How Much On-Screen Text Should a Reel Have?

Most Reels that lose viewers to on-screen text do not have too little of it. They have a paragraph in frame one, a second paragraph at second four, and a viewer who stopped reading and started scrolling somewhere in between.

Here is the short answer to what you searched. A Reel needs one short hook line in the opening second, then text that never asks the viewer to read for longer than the shot is on screen. Past the hook, on-screen text should carry one idea per card, in a handful of words, and it should add something the voice or the picture is not already saying. More text is not more clarity. On a feed, it is more work, and work is what people swipe away from.

We covered whether subtitles help at all in Do On-Screen Captions Actually Increase Views?. This post is about the amount: how much text a Reel can carry before the text starts costing you the viewer it was meant to keep.

The real limit is reading time, not screen space

Creators usually judge on-screen text by how it looks in the editor. Does it fit, is it centered, is the font clean. The viewer judges it by something else entirely: can I finish reading this before the shot changes?

People read short phrases on a phone at roughly three to four words per second, and that is when they are paying full attention. A viewer deciding whether to keep watching a Reel is not paying full attention. So a practical budget looks like this:

• A card on screen for one second can carry two or three words.
• A card on screen for two seconds can carry about five or six.
• A card on screen for three seconds can carry a short sentence, and only if nothing else on screen is moving.

When the text outlasts the shot, the viewer is still reading while the picture moves on, which feels rushed. When the shot outlasts the text, the viewer has read everything and is waiting, which feels slow. Both feel wrong, and both push the thumb. The sweet spot is text that finishes a beat before the cut.

A quick test you can run in the editor

Play your Reel at normal speed and read every card out loud at a calm pace. If you cannot finish a card before it disappears, it has too many words. If you finish and then sit through half a second of silence on the same frame, the card is either too short for its shot or the shot is too long. Fix the shot first. Trimming footage is usually better than padding text.

The hook line: one sentence, readable in a glance

The first second of a Reel is the only moment where on-screen text decides whether anyone sees the rest. In that second, the text has one job: tell a muted viewer why to stay.

That means the hook line should be:

• Short enough to read in one glance. Six to eight words is a comfortable ceiling. Two lines at most.
• A stake, not a label. “Morning routine” is a label. “I stopped doing this and slept better” is a stake. A label tells the viewer what the video is. A stake tells them why it matters to them.
• On screen from the very first frame. Text that fades in at second one has already missed the viewers who decided at second zero.

The hook line works alongside the picture, not instead of it. The frame communicates category in an instant (a kitchen, a gym, a face mid-reaction), and the line adds the tension. We break that split down in visual vs verbal hooks. If the frame and the line say the same thing, one of them is wasted.

Full subtitles vs text cards: two different tools

There are two kinds of on-screen text in a Reel, and creators often stack both without noticing.

Subtitles follow your speech word by word. Their job is to make the spoken video work muted. They should be short chunks timed to your voice, sitting low in the safe middle band of the frame, and they can run the whole length of the video.

Text cards are written separately. A number, a step name, a twist, a label on what we are looking at. Their job is to add structure the voice is not already giving.

The problem starts when both run at once. A subtitle at the bottom reading your sentence, plus a card at the top restating the same idea in different words, gives the viewer two things to read and only one thing to learn. Pick one role per moment:

• If you are talking to camera, let subtitles carry the speech and use cards only for the few beats that need a label (a step number, a price, a before and after).
• If you are not talking, cards carry everything, so keep each one to a single idea.
• If there is a voiceover over B-roll, subtitles plus the footage are usually enough. Add a card only when the picture needs a name.

One idea per card

The fastest way to lose a viewer mid-Reel is a card that tries to say two things. “Step 2: add the oil, but only once the pan is hot, otherwise it sticks” is a full instruction. It is also three ideas, and the viewer reads it once, gets one of them, and feels behind.

Split it:

1. “Step 2: oil.”
2. “Wait until the pan is hot.”
3. “Otherwise it sticks.”

