How Many Slides Should a TikTok Carousel Have? (Data Says 5-10)

Five to ten slides. That's the answer, and it's one of the more consistent findings in short-form content: TikTok allows up to 35 images per carousel, but performance data and creator testing keep converging on the 5-10 band, with 7 as the most dependable default — hook, five value slides, CTA.

The interesting part is why, because the mechanics explain almost everything else about how to structure a carousel: why too short underperforms just like too long, why different content types have different sweet spots inside the band, and why slides 1 and last deserve more effort than everything in between combined.

(New to the format? The full build guide is how to make TikTok slideshows; the data on why carousels are worth making at all is in the photo-mode algorithm post.)

The completion math behind 5-10

TikTok's recommendation system ranks posts largely on the signals viewers generate, and a carousel generates two that matter most: swipes and completion.

Every swipe is a deliberate act of continued interest — passive video watching has no equivalent. More slides means more potential swipe signals per viewer. That's the force pushing slide counts up.

But every slide is also a chance to lose the viewer, and completion — finishing the post — is one of the strongest signals a carousel produces. A viewer who bails on slide 4 of 12 tells the algorithm the post didn't hold; a viewer who finishes 7 of 7 tells it the opposite. That's the force pushing counts down.

The 5-10 band is where those forces balance:

• Under ~5 slides: the post finishes before the viewer has meaningfully interacted. Two or three swipes is thin engagement data, and the post can't build the "kept choosing to continue" pattern the system rewards. It also just feels thin — a 3-slide listicle reads as an unfinished thought.
• 5-10 slides: 10-20 seconds of swiping. Most viewers who start, finish; each of them hands you 4-9 intentional interactions plus a completion. Maximum signal density.
• Past 10-12: completion collapses. TikTok's own engagement pitch for photo posts — 1.9x likes, 2.9x comments vs video — is built on posts people actually finish; a 20-slide informational post is a completion-rate donor. The exception is narrative content with a genuine pull to the ending (storytime dumps, "swipe to see how it ended"), which is the only genre that sustains 12-15+.

One more practical reason: effort-per-slide is zero-sum. The founder making a 15-slide post spreads the same thinking across twice the surface as the one making 7. Constraint concentrates quality.

Slide count by content type

Inside the 5-10 band, content types have their own sweet spots. From creator testing patterns (see e.g. Swypr's roundup of slide-count guidance) and our own runs across app-marketing accounts:

Content type	Sweet spot	Why
Meme / relatable dump	4-6	Jokes thin out fast; leave them wanting one more
Before / after reveal	5-6	The contrast is the content; padding dilutes the reveal
Feature spotlight / mini-tutorial	5-7	One step per slide; more than ~6 steps means the tutorial is too big for one post
Listicle / ranking	7-10	One item per slide plus hook and CTA; the number in the hook sets the count
Educational deep-dive	8-10	Enough room to teach; caps where completion starts sliding
Story / narrative dump	10-15	The one genre where a strong ending sustains length

Notice the pattern: the more structural the content (numbered items, steps), the longer you can run, because each slide re-earns the swipe with a visible progress marker ("3/7"). The more emotional the content (memes, reveals), the shorter — you're spending a fixed budget of surprise.

Treat the table as a starting prior, then test against your own account: run the same content archetype at two counts (say, listicles at 7 and at 10) for a couple of weeks each and compare median reach and saves — not best-case views, which are noise. Most accounts find their audience has a real preference, and it's discoverable within about twenty posts. Slide count is one of the few carousel variables you can actually A/B cleanly, because everything else (niche, hook style, posting time) can be held constant.

If you're marketing an app, the 20 slideshow ideas for apps post maps every idea to its archetype and count, and the carousel vs video framework covers when a topic shouldn't be a carousel at all.

Slide 1 has one job (and it isn't summarizing)

Slide 1 is simultaneously your hook, your thumbnail on the profile grid, and your result card in TikTok search. Its one job: make the swipe feel owed.

• Promise with a number. "6 tools that replaced my $200 stack" — the number tells the viewer exactly what finishing buys them, which is what makes completion feel achievable and worth starting.
• Readable at thumbnail size. If the hook isn't legible when the slide is an inch tall, cut words until it is.
• Open a loop; never close it. The cardinal sin is answering on slide 1. "This app saved me 10 hours a week" with the app's name visible = zero swipes owed. Same line with the name held to slide 3 = a reason to swipe.
• No logo slides, no title-card slides. Slide 1 that says your brand name is a post that dies with single-digit completion.

