How Does the TikTok Algorithm Work? A 2026 UK Creator's Guide

Quick answer: TikTok's algorithm ranks videos by three signal groups it names publicly: user interactions, video information and device/account settings. Watch time and completion sit at the top, then saves and shares, then likes. Follower count is not a direct ranking signal. The Nearby Feed launched in the UK on 4 December 2025 and reshapes local reach.

You post, and TikTok shows the video to almost no one. Next post: 40,000 views. Most "how does the TikTok algorithm work" guides recycle 2022 lines and skip what TikTok has rewritten in the last twelve months.

Here is what the algorithm rewards in 2026, sourced to TikTok's newsroom and Ofcom, with a UK-first read on the new Nearby Feed, teen filters and honest guidance on paid help. Every claim links to a primary source.

What is the TikTok algorithm in 2026?

The TikTok algorithm is the recommendation system that decides which videos land on your For You feed. It watches how you behave, reads the videos it might show you, and predicts which ones you will keep watching. Scoring runs on every video and every viewer separately, which is why a brand new account can put up a first post and hit a million views by the weekend.

What TikTok says: "Neither follower count nor whether the account has had previous high-performing videos are direct factors in the recommendation system."

That single sentence, from TikTok's own recommendation post, sits under almost every "how the algorithm works" question. TikTok has no interest in your follower count. It does not remember that your last three videos flopped, and it scores this video on its own merits, against the signals it collects from the small first batch of viewers it decides to show it to.

One useful side effect follows. If your last few TikToks did nothing, the next one starts on the same footing as anyone else's. Buying TikTok followers can shape what a first-time viewer sees on your profile, but it does not tip the algorithm. This distinction runs through the rest of the guide.

The three signal categories TikTok's own docs name

Every ranking-signal list you see online rolls up into three groups TikTok published years ago and has not renamed since. Same three, still the map.

User interactions

User interactions are what the viewer does when your video appears. Finishing the video, watching it again, saving it, sharing it to a DM, commenting, following you afterwards, or tapping "Not interested" all count. TikTok says the stronger the action, the more weight it carries.

A save is louder than a like. A share to a DM is louder still. Someone finishing a 90-second video from cold is the loudest signal a Tier 1 action can send.

Watch time lives here, and it is the biggest lever any creator has on the For You feed. Ignore the rest and get this one right.

Video information

Video information is what the app reads about the video itself. Captions, on-screen text, hashtags, sounds and spoken words in the audio all count. TikTok's system transcribes speech and reads the on-screen text with computer vision, so the first line you say and the words you overlay both feed the same categoriser.

Sounds do a second job: they carry the trend signal, and they group your video with everything else using that same audio. That is why a small account can slip into a big trend on the back of a matching sound. Write for the transcript first and the human reader second.

Device and account settings

Device and account settings are the lightest of the three groups. Language, country and device type sit here.

TikTok calls these "not weighted as heavily", because you chose them once at sign-up rather than through what you keep watching. For a UK creator these settings do two jobs. They put your account in the UK-first pool that the Nearby Feed draws from.

And they help route videos in English to English-speaking viewers first, which is why a Manchester-based cooking channel usually reaches UK feeds before it reaches anyone in Australia. None of them alone will push a video, but together they narrow the first test batch to viewers likely to give it a chance.

The ranking-signal weight ladder

TikTok tells you the ranking system weights signals "based on their value to a user". That single line gives you a usable mental model, and it groups every published signal into three tiers, plus one row for the signals TikTok explicitly leaves off.

Tier 1: the heavyweights. Completion, average watch depth and replays. If most viewers finish the video, replay it, or watch it for a long fraction of its length, the next test batch grows.

Any Tier 1 miss is expensive. TikTok reads a swipe in the first two seconds as a strong negative on the same signal it rewards a full watch on. That symmetry is why the first 2-3 seconds decide most of a post's reach. Get the hook right or nothing else on the page rescues it.

Tier 2: the amplifiers. Saves, shares, comments.

A save says the viewer plans to return to the video later. Shares to DMs say "someone I know needs to see this". Comments say "I have something to add".

These do not carry the same weight as a full watch, but they compound quickly on a video that already has a high completion rate. Viral videos almost always have an outsized share ratio; the like count is a side effect.

Third tier: the light stuff. Likes, follows, taps on caption text. A like is one tap and no cost, and the algorithm treats it that way, weighting it well below any Tier 1 signal you might have earned in the same viewing session. Chasing likes with a big caption does not push a video that no one finishes.

Follows after the video do count, but only in the context of everything else that happened first. Bought TikTok views can lift the visible count on a video, which lifts the social-proof read a first-time viewer gets, but they will not force a bigger next batch on their own. Only real Tier 1 signals do that job.

Not on the ladder. Follower count sits nowhere in the mix. Past video performance is out too, because TikTok scores each new post independently rather than holding a grudge from last week.

