Rate limits and retries: how to avoid double posting with an API

By Ranjit Bhinge, founder of Blur Studio · Updated · 9 min read

Short answer: A double post happens when a program asks a posting service to publish, never hears back, and asks again, but the first request had worked. Blotato, a software subscription that writes, schedules and posts social media content, says to check the post's status before you ever retry a create call, and says it offers no idempotency key (a unique tag that lets a server ignore a repeat of the same request). Its documented limit for creating posts is 30 requests a minute. So: save every ID, check status, wait at least 10 seconds between checks, cap your attempts.

Ranjit Bhinge

The scary failure in automation is not the post that fails. It is the post that worked twice.

I'm Ranjit, and I run Blur Studio: 15+ years building software, 120 apps taken live, and clients who have gone on to raise $10M+. I build AI systems and processes that help businesses do a lot more with a lot less time. I also work differently from most of what else is out there, more on that further down this page. Say hello.

An API is the set of web addresses a program calls to do things in another service. When your automation, or an AI agent, publishes through Blotato's API, it is sending a request over the internet and waiting for an answer. Networks drop answers. A request can succeed while the reply never arrives. That one fact causes most double posts, and the fix is a habit, not a clever setting.

This page is written for people who are not developers. It uses plain words first, then a code sample and a test you can run.

This page may contain affiliate links from which Blur Studio earns commissions.

Why double posts happen

  1. Lost reply, blind retry. Your tool asks Blotato to publish. Blotato does it. The "done" message is lost. Your tool, seeing nothing, asks again. Two posts.
  2. Replaying a whole run. You press replay on a failed automation run, and the steps that already worked run again. Zapier's own help says a replay re-counts the steps that had already succeeded, so check whether a replay will run a post step a second time.
  3. Two triggers for one event. A feed updates twice or a spreadsheet row is saved twice, and both start a run.
  4. Two routes to the same destination. You select Instagram and Facebook in the post, and Instagram's own auto-share also copies it to Facebook. Blotato's help says not to assume a cross-post.
  5. An agent that "tries again" on its own. An AI assistant sees a timeout and calls create again because retrying looks sensible to it.

Blotato's rule: check the status before you retry

Publishing in Blotato is asynchronous. The create call returns a postSubmissionId, and you ask for the status of that ID. As of November 2026 its help pages say:

  • There is no client-supplied idempotency key, and you should not invent a header for one.
  • Its internal duplicate detection is not a guarantee for changed payloads, changed times or retries.
  • If a status call fails, retry the status call with the saved ID.
  • If the create call timed out and you have no ID, look at the dashboard, the post list and the schedules before creating again.
  • If a post failed, read its error message and fix the cause. An unchanged retry usually fails again.
  • Cap attempts or elapsed time.

Status values are in-progress, published (with the public link), scheduled and failed (with an error message). Two traps. A scheduled post is not published. And in-progress is what comes back when no record exists, so a wrong ID also returns it. Three kinds of ID exist and must not be mixed: the submission ID from create, the scheduled post ID from the schedules list, and the published post ID from the list of posts.

Blotato has no webhook (a message the service sends to your system when something happens) for published or failed posts. You poll, meaning you ask repeatedly. See webhooks and Blotato for what does exist.

Rate limits in plain words

A rate limit is a cap on how many requests you may send in a window of time. Blotato's rate limits page lists them per route, per minute, as of November 2026:

What you doRequests per minute
Create a post30
List posts60
Check a post's status60
Upload media from a link30
Get an upload address for a file120
Post a comment30
Send a message30
Create or edit a DM automation30

Two corrections to common claims. It is 30 create requests a minute, not 30 posts. And "1,800 requests an hour" appears on Blotato's home page as a stat tile with no explanation, while the documentation lists only the per-minute limits above. Thirty a minute times 60 minutes is 1,800, but that is arithmetic for one route, not a documented hourly cap. Also, MCP calls (the plug-in route AI assistants use) go through the same routes, so status polling and tool steps all count.

These are Blotato's limits, and each social platform has its own separate daily caps. Blotato lists them on its posting limits page. For the platform caps, Instagram is 50 posts a day on the safe reading, TikTok is typically around 15 a day per creator, X lists 100 post requests per 15 minutes per user, and Threads is 250 a day. Starter, at $29 a month billed monthly as of November 2026, also holds 200 queued scheduled posts. See the Blotato pricing page.

