn't. Between publicly-confirmed guidance (Adam Mosseri on Instagram, TikTok's Creator Portal, Meta's Threads FAQ) and community-measured data from tools like Later, Hootsuite, Metricool and Buffer, we know almost exactly how many hashtags to use, what caption length converts, and when to hit publish. This is the copy-paste playbook — one section per platform. Save the infographic above, then read below for the copy-and-example version.

The Hashtag & Caption Playbook

1. TikTok — 3 to 5 hashtags, hook in the first 40 characters

TikTok's For You Page uses a two-stage ranker: a broad candidate-generation pass (where hashtags help you enter the pool) and a fine ranker (where watch-time and completion decide who sees you). Too many hashtags confuse the candidate generator; too few and you don't get sampled at all. 3–5 is the empirically-observed sweet spot from the largest creator datasets.

2. Instagram — 3 to 5 hashtags (Mosseri confirmed), saves are the #1 signal

Instagram's ranking model in 2026 optimises for retention + save-rate over impressions. Saves signal 'reference value' to the algorithm, which then re-shows the post to your followers over the next 7 days, boosting cumulative reach. Hashtags only get you into the pool — saves keep you in it.

3. LinkedIn — 3 hashtags maximum, first 2 lines decide the click

LinkedIn's feed uses a two-hour 'golden window' — if your post gets strong dwell-time and comments in the first two hours, it gets amplified for the next 24. Hashtags help LinkedIn categorise you but they're a much weaker signal than dwell-time and reply-depth. Prioritise a hook that stops the scroll and a question that triggers a reply.

4. YouTube Shorts — hashtags belong in the title, #Shorts is non-negotiable

YouTube Shorts uses a hybrid of Shorts-shelf ranking and semantic-search matching. Hashtags in the title feed both systems; hashtags in the description only feed one. Ranking = watch-through % first, hashtag relevance second — so a punchy title + hook matters 10× more than a long tag list.

5. Threads — one topic tag, reply-bait > hashtags

Threads is optimising for daily-active-users to catch Twitter/X, so its ranker heavily rewards content that drives replies (the metric Meta most needs to prove Threads is 'alive'). Hashtags in Threads are almost cosmetic — reply-count is the real currency.

6. The universal caption formula — regardless of platform

Every platform's ranker rewards the same 3 things in some weighted combination: (1) watch-through / dwell-time, (2) engagement (save, reply, share), (3) predicted retention on next scroll. A tight hook + a specific ask + the correct hashtag stack triggers all three at once. Everything else is decoration.