Guide · 9 min read

By The Content Engine (organization). Written and reviewed by the Content Engine editorial team. Published 2026-10-30 · Last reviewed 2026-10-30

Instagram Carousel Hooks: 7 First-Slide Patterns From Real Covers

A carousel hook is the promise on the cover slide, and it gets about one glance to land. We hand-sorted the 24 covers with the highest likes per follower in our corpus into seven patterns. Plain labels, fill-in-the-blank definitions and versus pairs showed up most. Numerals barely did.

Where did these patterns come from?

Most "carousel hook" lists are written from memory. Ours came from looking. Our research folder holds 130 carousel posts from 18 accounts in AI, tech, business and design, collected logged out in October 2026. Of the 123 cover slides in the slide-level corpus, 109 show a public like count. We ranked those by likes divided by follower count, took the top 24 and read every cover by eye.

Three honest limits before the patterns. First, one person did the labelling in one pass. Second, the 24 come from ten accounts, and one account (simplifythefuture) supplies six of them, so a house style is doing some of the work. Third, we did not label the other 85 covers, which means we cannot tell you that these patterns over-index among winners. This is a catalogue of what winning covers look like, not proof that the pattern is the reason. If you want the reason, the follow-up is a controlled test on your own account, and the section on that is below.

It also fits what we found when we tested design variables: within a competent template, small visual choices did not separate winners from the median (see Does carousel design affect reach?). The words on the cover are a different variable, and a more promising one.

Which first-slide patterns showed up on the best covers?

Our own hand-sort, 2026-10-30. n = 24 covers, 10 accounts. The seven buckets are ours and are not mutually exclusive in real life; each cover was put in the single best-fitting bucket.

Two things surprised us. The first is how short these covers are. The wordiest in the set is a ChatGPT-prompts cover from gptprompts.ai that stacks a headline of about thirty words over a photo, and it is the exception. Most covers say a thing in three to twelve words and then leave space around it, which matches what we found earlier: cover slides carry more type than inside slides (27.7% ink share against 9.0%), but that is bigger and fewer words, not more of them.

The second is the near-absence of numerals. One headline out of 24 contains one ("1 billion monthly users"), and the claim is about ChatGPT, not about the post. The "7 ways to..." and "5 tips for..." genre that fills every hook listicle is close to missing from the covers that did best relative to their audiences. That is not proof that numbers fail. It is a reason to doubt that you need one.

How do you write a carousel hook in ten minutes?

Here is the drill we use when we plan posts. It is deliberately mechanical, because hooks go wrong when people wait for inspiration.

1. Write the outcome in plain words. Not the topic, the thing the viewer gets. "How to post less and keep reach" beats "Instagram posting frequency".
2. Cut it to eight words or fewer. This is our working rule, not a research finding. Creator Alex Kallaway calls the idea compression, not speed: fewer words, not faster delivery. We took his four tests as a checklist: is the subject instantly clear, is there a stake, is it short, can it be read only one way.
3. Dress it in three patterns. Write the same outcome as a fill-in-the-blank ("Posting frequency is ____"), as a versus pair ("3 posts a week vs 7") and as a command ("Stop posting daily"). Do not keep the one you like. Keep the one that is clearest in a thumbnail.
4. Remove every adjective. "Powerful", "ultimate" and "simple" carry nothing a viewer can check.
5. Check the thumbnail. Shrink the cover to the width of a grid tile, about 110 pixels. If you cannot read the words, they are too small or too many.
6. Do the slide 2 test. Cover the first slide and read the second alone. If it begins "And", "But" or "So", rewrite it.

Step 6 matters more than it looks. Adam Mosseri said in a Reel, as reported by CC Startup on 21 October 2024, that when someone sees a carousel but does not swipe, Instagram will often show it again starting from the second slide. We could not find the original Reel, so treat the wording as second hand. What it implies is easy to act on: some of your viewers meet slide 2 first. In our corpus operators mostly do not plan for this. Slide 2 was darker than the cover in 52% of 117 decks and otherwise indistinguishable from other inside slides. That measures what people do, not what works.

How much should a hook give away?