If you hit a limit you get a 429 error. Stop sending to that route, wait for the retry time if the message gives one, and slow down. Blotato's pages do not name a retry header, so read the message text.

What to do when a create call times out

What you seeWhat it meansDo this
You got an ID backThe request was acceptedSave the ID. Check status. Do not create again
Timeout, no IDUnknown: it may or may not have workedSearch the post list for your window and the schedules. Check the destination. Only then decide
Status call failed (network or 5xx)Only the question failedRetry the status call after a delay, with a limit
429Too many requests on that routeWait, slow down, then continue
Status failed with a messageThe post did not go outFix the named problem, then create once
Status in-progress for a long timeSlow, or a wrong IDConfirm the ID type, then check the dashboard before anything else
Slack or Discord says it may have been postedUnknownLook in the channel before posting again

Your own idempotency ledger

Since there is no key to lean on, make one with a spreadsheet or a database table, one row per post you want to exist. Columns: your own content ID, destination account ID, state (new, submitted, published, failed), submission ID, public link, error message and timestamp. The rules:

  1. Before create, look up the content ID and destination. If the state is anything but new or failed, stop.
  2. Write the submission ID into the row before doing anything else with it.
  3. Only the status check moves a row to published or failed.
  4. A human, not the program, resets a row to new after reading the error.

This is how a simple Google Sheets queue can work, and it is the same idea Blotato's own post-scheduler skill follows when it says never to replace a status check with another create request.

Backoff and polling

Backoff means waiting longer after each failed attempt, for example 10 seconds, then 20, then 40, so you do not hammer a service that is struggling. Blotato asks for status checks at least 10 seconds apart, so start there. Set a ceiling on tries or total time and report when you hit it. Here is a Node.js sample (Node 18 or later, which has the built-in fetch function) that does one status check loop. It stops on published, failed or scheduled, because Blotato's status page says a scheduled result confirms the schedule, not publication. It never creates a post. Keep the key in an environment variable, not in the file. Confirm field names against Blotato's Get Post Status reference before you rely on it.

// Poll a Blotato post until it is published, failed or scheduled. Never creates a post.
const BASE = "https://backend.blotato.com/v2";
const KEY = process.env.BLOTATO_API_KEY;

const sleep = (ms) => new Promise((r) => setTimeout(r, ms));

async function checkPost(id, maxTries = 6) {
  let wait = 10000; // at least 10 seconds between checks
  for (let i = 1; i <= maxTries; i++) {
    try {
      const res = await fetch(BASE + "/posts/" + encodeURIComponent(id), {
        headers: { "blotato-api-key": KEY },
      });
      if (res.status === 429) {
        wait = Math.max(wait, 30000); // slow down, then try again
      } else if (res.ok) {
        const body = await res.json();
        if (["published", "failed", "scheduled"].includes(body.status)) {
          return body; // done: report it, do not retry the create
        }
        // in-progress: keep waiting
      } else if (res.status >= 400 && res.status < 500) {
        throw new Error("Request problem " + res.status + ", fix before retrying");
      }
    } catch (err) {
      if (String(err.message).startsWith("Request problem")) throw err;
      // network error on a READ is safe to retry
    }
    await sleep(wait);
    wait = Math.min(wait * 2, 60000);
  }
  return { status: "unknown", id }; // limit hit: keep the ID, ask a human
}

The key idea is in the last line. When you give up, you keep the ID and hand the decision to a person.

Settings in n8n, Make and Zapier

  • n8n (a no-code automation tool). A node's Retry On Fail setting allows at most 5 tries with at most 5,000 milliseconds between tries, per n8n's docs, and its HTTP Request node does not read a retry-after header. For longer waits use Loop Over Items with a Wait node, or the HTTP Request node's batching with Items per Batch and Batch Interval. Do not turn Retry On Fail on for a create-post node. Set an Error workflow in Workflow Settings, starting with the Error Trigger node, so you hear about failures. See the Blotato n8n node setup.
  • Make (a visual automation tool). It has error handler modules named Rollback, Break, Resume, Commit and Ignore. Use Break and Resume only on status checks, never on a create step. Error handlers and the Router consume no credits per Make's pricing FAQ.
  • Zapier (an automation tool built around triggers and actions). Replaying a whole run re-counts steps that already worked, and re-runs them. If the Zap contains a post step, replay only the step that failed, and check first whether the post exists. Zapier has no official Blotato app, so posting goes through its Webhooks by Zapier action. See Zapier recipes for social media.

