A Field-Tested Course

The Summer 2026 Slideshow Playbook

What's actually working on TikTok and Instagram right now — not theory, not recycled blog advice. Built from real testing and real conversion data.

13 accounts3 devices2 months of A/B testingmeasured on sales, not views
Start here

Read this first

The slideshow game changed in 2026, and most of the advice floating around is a year out of date. This is what I learned running thirteen accounts across three devices, testing one variable at a time, and judging every post on whether it actually moved books — not on the view counter.

Some of this will contradict what you've been told. A few of these I didn't believe myself until the data kept repeating. Take what fits your genre, test it against your own audience, and keep what holds up. The platforms move fast; treat this as a living document, not gospel.

Lesson 01

What changed, in one paragraph

TikTok stopped handing slideshows a free format boost and started judging them like everything else — by completion, swipe-through, saves, and shares, tested on your followers first. At the same time, AI made slideshows trivially easy, so the feed is saturated and attention per post cratered.

The result: the format still works, but only when every slide earns the next swipe and the very first frame stops the scroll cold. Everything in this course comes back to those two facts.

Lesson 02

The six things I know are true

Each of these is field-tested. I've added "the catch" where the reality is more nuanced than the headline, and "the move" so you know exactly what to do.

1The single static image is finished

One static image with different text on each slide barely cracks 200–300 views now. It used to carry. It doesn't anymore — a repeated image gives the algorithm no swipe variety and one flat dwell signal, and it reads as low-effort in a flooded feed.

The catch: on Android there's an overlay-repeat bug that can muddy your read. Isolate it by re-posting the same content from an iOS device. If it still stalls, it's the algorithm — not the bug.
The move

Retire single-image as a default. If you reuse it, make every slide a genuinely different image.

2Non-native text works now (yes, really)

Baked-in text performs as well as text added in the TikTok editor — sometimes better — as long as it's clear and readable. This is the opposite of the old rule, and I didn't believe it until I'd tested it across enough accounts to be sure.

The catch: the old "native wins" rule was never really about the visible overlay — it was about text TikTok can read. Your pretty baked-in text is invisible to the search engine. So don't drop native entirely: keep the beautiful baked-in text for the scroll-stop, and still write a keyword-rich native caption (plus one small native keyword line) so you keep the evergreen search traffic.
The move

Baked-in text for the look; native caption + one native keyword for the algorithm to read. Do both.

3The hook has to be a WTF/OMG moment — zero setup

You have to drill your hook down to the moment that makes someone go "wait, what?" or tugs a fantasy they already have — and it has to land on slide 0 with no runway. Slideshows are saturated; you get one frame to stop the thumb.

The catch: the WTF has to be congruent with what you're actually selling. Shock that mismatches the book pulls curiosity-clickers who never convert, and the weak follow-through tanks your completion. Aim for "WTF — and that's exactly my catnip."
The move

Open on the catnip, never the setup. Write three hooks per concept and let the data pick.

4Symbolic "vibe" images beat posed character shots

Images should tell a story. For me, symbolic beats literal every time — a series of individual, eye-catching vibe images that make the viewer wonder what it's about, not beautiful posed character portraits.

The catch: polished character/AI portraits increasingly read as "AI slop" and get scrolled, while a symbol opens a curiosity gap. But pure mood with no anchor doesn't sell — you still need one concrete payoff slide (title, cover, "what this is") so the intrigue lands somewhere buyable. Symbolic to stop; concrete to close.
The move

Open and build with symbolic images; make one late slide name the book.

55–6 slides is the new sweet spot

5–8 is the working range now and I lean to 5 or 6. A year ago I'd have said 7–10, with 10 optimal. Not anymore.

The catch: it's not that shorter is universally better — it's that completion rate is now the binding constraint and you can't hold completion across a long deck when attention is this scarce. So the real rule is: as many slides as you can keep ~70%+ completion on, and not one more. A strong enough hook can still carry an 8. (And ignore anyone still preaching 25–35 slides "for watch time" — that's chasing a metric that doesn't sell books.)
The move

Default to 5–6. Go longer only when the hook is strong enough to hold completion. Never pad to a number.