The natural fear is that a clear hook spoils the swipe, so people write vague ones ("You will not believe what happened"). The best evidence we found says that is half right. Aubin Le Quere and Matias analysed 8,977 headline A/B tests and found a curved relationship: when the baseline headline is too vague, being more concrete raises clicks, and when headlines are already very concrete, more concreteness lowers them. In their numbers, about 8.7% of items gained from more concreteness and about 50.9% lost from it. The authors say the right amount of information maximises clicks.

Two cautions on using that. The headlines were news-style ones from one test archive (the authors warn that the audience may have preferred less concrete headlines), and a carousel cover is not a headline. But the practical reading is consistent with the covers above. "Finance is ____" tells you the subject and withholds the answer. "Economics VS Finance" names both sides and withholds the verdict. Neither says everything, and neither is vague about what the post covers.

What did our own covers get right and wrong?

We ship seven example carousels with the Content Engine, each in a different look, and we can see our own house habits in them.

What we got right: one claim per cover, large type, a swipe cue in the corner, and an outcome in plain words. What we got wrong, or at least have not tested: we lean on numerals ("1 client win = 7 posts", "4 pillars. 40 ideas. 30 posts."), which the sample above suggests is not how the best-performing covers are written. We also have no fill-in-the-blank or versus format at all. Both are cheap to add, and both are series formats, which gives a page something a reader can learn to recognise. When we add them we will test them against our own median rather than assume.

That test is simple and you can run it without any software. Pick one idea, write it as two different hooks, and publish each on a different week in the same slot. Compare each against your own median for saves and shares, as we describe in how to measure a carousel. One pair proves little. Ten pairs start to show a pattern on your page.

What should the caption do for the hook?

The cover earns the swipe. The caption's first line is the first text a viewer reads, and it does a different job: it repeats the promise in new words and adds one fact the cover did not. We call this echo, not copy. Our caption template starts with the same hook words as the cover; looking back, that wastes the line. The next post covers the whole caption, including where the call to action goes: how to write a carousel caption that gets saves.

Where do the seven patterns fail?

• Fill-in-the-blank fails when the answer is obvious. "Water is ____" has no loop to close.
• Versus fails when the two sides are not really opposed, and the verdict on slide 8 is "it depends". Put the verdict on slide 2 or 3 if you have one.
• Command or negation fails when it contradicts something the audience already believes and you cannot support it. "Stop posting daily" works when the next slides explain what to do instead.
• "Happening now" fails with nothing new inside. Do not use it for evergreen content.
• Plain labels fail without a strong image. They are the pattern most dependent on the picture.

If you are producing a month of posts at once, give each pattern a slot in the rotation instead of using one forever. We lay that out in how to plan 30 days of carousels in one afternoon.

Frequently asked questions

Sources and our data

• Aubin Le Quere and Matias, Scientific Reports, 6 January 2025: "When curiosity gaps backfire" (8,977 headline experiments; about 8.7% of items helped and about 50.9% hurt by more concreteness)
• CC Startup, 21 October 2024: reporting Adam Mosseri's remark about carousels moving to the second slide
• Our own data: cover sort of 24 slides, chosen from the 109 covers with visible like counts in our carousel corpus (130 carousel posts, 18 accounts, harvested logged out in October 2026), ranked by likes divided by followers and labelled by hand on 2026-10-30. One reader, one pass, no second rater.
• Our own data: the cover text of our seven example carousels, and the 1,064-slide design corpus of 2026-10-10 (ink share, slide 2 comparison).
• Alex Kallaway's hook framework: our notes from his YouTube transcripts, September 2026. We cite it as a checklist, not as a statistic.

Go deeper: how The Content Engine works · pricing and the 7-day refund · compare the options · glossary of Instagram terms · all guides

Keep reading

GuideHow to Write an Instagram Carousel Caption That Gets SavesWhat 130 real captions look like and a five-part caption you can test on your own page.9 min readResearchDoes Carousel Design Affect Reach? We Tested Nine VariablesNine design variables tested on 70 posts: none significant.4 min readGuideHow to Make Instagram Carousels With ClaudeClaude writes one idea per slide, code renders the slides, you check every frame.4 min read

Results vary. No promise of views, followers or income. The Content Engine is an independent product, not made or endorsed by Anthropic. Claude is named only as the tool it runs on.