Neither is having #fyp in the caption. Nothing TikTok has published in the past six years puts any of those in the ranking mix. Beat one Tier 1 signal and you outrank a caption stuffed with feed hashtags.

How does the TikTok algorithm work when you post a new video?

Every post moves through the same four stages, whether the account is new or well-established. The specific numbers shift with the account's history and its usual audience, but the shape stays the same whether the video comes from a first-time creator in Cardiff or from a two-year-old account in Manchester. Knowing the stages saves you from panicking at hour one and helps you read a post's real trajectory instead of guessing.

Stage 1: the first test batch (minutes). TikTok pushes a fresh video to a small group whose interests match what the video seems to be about: the topic, the sound, the caption, the on-screen text. This batch is usually 200-500 people, and it decides how much oxygen you get next.

Second: response scoring (first hour). TikTok grades the batch's behaviour on Tier 1 signals first.

A strong completion rate earns a bigger second batch, typically several times the size of the first. Middling response holds you steady. Poor response caps the reach quickly and quietly, and no amount of engagement chasing brings the video back up.

Stage 3: compounding (hours to days). Each batch that responds well earns a wider one. This is where the "goes to 40,000 people while nothing else you did looks different" moment happens. Compounding can run for a day, a week, or longer if the completion rate holds.

Fourth: second life (anytime). An older post can re-enter distribution if it starts collecting signals in a new context, such as a trending sound picking up, a search spike, or a share into a large group DM. That is why an evergreen post can suddenly re-appear weeks later, sometimes carrying more reach on the revival than it did on release day.

Practical read-out: judge a post at 24 to 48 hours. If completion is below about 50%, rework the hook or the length before your next post. Waiting for the real number is the habit that separates creators who improve from creators who repost the same weak hook.

What changed in 2025-2026 (the parts other guides missed)

Three changes in the last twelve months matter more than any tweak to the ranking weights. Most guides skipped all three.

TikTok Nearby Feed

TikTok launched the Nearby Feed in the UK, France, Germany and Italy on 4 December 2025. It is a second surface on the home screen, alongside For You and Following, and it shows posts tagged to your rough location.

Location, topic and recency are the three things TikTok names as inputs. Recency matters most. The feed skews to the current week and drops anything over 90 days old. Four eligibility rules apply: the creator is 18 or older, the account is public, the post is not set to "Friends" or "Only You", and the video is under 90 days from posting.

For UK small businesses this adds a second distribution surface. A shop tagging its location on one video a week gets a shot at a local first batch alongside the general For You test. Nothing like it existed last year.

Some earlier guides call it the "Local Feed", but TikTok's own newsroom names it Nearby Feed. Search for the correct name in your app. If you already post local content, UK-based TikTok views help your profile read as established when the Nearby batch arrives.

The Online Safety Act and Ofcom's TikTok investigation

In the UK, the Online Safety Act turned on its child-protection duties on 25 July 2025. Platforms have to run risk assessments and make algorithmic feeds measurably safer for anyone under 18.

In July 2026 Ofcom opened a formal investigation into TikTok's age-assurance systems. Ofcom can fine a platform up to 10% of qualifying worldwide revenue or £18 million, whichever is higher, and it has already used that power against smaller platforms. It is not a paper threat.

The practical read for creators is simple. Teen accounts in the UK see a more heavily filtered For You Page than adult accounts. TikTok quietly caps or removes content on self-harm, cosmetic surgery, gambling, extreme weight loss or age-restricted topics before the ranking scores it for teen feeds.

If your target audience is under 18, treat that filter as part of the algorithm and plan around it. Under-18 exposure has become a scored risk for TikTok itself, and Ofcom will next report in October 2026.

What the DSA transparency report tells us

TikTok's seventh DSA transparency report, published on 31 August 2026 and covering January to June 2026, is a useful sanity check on the algorithm's other job: keeping violating content out of the feed. That job runs quietly in the background of every post you make.

Two numbers to bank: TikTok removed about 104 million pieces of violating content in the six months, and its automated systems handled 94.1% of that without human review. Across the EU, TikTok reports 189 million monthly users.

For a creator the read-out is unglamorous but useful. Borderline hooks, misleading claims and off-policy hashtags will usually trip the automated filter before a human sees them, and that cuts reach quickly. Stay well inside the community guidelines and you avoid the fastest reach-killer on the platform.

What the algorithm doesn't reward

TikTok has spent five years correcting the same handful of misconceptions in public. The ones below come up most often in reader questions, and every row cites the primary source.

A few deserve a longer note than a table row allows. The next three come up in almost every reader question.

Watching your own videos in a loop. Personal replays sit alongside a small set of "related device" signals that carry almost no weight. Doing it forty times a day will not lift the video, and it can distort your own For You feed if you keep doing it.

Deleting flops. Poor past performance is not a ranking factor to start with. Deleting the video does not clean anything up on the algorithm's side, because there was nothing there to clean. If a video is genuinely off-policy, though, deleting it is still the right call for the account's overall safety score.