The last slide converts the completion

Whoever reaches your last slide is your highest-intent viewer of the day — they just chose to interact with your post seven times in a row. The last slide's job is to convert that intent into the next action:

• A comment prompt — usually the best pick, since comments are the metric carousels win biggest on ("which one did I miss?", "which type are you?").
• A save-bait summary — the whole list on one slide, built to be screenshotted and saved. Saves buy carousels their long distribution tails.
• A soft CTA — "follow for the next one" or a link-in-bio pointer. Keep it to one CTA; a last slide asking for follow + comment + save + click converts none of them.

What the last slide must never be: filler. If your list ends at slide 8, end the post at slide 9. A trailing "thanks for reading" slide exists only to be abandoned on — and that abandonment is, mechanically, a failed completion.

Three slide-count myths worth killing

"More slides = more engagement, since every swipe counts." Half true, and the half that's false is expensive. Swipes only accumulate from viewers who keep going; slides 11-20 of an informational post are mostly swiped by nobody, while their existence suppresses the completion rate of the whole post. You're not adding engagement surface, you're adding abandonment surface.

"TikTok auto-advances slides anyway, so count doesn't matter." TikTok can auto-advance a photo post like a video when the viewer doesn't swipe, but that's the low-engagement path — the viewer is passively watching, generating none of the swipe intent that gives the format its edge. Design for the swiper, not the auto-advance: if your post only works when TikTok plays it as a de facto video, it should have been shorter, or an actual video.

"The 35-slide dumps I see going viral disprove the sweet spot." The viral long dumps are almost always narrative — a story where the ending is the payoff, or an aesthetic photo dump where completion isn't the point because viewers save it two slides in. They're real, and they're a different genre. For lists, tips, product content, and everything else an app marketer posts, the completion mechanics rule, and the completion mechanics say 5-10. Survivorship also does heavy lifting here: you see the one 30-slide post that broke out, not the ten thousand that died on slide 6.

Caption and music: the two inputs outside the slides

The caption is your search surface. TikTok search indexes captions heavily, and photo posts increasingly surface in search results — so write one keyword-rich line of context ("7 budgeting apps for freelancers who hate spreadsheets") plus a comment-prompt question. Don't restate slide 1; the caption is for the algorithm and the commenters, the slides are for the swiper.

Music sets the swipe tempo. One track plays across all slides, viewers control the pace, and the pairing is more of a ranking input than most people assume: a trending sound gives the post a second discovery surface via the sound page. Match energy to archetype — ambient/lo-fi for educational listicles, sped-up aesthetic tracks for dumps and memes, something with momentum for before/afters. A mismatched or missing sound reads as broken in a sound-on feed.

The 7-slide default, spelled out

When in doubt, ship this:

1. Slide 1: Hook — numbered promise, big text, loop open.
2. Slides 2-6: One idea per slide, max two short lines each, visible progress if it's a list.
3. Slide 7: Payoff + one CTA (comment prompt or save-summary).
4. Caption: one searchable line + one question.
5. Sound: rising, energy-matched.
6. Specs: 1080x1920, text out of the bottom ~250px.

Then post it daily, because slide count is a tuning parameter and volume is the strategy — carousel performance is high-variance, and the 1-in-10-to-20 winner math only works if you keep taking shots. Daily hand-built carousels is where founder discipline goes to die, which is why the sustainable version is automated: ReelFlood generates finished slideshows from an image collection — hook, slide copy, and sequencing already inside the 5-10 band this post argues for — and publishes them to your connected TikTok account on the schedule you set, via TikTok's official Content Posting API. Setup details in how to auto-post to TikTok.

The short version

5-10 slides; 7 as the default. Short enough to finish, long enough to generate real swipe engagement. Memes run 4-6, listicles 7-10, and only stories earn 12+. Slide 1 opens a loop with a number; the last slide converts your best viewers with exactly one ask; the caption feeds search and the comments; the sound feeds discovery.

And if you'd rather test the sweet spot than hand-count slides every morning, ReelFlood's free tier — 100 credits, no card — will generate and schedule your first two slideshows, and the $29 Starter plan (roughly 60 a month) runs the full two-week experiment for you.