How to make an Instagram carousel that gets saves

Where do you start — the idea or the design?

The idea, and specifically the answer to "why would anyone open this twice?". Saves come from residual value, so a carousel with nothing left over after one read cannot earn them no matter how it looks.

The most reliable sources of save-worthy ideas are unglamorous: a question you get asked repeatedly, a mistake you watch people make, a set of numbers nobody can memorise, or a process you have run enough times to have opinions about. All four have the property of being needed again later.

What does not work as a starting point is a template. Opening a design tool and looking for something to fill it with produces decks that look finished and say nothing — the format is satisfied, the reason to save it is missing.

How do you write the slides?

Write the hook first, the close second, and the middle last. The hook determines whether the rest is seen and the close determines whether anything happens afterwards, so writing them while you still have energy is not an accident of process — it is the whole allocation of effort.

• Hook (slide 1): one specific promise in six to twelve words. See how to write carousel hooks that stop the scroll.
• Slide 2: treat it as a second cover. Instagram will sometimes show it to people who did not swipe, so it must stand alone.
• Body (slides 3 to N−1): one idea per slide, each with its own payoff. Cut any slide that only sets up the next one.
• Close (final slide): name the moment they will need this again — "save this for the next time you plan a week of posts" — rather than a generic "save for later".
• Then reread the whole deck and delete the weakest slide. There is almost always one, and removing it improves completion more than anything you could add.

What dimensions and settings should you export at?

1080 pixels wide, every slide the same shape, exported as PNG. Square is 1080×1080 and portrait is 1080×1350; portrait takes more feed space, square reposts more easily and suits charts better.

Two things to get right at export. First, the first slide sets the aspect ratio for the entire carousel — mismatched slides get cropped, not padded. Second, keep cover text inside the middle ~74% of a square slide, because Instagram’s 3:4 profile grid removes about 25% of a square image’s width, 135 pixels from each side.

Do not export wider than 1080. Instagram downscales anything larger and re-encodes it, so oversized exports commonly look worse than correctly sized ones. Full detail in Instagram carousel sizes and specs.

What goes in the caption, and what should stay in the slides?

Reference material belongs in the slides; context and voice belong in the caption. Someone who saves your post re-opens it to look at the slides, and a caption is much harder to scan than a slide.

Captions run to 2,200 characters, but Instagram truncates the preview at roughly 125 characters behind "…more", so the opening line has to earn the expansion on its own. Lead with the sharpest version of the promise, then give the context underneath.

On hashtags: Instagram permits 30, which is not a reason to use 30. Four to six specific, genuinely relevant tags do more than a wall of broad ones, and a caption stuffed with tags reads as spam to the humans you are trying to convert.

Why should you write alt text for every slide?

Because it is the only meaningful text-search surface on an Instagram post and because a significant number of your followers cannot see the images. Both reasons are good; the first is the one that gets people to actually do it.

Write one descriptive sentence per slide covering what is shown and what it means — "bar chart showing carousel saves at nine times single-image saves" rather than "chart". Around 125 characters is a sensible ceiling since screen readers often truncate near there; the widely quoted 100-character limit has no official source behind it.

This is the step everyone skips, which is precisely why doing it is cheap differentiation. It takes about a minute for a whole deck.

How do you know whether it worked?

Look at saves and sends per reach rather than at likes, and look at them a week later rather than an hour later. Saved carousels accumulate slowly and a post that looks mediocre on day one often looks different by day seven.

Sends are worth watching most closely. In January 2025 Mosseri named watch time, likes and sends as Instagram’s three most important ranking signals, and noted sends matter slightly more for reaching people who do not already follow you. Saves, despite what most marketing content claims, were not in that top three.

The other number that matters is how far people got. If reach is fine but saves are low, the content was consumable in one pass — it needed more density, not a better hook. If almost nobody swiped past slide one, the hook was the problem and the rest of the deck never got a fair test.

Can you shorten this workflow?

Most of it, yes — the writing and layout are the slow parts and both are largely mechanical once the idea exists. What cannot be shortened is choosing something worth saving, which is the step that actually determines the outcome.

Carousel Studio compresses the middle: give it a topic and it researches it with live web search, writes a 3–8 slide deck with the hook and close in place, pulls real photos, applies your brand colours and headline font, and exports 1080×1080 PNG slides plus an MP4 for any chart, a 9:16 reel, a caption and per-slide alt text. It runs on your own computer and your own Gemini or OpenRouter key, so there is no subscription — $19 once, with a 7-day free trial in the app.

It will not tell you which idea your audience is waiting for. That judgement stays where it has always been.