Three cards, each readable in under a second, each landing on its own beat. The video gets more rhythm, not less, because every card is a small payoff. This is the same reason cuts per second matter: a viewer who keeps getting small resolutions keeps watching for the next one.

One idea per card is also what makes a Reel easy to save. A viewer who wants to come back to the steps can scan the cards and find them. A viewer faced with paragraphs will not bother.

Sound-off viewers decide more than their share

Plenty of people watch with sound on. The ones who do not are often the ones scrolling fastest: on a train, in a waiting room, late at night next to someone sleeping. They are also the people your video most needs to convert, because they arrived from the feed, not from your profile.

For them, the on-screen text is the whole verbal channel. So check the Reel with the sound off before you post, and ask three questions:

• Does the opening second tell me why to stay, with no audio?
• Can I follow the main point from text and picture alone?
• Is there any stretch longer than a few seconds where nothing on screen tells me what is happening?

If the answer to the last one is yes, that stretch is where muted viewers leave. You do not need to subtitle everything to fix it. One card naming what is happening is often enough.

When text starts competing with the picture

On-screen text is not free. Every word sits on top of the footage, and some footage needs the space.

Reveal moments. A before and after, a finished dish, a transformation. If a card covers the reveal, the viewer reads instead of looks, and the payoff lands on the text rather than the picture. Clear the frame for the reveal and let it breathe for a beat.

Faces mid-expression. In talking and reaction videos, the face is the content. Text slapped across the mouth or eyes hides the most watchable part of the frame. Keep text in the band below the face.

The edges. Instagram puts its own interface on the right side and along the bottom of a Reel. Text placed there gets covered by the caption, the handle, or the buttons. We cover the exact areas in safe zone composition. Check the Reel in the app, not just the editor, because the overlay is what the viewer actually sees.

Busy footage. Fast motion, patterned backgrounds, and bright scenes all fight with text for the eye. If a shot is visually loud, cut the text down or give it a solid backing so it reads without effort.

A useful rule: if the shot is the payoff, the text gets out of the way. If the shot is filler, the text can carry the beat.

How much text by format

The right amount depends on what kind of Reel you are making. Rough guides from the formats we see most often in our library of analyzed videos:

• Talking to camera: subtitles throughout, plus one hook line and maybe two or three label cards. Nothing more.
• Tutorial or how-to: one card per step, each a few words. The result gets its own clear frame. See the tutorial people save for structure.
• Listicle: a card naming each item as it starts. The number on the card is the retention contract, so keep it visible.
• POV and text-led Reels: the text is the script, so it carries more words, but each card still holds one idea and stays on screen long enough to read twice.
• B-roll with voiceover: subtitles or nothing. Extra cards usually compete with the footage.
• Aesthetic or music-led Reels: a hook line, then mostly clean frames. Text here works as punctuation, not narration.

A pre-post checklist for on-screen text

Before you publish, run through these:

1. The hook line is on screen from frame one and readable in a single glance.
2. Every card can be read out loud before it disappears, at a calm pace.
3. Each card holds one idea. Split anything with a comma or “but” in the middle.
4. Subtitles and cards are not saying the same thing at the same moment.
5. The reveal frame is clear of text.
6. Nothing important sits on the right edge or bottom where the app overlay goes.
7. The Reel makes sense muted. No long stretch without a visual or text cue.

The rule

On-screen text is a budget, not a decoration. Spend it on the hook line, on labels the picture cannot give, and on making the video work muted. Stop spending it the moment the viewer has to choose between reading and watching, because they will usually choose scrolling.

---

You can feel when a Reel has too much text, but it is hard to see in your own edit, because you already know what every card says. Blossom reads the video the way a stranger would. It scores your hook 1 to 10 and explains why it holds or loses a muted viewer, maps where attention is likely to drop with timestamps, and points to the one change with the most impact, so you know whether the problem is the text, the opening, or the pacing. Get your first breakdown, or browse the short answers in the FAQ.