"TikTok captions" is two different things wearing one word, and most of the confusion in this topic comes from mixing them up. One is the description — the text field under the video where you write your hook, your context, and your hashtags, and which TikTok increasingly treats as search-and-recommendation copy rather than a throwaway line. The other is the subtitles — the words burned onto the video itself, the ones that let the roughly four in five people watching with the sound off follow what you are saying. They are governed by different rules, sized by different constraints, and written for different readers, and getting good at TikTok means getting both right on purpose. This guide separates the two cleanly, then goes deep on each: how long a description can actually be and why only the first line does the heavy lifting, the 1080×1920 canvas and the safe zones the interface eats, how to style subtitles so they read on a phone at arm's length, and how automatic caption generation works in 2026 — in-app versus before upload, what you can edit, and the one hard limitation (you cannot add captions after a video is posted). The goal is a caption practice you can run across every video without re-deciding it each time.
Almost every mistake people make with TikTok captions starts with a single confusion: "caption" refers to two completely different things, and they follow different rules. The first is the description — the text field you type into when you publish, holding your hook, any context, and your hashtags. TikTok reads that text to help understand what the video is about and who to recommend it to, so it is copy with a discovery job, not a throwaway line. The second is the subtitles — the on-screen text that transcribes what is being said, so someone watching with the sound off can follow along. The large majority of short-form video is watched muted, which makes subtitles less of a nicety and more of a requirement for the video to land at all.
These two get merged constantly because the platform and most guides use "caption" for both. Keep them apart and the whole topic gets simpler: the description is a writing problem (what words get you found and get the tap), and the subtitles are a design problem (what text is legible and correctly placed on a moving phone screen). This guide handles them in that order, then covers how automatic caption generation fits, because auto-captioning tools sit squarely on the subtitle side and are frequently mistaken for a substitute for a good description. They are not — you need both.
The TikTok description has grown a lot. It started at a few hundred characters, expanded to 2,200, and native-app reports now put the ceiling as high as 4,000 characters — while third-party schedulers and the posting API often cap lower, commonly at 2,200. The practical rule: treat 2,200 as your safe planning limit so a description written once works whether you post in-app or through a tool, and do not build a workflow that assumes 4,000 everywhere. If you are unsure of the exact current cap for your posting method, test it rather than trusting a number, because this is precisely the kind of spec platforms change quietly.
Here is the part that matters more than the limit: TikTok only shows the first line or two of the description before it collapses behind a "more" cut, which is roughly the first 100 characters. Everything a scrolling viewer sees at first glance — and much of what TikTok Search reads with the most weight — lives in that opening. So the length of the field is almost irrelevant; the discipline is front-loading. Your hook and your main topic keyword have to be in that first line, before the fold. "POV: you finally fixed your product photos" earns the tap; "Hey guys, so in this video I wanted to talk about…" wastes the only real estate that counts. Write the description as a hook plus context, not as a summary of the video.
Subtitles are a layout problem, and TikTok's interface is the constraint. Build the video on a 1080×1920 canvas at the 9:16 vertical aspect ratio — that is the native TikTok frame. The trap is that the app draws its own elements over that frame, and anything you place under them gets covered. The bottom band, roughly the lower 400-plus pixels, is occupied by your username, the description, and the scrolling audio ticker. The right side, about 140 pixels in, holds the profile picture and the like, comment, and share buttons. The top holds the status bar and the For You / Following tabs. And the outer edges can clip on some devices.
The consequence is counterintuitive but consistent: the bottom of the frame — where most default caption tools drop subtitles — is the worst place for them on TikTok, because that is exactly where the interface lives. Put subtitles in the middle band of the frame, slightly above center, keeping them clear of the top status area and well above the bottom UI. Two to three lines maximum, with two lines being ideal, so the text never becomes a wall. And there is no substitute for previewing on a real phone before you publish; a caption that looks fine in a desktop editor can sit directly behind the audio ticker on an actual device.
Legibility beats style. The single most reliable subtitle look is white text with a dark stroke or a solid background block behind it, because it stays readable over bright footage, dark footage, and everything in between — the one thing a caption cannot afford is to vanish into the video behind it. Keep the type large enough to read on a phone held at arm's length, which is bigger than most people default to. The test is simple: if you have to lean in to read it, so does every viewer, and most will just scroll instead.