Test plan: break it on purpose

Use a private test channel (Blotato supports Slack and Discord channels as destinations). Record each run.

TestHowPass ifResult
Lost replyCut the network after sending create, then restart the flowLedger shows a row with no ID, the flow checks the post list, and creates no second post
ReplayFail a later step, then replay the runThe post step does not run twice
Bad media linkUse a link that does not loadStatus is failed with a message, the row is not marked done
Rate limitSend 70 status checks in a minuteYou see a 429 and the flow slows down, not crashes
Wrong IDCheck a made-up IDFlow does not treat in-progress as proof of a live job
Two triggersSave the same sheet row twiceOnly one post exists

The status route allows 60 requests a minute, so 70 is over the cap. Use read-only status checks for this test and never create posts in a burst to find the limit. Run it only against a test account.

Who can skip this

If you post by hand in the web app, none of this applies. If an AI assistant posts only while you watch and approve each call, you still need the status rule, but not the ledger. Everyone running unattended automations needs both.

How this page was checked

The Blotato facts come from its help pages on rate limits, error handling, async jobs and polling, post lifecycle, publishing status and identifiers, plus its pricing page. The n8n, Make and Zapier facts come from their own documentation and pricing pages. Last checked 21 November 2026, refreshed every 90 days. Blur Studio has not run Blotato itself, and the sample code is a pattern to check against Blotato's current reference, so run the test plan on a paid month you can cancel. The API is not part of the 7 day trial.

How I'd approach this

I would build the ledger before the automation. First week: day 1, make the sheet with the columns above. Day 2, connect a private Slack channel as the test destination. Day 3, build create then save the ID. Day 4, add the status loop with a cap of 6 tries. Day 5, run the test table. Day 6, add a failure alert. Day 7, run one real post with a human approving it first. Decision rule: if any test row fails, the automation does not go near a live account. For help sketching this, the free AI brainstorm call is open and /free has no-cost tools. To look at Blotato yourself, visit Blotato's site and read its rate limits help page first.

Questions people ask

Does Blotato prevent duplicate posts?

Its pages say it has internal duplicate detection, but not as a guarantee for changed payloads, times or retries, and it has no client-supplied idempotency key. Build your own ledger and check status before any retry.

What is Blotato's rate limit for posting?

Per its rate limits page as of November 2026, creating a post is 30 requests a minute, and checking a status is 60 a minute. It is a request limit, not a count of live posts.

Is there a 1,800 requests an hour limit?

Blotato's home page shows that number as a stat tile, but its documentation lists per-minute limits by route. Thirty a minute times 60 is 1,800 for the create route, which is arithmetic, not a stated hourly cap.

What should I do if a create call times out?

Do not create again. Look in the post list and schedules for a post in your time window, and check the destination. Only create again if nothing exists. See post stuck on processing or scheduled.

How often can I check a post's status?

Blotato asks for at least 10 seconds between status checks, and the route allows 60 requests a minute. Set a limit on tries and keep the ID when you stop.

Does Blotato send a webhook when a post is published?

Its documentation describes none for published or failed posts. You check status by polling. See webhooks and Blotato.

Can I test this in the free trial?

No. Blotato's pricing page says the 7 day trial excludes the API and MCP, and creating an API key ends the trial. Plan prices are on the Blotato pricing page.

More in this topic: Social media: Claude, ChatGPT and AI agent workflows (24 guides) →

Sources

Ready for the next step?

A free 15-minute AI brainstorm. We find 3 broad areas where AI could help your business.

Ranjit Bhinge You talk to me, Ranjit. I reply within a few hours.
Book a free AI brainstorm call

Not sure what you need? Tell me what is going on. I will point you the right way, even if it is not me.

Blur Studio · 30 minute call

Book a 30 minute call

You talk to me, Ranjit. If you do not need me, I will tell you.

  • We talk about your business and what is going on.
  • You leave with a clear next step, even if it is not me.
  • I reply to every message within a few hours.
Loading the calendar…