"Only videos under 15 seconds go viral." TikTok has been rewarding longer videos with high completion rates since late 2024, and its own creator resources encourage 1-3 minute posts for topics that need more room. Watch depth is the constraint here. Raw length is a red herring.

Does buying followers or likes help the algorithm?

The honest answer is no. Follower count is not a direct ranking signal on TikTok, so a bigger number does not force the algorithm to distribute your video more widely. Nothing has changed on that point since TikTok's original 2020 recommendation post, and none of the 2026 updates put it back on the table.

Where paid help matters is a step earlier: the human decision to watch. A profile with 12 followers and a video with 3 likes reads "empty" to a first-time viewer, and first-time viewers skip empty profiles before the ranking system ever sees the retention data. That is why cold-starting a TikTok account is so brutal, and why the cold-start problem is real even for good creators.

A bought boost affects that first read. Numbers on the profile and on the post do the same work the first genuine viewers used to do: they make the account look worth watching. Those extra numbers shift the Tier 1 signals the next real viewer sends. It does not create them.

One or two practical limits. Drip-fed delivery from real UK-based accounts reads more naturally than a big overnight spike. Bought engagement will not save a weak hook or a video with 20% completion. And the safest bridge is a service page for the platform you actually post on, such as TikTok likes matched to the videos already carrying strong watch time.

How UK creators can work with the algorithm in 2026

Six habits that map directly to signals the algorithm rewards. Work through them in order; the first two do most of the lifting.

1. Front-load the hook. The first 2-3 seconds decide the Tier 1 completion signal. If a viewer scrolls before the third second, that is one of the strongest negatives you can send. Open with the payoff, the twist or a face-camera line, and hold the visual movement steady for at least a second before cutting.

2. Write for the transcript. TikTok's system reads speech and on-screen text. Say the topic out loud in the first sentence, and put the same words on screen. That gives the video-information signals two chances to categorise the post correctly.

3. Post to GMT and BST rhythms. UK evenings between roughly 7pm and 10pm compound signals faster because the first test batch is bigger. Match posting time to when your target UK audience is on the app; a generic US-based schedule usually misses the peak.

4. Book one Nearby Feed anchor per week. Tag a location or a service on at least one video per week. That single habit makes the account eligible for the local surface, which is a second batch of eligible viewers on top of For You.

5. Do not repost with the same sound in quick succession. TikTok says two videos in a row from the same creator or with the same sound usually will not both appear in a feed. Space out reuse of any sound by at least a couple of days.

6. Treat paid help as social proof; distribution is a separate lever. Bought numbers change what the first organic viewer sees on your profile and post, and that shifts their Tier 1 behaviour. Keep any delivery slow, keep it on-topic per platform (TikTok comments for engaged posts, TikTok Live viewers for streams), and pair it with a real hook you have already tested. Our FAQs cover the 30-day refill guarantee if a delivery misses its window.

Get the first two right and the rest compound. Everything on this list is a lever the algorithm has told you it uses.

Frequently asked questions

How long does it take the TikTok algorithm to pick up a new video?

Minutes for the first exposure, but 24 to 48 hours for a real read. Distribution starts with a small test batch immediately, then widens based on watch depth, replays and saves. Rework the hook if completion sits below 50%. See why your TikTok views may be low for symptom-level triage.

Can I reset my TikTok For You Page?

Yes. In Settings and privacy > Content preferences you can refresh your For You feed. TikTok clears the interest signals it has built for your account and starts again from your device and country settings. Expect a few days of rough recommendations while it re-learns what you watch, then it stabilises.

Do #fyp and #foryou hashtags help you rank on TikTok in 2026?

No. Nothing in TikTok's public docs shows #fyp or #foryou lifts recommendations, and the algorithm reads all hashtags as one of many video-information signals. Specific, relevant hashtags help TikTok categorise the video and match it to the right first batch. Generic feed tags do not, and they waste a caption slot.

Does watching your own videos on repeat boost their reach?

It doesn't help. Owner replays sit in a small "related device" bucket that carries almost no weight. TikTok scores how a broad, unfamiliar audience behaves. Personal replays and small-circle rewatches carry almost none of that weight; real reach comes from strangers finishing the video.

Is the TikTok Nearby Feed available to UK users under 18?

No. Content only appears in the Nearby Feed if the creator is 18 or older, the account is public, the post is not restricted to "Friends" or "Only You", and the video is under 90 days old. Under-18 accounts can browse the feed, subject to TikTok's teen safety rules, but they cannot publish into it.

How often does the TikTok algorithm change?

Continuously in small ways, and every few weeks in named ways. The three signal categories, user interactions, video information and device/account settings, have stayed the same since 2020. What changes is the surfaces (Nearby Feed, Manage Topics) and the safety filters. Trust TikTok's newsroom over old listicles for the current picture.