Beyond color, subtitle styling is mostly about timing — how the words appear relative to the speech. Three patterns cover almost everything. Word-by-word highlighting, where each word pops as it is spoken, suits fast-paced, punchy content and short clips; it is the energetic "TikTok" look, best kept to videos around 30 seconds or under so it does not exhaust the viewer. Static blocks, where a full phrase sits on screen for the length of the line, suit explainers and demos where the viewer needs a moment to read and absorb. Karaoke-style timing, where a line builds as it is spoken, works well for talking-head content and gives a middle ground between the two. Pick one and hold it across a series — consistency reads as a brand, and switching styles video to video reads as unfinished.
Auto-captioning is where subtitles get fast. TikTok has a built-in auto-captions feature: in the editing screen after you record or upload, you turn it on and the app transcribes the spoken audio into on-screen subtitles, in a growing list of languages whose availability varies by region. Crucially, you can then edit the generated text to fix transcription errors — and auto-transcription reliably mangles proper nouns, brand names, and numbers, so this review step is not optional — and you can adjust the font, color, and placement. Viewers who want to hide captions can do so from the share panel, so what you add is closer to real accessible captions than to permanently burned-in text.
There is one hard limitation to plan around: you cannot add captions to a TikTok after it is posted. Captioning has to happen before you publish, full stop. That single rule is the strongest argument for the alternative approach — captioning before upload, in an external tool, rather than in the app. Doing it before upload gives you three things the in-app feature does not: full control over wording, line breaks, styling, and placement; consistent branding across every video instead of TikTok's defaults; and reusable templates you can carry across TikTok, Reels, and Shorts so the same clip is captioned once and posted everywhere. The in-app auto-captions are genuinely good for a one-off post; a before-upload workflow is what you want once captioning is a repeated job, not a one-time task. The step-by-step for both routes is in the how-to on adding captions to a TikTok video.
Back to the writing side, because the description is where discovery is won or lost. Three rules carry most of it. Front-load the hook and your main keyword into the first ~100 characters, before the fold — this is the same discipline as the subtitle-placement rule, applied to text: the important thing goes where it is actually seen. Write for the viewer who just watched, not as a pre-summary — a description that adds a line, asks a question, or extends the video's point earns comments in a way that restating the video does not. And use a focused group of relevant hashtags rather than a wall of them; a handful of precise tags that actually describe the video helps categorize it, while twenty generic tags read as spam and do little for reach.
The reason keywords in the description matter more than they used to is that TikTok Search and recommendations increasingly read the text to understand a video, the same shift that made the caption the most important copy on Instagram — covered in the Instagram caption strategy guide — playing out on TikTok too. So the words you would want a viewer to search should appear naturally in the description. This is also where the two meanings of "caption" reconnect: your on-screen subtitles carry the spoken keywords for the muted viewer, and your description carries them for the search system, and a video that does both is legible to a person and to the algorithm at once. For the front-of-video half of this — the first three seconds a hook has to survive — see the how-to on optimizing the first three seconds, and for why captions drive retention specifically, video captions for retention.
The through-line of everything above is that captions — both kinds — are load-bearing, not decorative. Subtitles are what let most of your audience follow a muted video, which protects the watch time and completion the algorithm rewards; the description is what tells TikTok what the video is and gets it recommended and found. Skipping either is the quiet reason a genuinely good video underperforms: it was unfollowable on mute, or the system never understood it. This is the argument of the broader captions-first video strategy guide — designing for the sound-off, subtitle-native viewer from the start rather than adding text at the end.
The operational challenge is that doing all of this — correctly placed, well-styled, error-checked subtitles plus a front-loaded, keyword-aware description — for one video is easy, and doing it for every video across a whole calendar is where it decays. The captioning gets skipped on a busy week, the description reverts to "new video ✨," the styling drifts. Consistency, not any single perfect caption, is what compounds on TikTok, and consistency is a systems problem. The last section is about making the whole caption practice repeatable so it survives contact with a real posting schedule.
