Four numbers appear on the home page. This page defines each one precisely enough that someone inside the company could recompute it and get the same answer, and precisely enough that you can tell what it does not say.
None of them is typed into the page. All four are read from the production
database by a single query behind GET /api/v1/public/stats, cached for ten
minutes, and re-fetched by the site every fifteen. A number the query cannot
compute is rendered as nothing at all — the page shows three figures, or two,
or none, rather than a stale one carried forward.
Creators on PostEngage
Counts: distinct workspaces with at least one Instagram connection that has been connected and not since disconnected, where the workspace itself is active.
Excludes: accounts that signed up and never connected Instagram, connections that were removed, and suspended workspaces.
Does not mean: monthly active users, paying customers, or accounts sending traffic today. Some of them connected an account months ago and have not logged in since. We report the number we can define, not the flattering one next to it.
Replies sent this year
Counts: the sum of actions actually dispatched, across every automation run since 1 January of the current year. Both templated and AI-written replies count, and a run that sent two actions counts two.
Excludes: anything the gate refused, anything a guard held, test runs, and shadow runs — the v1-versus-v2 comparison traffic that is processed with dispatch structurally impossible.
Rounded down to one decimal place at the nearest hundred thousand. The precise figure is always higher than the one shown.
Median time to reply
Measures: the recorded duration of every run that sent at least one action — from the webhook arriving at our servers to Instagram accepting the reply. It covers ingestion, matching the automation, generation where the reply is AI-written, all ten policy checks, and the Graph API call.
Does not include the time Instagram itself takes to deliver the webhook to us, which we cannot see, or to surface the reply publicly, which we cannot see either. The number is what we are responsible for, not what a stopwatch held by the commenter would show.
A true median, computed in the database with percentile_cont(0.5) rather
than derived from a stored average. A mean here would be dominated by the small
number of actions that wait behind a rate budget — a queued reply that goes out
forty minutes later is a correct outcome and a terrible average.
Split, because it matters: a templated reply is faster than a generated one, and the figure shown is the median across both. If you only use templates, yours will be lower.
Held back on purpose
Counts: runs that ended blocked, as a share of runs that reached a
send decision at all — that is, blocked divided by sent plus partial
plus blocked, over the same window.
Excludes from the denominator every run where nothing was watching
(no_match) or a condition simply did not apply, because neither was an
attempt to send.
The largest contributors, in order, are: the creator was already replying in that conversation; Instagram's reply window had closed; the recipient had already been answered by that automation; and quiet hours.
Does not mean that share of replies failed. Almost every one of these is the system working: sending would have duplicated a human, broken a platform rule, or woken someone at 3am. Each one is visible in the Blocked view with the check that stopped it.
Includes holds you later approved by hand.
Comparisons with other products
Where we compare PostEngage to another tool, the claim comes from that company's public pricing or documentation page on a stated date, and the date is on the comparison. Pricing pages change; we re-check monthly and correct what has moved.
We do not test competitors' products to make performance claims about them, and we do not publish a comparison a competitor could not verify from their own site.
Testimonials and screenshots
There are currently no testimonials on this site. We have customers who would give one, and no process yet for collecting consent in writing, so the section is not there rather than filled with something approximate. When it returns, every quote will be from a named customer who agreed to be quoted, unpaid, cut for length but never rewritten.
Product screenshots come from the real interface, with data anonymised where a third party's handle would otherwise appear. Where a screen is a mock-up rather than a capture, it is labelled.
Numbers you will not find here
- Average revenue, sales, or follower growth attributed to using PostEngage. We cannot see your sales, and any figure we published would be someone else's arithmetic with our logo on it.
- Response-rate uplift, conversion uplift, or hours saved. We would have to know what you did before, and we do not.
- Accuracy percentages for the AI replies. There is no benchmark for "sounds like you" that we would trust, so we describe the guardrails instead of scoring the output.
Corrections
If a number here looks wrong, write to support@postengage.ai and say which one. We will recompute it, publish the correction, and change the date.
No corrections have been issued since this page was published.