The best time to post on TikTok, from the API's own data
The best time to post on TikTok, by the API's own data, is when the account's followers are most active: audience_activity, a daily metric of GET /business/get/ in TikTok's API for Business, counts the followers active in each hour, in UTC. Summed over several weeks and converted to the followers' time zone, its highest hours are the slots to post in, and each post's reach then tests the choice.
easyCoexistence connects a TikTok account through TikTok's API for Business, so Claude, ChatGPT or your own agent reads its audience_activity with get_social_insights and schedules posts into the busiest hours with publish_tiktok_post and scheduled_at, up to 90 days ahead, with each connection monitored 24/7.
How far back start_date reaches on /business/get/, which returns one set of metrics, audience_activity included, per UTC day.
Each post's publish time as a UTC Unix timestamp, returned in real time by /business/video/list/ beside the post's reach and views.
Most posts /business/video/list/ returns per page; cursor, a UTC timestamp in milliseconds, fetches posts published before it.
What sets a post's reach, per TikTok: the account's region, not the IP address of the API call that publishes it.
What does audience_activity count?
Followers, hour by hour. TikTok describes it as the hourly breakdown of followers active on TikTok during the day, in UTC+0, with TikTok Studio as its source. Each day in the response's metrics list carries its date and an audience_activity list of pairs:
- hour: the hour of that UTC day, as a string such as "0" or "23".
- count: how many followers were active on TikTok during that hour.
- Order: TikTok's own example lists the hours out of sequence, "20" before "7", so sort them as numbers before reading.
How many days of data should you read?
As many as the API allows. start_date and end_date take dates such as 2026-10-31, in UTC, and start_date can reach 60 days back; without them, the response covers the 7 days up to yesterday. A single day shows that day alone, while a peak that returns week after week is a pattern to plan around. The data also starts no earlier than the day the owner turned Analytics on in the TikTok app.
How do you turn the counts into posting times?
Add up each hour's count across the days, rank the hours, and read them in your followers' local time: a peak at hour "23" in UTC is 20:00 in São Paulo and 18:00 in New York in winter. Converting each date and hour on its own, as below, also lands each peak on the right local weekday, which turns a list of hours into slots such as Monday at 20:00:
// metrics: data.metrics from GET /business/get/ with fields ["audience_activity"]
const zone = "America/Sao_Paulo"; // where most followers live
const totals = {};
for (const day of metrics) {
for (const { hour, count } of day.audience_activity ?? []) {
const utc = new Date(day.date + "T" + hour.padStart(2, "0") + ":00:00Z");
const slot = utc.toLocaleString("en-US", { timeZone: zone, weekday: "short", hour: "2-digit", hourCycle: "h23" });
totals[slot] = (totals[slot] ?? 0) + count; // "Mon, 20": Monday, 20:00 local
}
}
const best = Object.entries(totals).sort((a, b) => b[1] - a[1]).slice(0, 5);Which time zone should the hours be read in?
The one most followers live in, and the same endpoint says which. audience_countries gives the share of followers per country or region and audience_cities the share per city, both once the account has 100 followers; where the top country spans several time zones, the cities settle it. For a following split between distant zones, rank the hours for each group on its own.
Why can the busiest follower hour mislead?
Because it describes followers only, and posts also reach people who do not follow the account. /business/video/list/ shows how far, under the video.insights scope: audience_types splits a post's viewers into new and returning, followers and non-followers, and impression_sources ranks where they found it, such as For You or Follow. When most viewers come from outside the following, treat the follower peaks as a first guess and let post results decide.
How do you check that the chosen hours work?
Compare posts. /business/video/list/ returns each post's create_time with lifetime counts such as reach and video_views, so group posts by local publish hour and compare them at the same age: the counts keep growing and arrive 24 to 48 hours late, which makes a two-day-old post and a month-old one an unfair pair. Accounts without audience_activity can start here, since these fields come with the video.list scope.
How does an AI agent schedule posts into those hours?
Through easyCoexistence's MCP server, from Claude, ChatGPT or your own agent. get_social_insights reads audience_activity for the dates in params, the agent ranks the local hours, and publish_tiktok_post takes the chosen slot as scheduled_at, ISO 8601 with its offset, from 1 minute to 90 days ahead; easyCoexistence copies the media then and publishes at that time. list_scheduled_posts shows the queue, update_scheduled_post moves a post to another peak, and TikTok's caps of six posts a minute and 15 a day per account apply as each one goes out:
get_social_insights {
"account_id": "<ACCOUNT_ID>",
"metrics": ["audience_activity", "audience_countries"],
"params": { "start_date": "2026-09-05", "end_date": "2026-10-31" }
}
publish_tiktok_post {
"account_id": "<ACCOUNT_ID>",
"video_url": "https://media.example.com/tiktok/breakfast.mp4",
"caption": "Five-minute breakfast #breakfastideas",
"scheduled_at": "2026-11-09T20:00:00-03:00"
}Common mistakes
- Shifting the whole table by one fixed offset. Daylight saving moves the offset during the year and the UTC date turns at midnight, so convert each date and hour separately before grouping by weekday.
- Grouping posts by create_time as it comes. It is a UTC Unix timestamp, so convert it to the followers' time zone before comparing publish hours.
- Counting the last two days as complete. audience_activity arrives 24 to 48 hours late, so end the date range two days back for an even sum.
- Writing a local hour with Z. 20:00 in São Paulo is 2026-11-09T20:00:00-03:00, or 23:00Z; the local hour marked Z publishes three hours early.
US$ 9 to US$ 0.75 per connection per month, falling as you add connections, with the first 3 days free.
easyCoexistence is a multichannel API and MCP server that connects AI agents such as Claude and ChatGPT to WhatsApp, Instagram, TikTok, Facebook and more. On WhatsApp it uses Coexistence: the business keeps using its number in the WhatsApp Business app while the AI works through Meta's official API.
Frequently asked questions
Is there one best time to post on TikTok for all accounts?
TikTok's API answers per account: audience_activity covers one account's own followers, so each account reads its own peak hours instead of a general rule.
Does posting from a server in another country change a post's reach?
No. TikTok's Accounts API FAQ says its publish endpoints have no IP-based restrictions and that the call's IP address does not affect a post's reach.
What if my TikTok account gets no audience_activity?
Compare your posts instead, by create_time and reach. audience_activity needs Advanced Access or a Verified Business Account with at least 100 followers, and Analytics turned on in the app.
Can Claude or ChatGPT choose the posting time by itself?
Yes, through easyCoexistence: it reads audience_activity with get_social_insights, picks the busiest local hour and passes it to publish_tiktok_post as scheduled_at, and the queue publishes then.
Keep reading
Connect your first channel in two minutes.
No platform reviews to wait for. Connect WhatsApp, Instagram, TikTok or another channel, then ask your AI to post, reply or report, and watch it work.
- 3 days free
- Cancel any time
- Official APIs only
- Monitored 24/7
Verified on