The reason captions decay across a calendar is that they are a per-video manual chore in the standard workflow — you record, then remember to open the caption editor, then style, then check placement, then write a description, for every single post. Kompozy, the BILT Kontent Engine, is a full content generation and multi-platform publishing engine, and it removes that chore by making captions part of how a video is produced rather than a step you perform afterward. When Kompozy renders a vertical short — a Persona Short fronted by an on-brand AI Influencer avatar, or a clipped vertical cut from long-form footage — the subtitles are burned into the frame during the render, styled from a caption preset and placed in the readable middle band, not dropped at the bottom where TikTok's interface would bury them. You never open the in-app caption editor, which also sidesteps the "cannot add captions after posting" trap entirely, because the video arrives already captioned.
The description is handled as its own job on the writing side. Because Kompozy generates the on-screen subtitle and the post description as two distinct outputs — the subtitle transcribes the video, the description is copy written for discovery — each gets the treatment this guide argues for: the description is drafted through a Persona Brief that carries your voice and a banned-word filter that kills the "hey guys, so in this video" warm-up openers that waste the pre-fold line, so the hook lands in the first ~100 characters by default. And because the whole thing runs from one source idea, the same clip becomes a captioned TikTok, a captioned Reel, and a captioned Short with descriptions shaped for each, plus a carousel or text post if the idea suits it — the reusable-template advantage of before-upload captioning, applied across every platform at once. Autopilot then schedules the approved posts across eight social platforms plus blog and email through a per-post review pipeline where you sharpen the hook or fix a transcription slip before anything ships.
The honest boundary: if you want to caption a single one-off TikTok, the app's built-in auto-captions are free, fast, and perfectly good — do that, and use the sizing and styling rules above to place them well. Kompozy is not a one-click captioner for a finished clip you filmed on your phone; it is the engine for the other job, where captioning and description-writing are a repeated task across many videos and platforms and the manual version is what quietly falls apart. For the wider vocabulary around this — subtitles, closed captions, sound-off design — the creator's AI glossary defines the terms by what each changes about the work of publishing.
They are two different things. The caption (or description) is the text field you type into when you post — your hook, context, and hashtags — and TikTok reads it to help decide what the video is about and who to recommend it to. Subtitles are the words displayed on the video itself, transcribing what is spoken, so people watching with the sound off can follow along. This guide covers both, because a strong TikTok post needs each one done deliberately.
The description field has expanded over time — from a few hundred characters originally, to 2,200, and native app reports now put it as high as 4,000 characters, with schedulers and the API often capping lower at 2,200. Treat 2,200 as the safe planning ceiling. But length is not the point: only the first line or two shows before the "more" cut, so your hook and main keyword have to land in roughly the first 100 characters no matter how long the field allows.
In the middle band of the frame, slightly above center. TikTok's interface covers the edges: the bottom roughly 400 pixels hold your username, description, and the audio ticker; the right side holds the profile and engagement buttons; the top holds the status bar and tabs. Subtitles parked at the bottom — where most default tools place them — get buried. Build on a 1080×1920 canvas and preview on a real phone before publishing.
In the editing screen after you record or upload, you can turn on auto captions and TikTok transcribes the spoken audio into on-screen subtitles. You can edit the text to fix errors and adjust font, color, and placement. Availability spans a growing list of languages but varies by region. The key limitation: you must add captions before the video is posted — you cannot add them to an already-published video — and viewers can toggle them off from the share panel.
Yes, indirectly and directly. On-screen subtitles let the large share of sound-off viewers follow the video, which lifts watch time and completion — signals the algorithm weighs heavily. A keyword-rich description helps TikTok Search and recommendations understand and surface the video. Neither is a magic reach button, but together they remove two common reasons a good video underperforms: nobody could follow it muted, and the system did not know what it was about.
"TikTok captions" means two things: the description field you type when posting (your hook, context, and hashtags, which TikTok reads for search and recommendations) and the subtitles burned onto the video for sound-off viewers. The description can run to about 2,200 characters (native app reports up to 4,000), but only the first ~100 show before the fold, so front-load your hook. Subtitles belong in the middle band of a 1080×1920 frame, styled white with a dark stroke, and must be added before a video is posted.
Get started → · ← All guides · Compare Kompozy vs other tools