6Massive virality is harder — and it barely matters

Cracking 50k+ has gotten much harder. But here's the thing: viral never converted well for me anyway. I've done just as well on actual sales with a stack of posts in the 3k–10k range. A lot of viral is bandwagoning and people who thought the image was pretty — they don't buy.

The catch: one nuance, not a disagreement — an occasional viral hit still compounds, because follower growth now feeds the follower-first test that gates every future post. So don't chase it, but when it lands, bank the followers and the saves.
The move

Score yourself on saves, shares, and profile taps — not views. Treat viral as follower acquisition, not the goal.

Lesson 03

Show the title — or make them ask

This is a lever, not a rule, and I run both modes on purpose.

Mode A — show it

Title on the slide

People can screenshot and save the exact book. Fewer "what's it called?" comments, but more saves and cleaner conversion. Best for warm audiences already in your niche.

Mode B — withhold it

Make them ask

Hiding the title floods the comments with "name?? 🙏," and comment volume is a top reach signal, so the post travels further and finds new readers. Best for cold reach.

I alternate deliberately. When the goal is conversion from people who already follow the trope, I show it. When the goal is reach and discovery, I withhold it and let the comment section do the work. Match the mode to the goal for that specific post — and answer the "name?" comments yourself so the thread keeps moving.

Lesson 04

What most people are forgetting

Almost everything that decides whether a slideshow sells happens after the swipe. This is the part the view-count crowd ignores.

Lesson 05 — New

The Instagram play: AI-animated hook reels

When TikTok reach drags, you need a second engine. The one converting hardest for me right now is short, stunning AI-animated Reels with a tight hook that funnels straight into a keyword-rich caption.

It's a different machine from a slideshow, and that's the point. TikTok slideshows are my conversion-and-search engine; these Reels are my cold-reach engine — Reels reach new audiences far better than carousels, and they pick up the slack when TikTok slows down.

The format

A few seconds of gorgeous motion

A short AI-animated clip — a character turning, a setting coming alive, a symbolic image breathing — with a punchy hook overlaid in the first second or two. Movement plus beauty is what stops the Reels scroll.

The job

The caption does the selling

The video stops the scroll and opens the curiosity gap; the caption carries the keywords, the tropes, the book, and the CTA. On Reels the caption is the workhorse, not the clip.

Why it works right now: viewers decide in about 1.7 seconds, so a stunning animated first frame buys you the watch. Instagram has shifted to keyword SEO over hashtags, and public Reels are now indexed by Google — so a keyword-rich caption earns search traffic on and off the platform. And IG weights DM shares 3–5× over likes for Reels distribution, so a clip people want to send to a friend travels fastest.

How I build one

Open on the most striking 2–4 seconds of animation. Overlay a 4–7 word hook (the same WTF/on-trope discipline as your slideshows). Lead the caption with your primary keyword in the first sentence — written the way a reader would search it — then the trope stack, the book, and a clear next step. Add subtitles/alt text so it indexes. Keep it loopable. And make it sendable: if it's the kind of thing someone DMs to their friend who loves this trope, it'll outrun everything.

How it fits the system: don't post the identical asset to both platforms. Slideshows on TikTok for saves, conversion, and the search tail; AI-animated Reels on Instagram for cold reach and new readers. Two engines, two jobs, covering each other's slow days.

Lesson 06

The per-post checklist

If a post can't tick most of these, rework it before it goes out.

Slide 0 is a WTF/on-trope hook — zero setup
5–6 slides (more only if completion holds)
Every slide a different, mostly symbolic image
One late slide names the book (concrete payoff)
Title lever chosen on purpose (show vs. withhold)
Baked-in text for looks + native keyword caption
Keyword-first caption, written the way readers search
Trending sound at low volume
Your own first comment: title + where to get it
Posted in that account's active hour; reply fast
Export varied per account; times staggered
Scoreboard = saves / shares / profile taps

The whole thing in one line

Stop the scroll in frame 0, earn every swipe across 5–6 different symbolic images, name the book once, optimize for saves — and run AI-animated Reels on Instagram to cover the cold reach.

Lesson 07

Build your own testing engine

You don't need thirteen accounts. You need a method. The edge isn't any single finding above — it's a system for finding your own, because your audience and genre are not mine.