How often should you post TikTok slideshows? A capacity method, not a rule

What do official sources say about posting frequency?

As of November 2026 we found nothing on it. TikTok's Creator Academy photo-posting article (opens in a new tab) covers how to build a post, not how many to publish. TikTok's Help Center (opens in a new tab) describes the signals behind the For You feed, such as interactions and watch time, and we found no cadence in it. Instagram's ranking explainer (opens in a new tab) of May 2023 lists signals for Feed and Explore, again without a number of posts.

That means every figure you read, whether "three a day" or "once a week", is somebody's practice or opinion. It might suit their account. It is not a platform rule, and this page does not offer one either.

Why is frequency a capacity question?

Because the number you can keep up is set by the time a finished post takes. Writing, choosing pictures, reviewing, exporting, uploading and writing a caption all count, and so do the minutes you spend answering comments and reading results. In our view a rhythm you keep for months tells you more than a burst you drop in week three, because you can only learn from posts you actually make.

How do you work out a sustainable cadence?

1. H: the hours a week you can honestly give to this.
2. R: the minutes a week that do not produce posts, such as replies, reading results and planning.
3. T: the minutes one finished post takes. Time three real posts from first idea to upload and average them. Do not guess.
4. Posts per week = (H × 60 − R) ÷ T, rounded down.

The table shows four fictional cases. The numbers are examples, not targets.

Cases B and C have the same hours and a different T. Making posts in a batch can lower T, because setup happens once. Whether it does for you is a measurement, not an assumption. The batch workflow shows one way to time it. To spread the posts over the days you choose and download a calendar file, use the posting schedule planner, which plans days and not times of day.

Where does generation time fit into T?

Generating the text is one step among several. Fyslide's own tests on 4 November 2026 timed ten slideshows from one topic at 88.0 and 101.6 seconds with a local Ollama model (llama3.1 8B), 207.4 and 259.2 seconds with Claude Code, and 434.2 and 443.8 seconds with Codex. That was one Apple M5 Pro Mac, six slides per post and local images, and the generation speed page lists the setup. Those numbers do not include the time you spend reading ten slideshows and fixing them. That review time belongs in T.

Do plan limits tell you how often to post?

No. A limit is a ceiling set by a price list. We found no source that ties any of these numbers to results. The units differ (slideshows generated, TikTok posts, scheduled posts), so the rows are not comparable and are not a ranking.

A plan that allows 100 slideshows a month is not telling you to post three a day. It tells you what the plan includes at that price.

How do you test whether a different cadence helps?

Change one thing at a time and keep the rest steady, so you can tell what moved.

1. Run a block at your current cadence for several weeks and log every post.
2. Run the next block at a different cadence, with similar topics and formats.
3. Log the same numbers at the same post age (how to set up the sheet).
4. Compare the medians of each block, not the best post of each.
5. Write down what else changed, such as a trend or a holiday week.

Two blocks of four weeks cannot rule out luck or timing. Treat a gap as a lead to check again, as what makes slideshows perform explains. If views are low whatever you do, work through the diagnostic for slideshows not getting views before you change how often you post.

What should you settle before you raise your cadence?

• I timed three real posts and know T.
• T includes review, export, upload and the caption.
• I counted the time for replies and results (R).
• The new cadence still fits in H with some room to spare.
• I will keep logging, so I can see whether the change did anything.