Instagram Reels Intelligence

Learn which of your Reels worked, which did not, and what to test next – from your own Instagram insights.

Two ways to feed it, both fully permitted: a CSV/JSON export of your post performance (Meta Business Suite) or the official Instagram Graph API with your own access token. It never logs in, scrapes, or touches accounts that did not authorise you.

Who is it for

Social media managers, creators and small agencies who already download Instagram insights and want a clear "what worked, what to test next" report instead of a spreadsheet.

Quick start

1. Press Start with the pre-ticked Demo mode: you get a full report on a fabricated account. Every row is labelled SYNTHETIC DEMO and demo runs are never charged.
2. For your own account, export your Reels performance from Meta Business Suite > Insights > Content (use the export/download option; exact menu names change between app versions) and paste the CSV text into Performance export (CSV text).
3. Press Start again. Provided data always wins over Demo mode, so you can leave it ticked.
4. Open Output for the structured rows, and the Key-value store for REPORT_WEEKLY_MD (weekly report) and findings.csv.

Example input

{
  "dataCsv": "Post ID,Permalink,Publish time,Post type,Description,Views,Reach,Likes,Comments,Shares,Saves\n...",
  "accountName": "My Brand",
  "timezoneOffsetMinutes": 180
}

Example output row (SAMPLE - synthetic, illustrative values; not from a real account)

{
  "recordType": "account_summary",
  "platform": "instagram",
  "account": "My Brand",
  "postsAnalysed": 48, "postsScored": 45,
  "medianValue": 5120,
  "billing": { "charged": true, "event": "account-analyzed" },
  "synthetic": false
}

Pricing

$0.25 per account analysed. The charge is made once per account, only when at least 6 posts could be scored (published at least minAgeDays ago with views or likes). Too little data = the results are still delivered, but not billed (billing.charged: false explains why). The billed summary is written last, after all detail rows. Demo runs are free.

Pay-per-event. Platform usage (compute, storage) is included in the price - there is no separate Apify usage bill for you. A negligible Actor-start event ($0.00005 per GB of memory, minimum one) also applies. You can cap spending with Apify's "Maximum charged results / cost per run" setting; the Actor stops cleanly at your limit.

What is verified, and what is not

• CSV/JSON export analysis (column matching, dates, hooks, baselines, weekly report, 14-day plan) is covered by automated tests using synthetic exports, including English, Turkish (; delimiter, Yayınlanma zamanı, 29.10.2026 14:30, 12.345 counts), Spanish and Portuguese headers. Real Meta/TikTok export files from other languages were not available, so an unusual header may need one line from you to be supported.
• Not verified with live credentials: the Instagram Graph API mode has only been tested against recorded/synthetic API-shaped responses. It has not been exercised with a real authorised account token. Use the export path for guaranteed behaviour.
• The Graph API needs your own token with the instagram_manage_insights permission for a professional account you control (Meta app review applies to other people's accounts).

Limitations

• Works on data you own or are authorised to use. It never logs in, scrapes or reads other accounts.
• Findings are correlations inside your own history, not proof of cause; confidence is shown on every pattern.
• At least 6 posts older than 3 days are needed for baselines; otherwise the Actor says so instead of scoring.
• No competitor scraping: competitor analysis only covers data you provide.
• Hook classification is rule-based and English-centred with partial ES/TR/PT.
• Retention curves, audience demographics and watch time are not part of the media insights used here.

Built by PRIVATEX. Questions or a data-format that does not parse? Open an issue on the Actor page and include the (anonymised) column headers or first lines.

What you get

• Top and weak Reels – ratio versus your own median of comparable Reels, with the comparison size. Weak posts list possible factors (correlations, labelled as such).
• Pattern table – opening-line style (question, number list, how-to, story, call-out…), caption length, numbers, hashtags, duration, weekday/hour, with sample size and confidence. Anything within ±15% of baseline is hidden as uninformative.
• Creative briefs and a 14-day test plan – built from your best topics and hook styles, with A/B hooks, a publish slot only if the data supports one, a success rule, and links to the posts they are based on.
• Weekly performance report – this week versus last week, best and weakest scored post.
• Account comparison – cadence, duration and hook mix across accounts, using only data you supply.
• JSON dataset + CSV + Markdown. No AI API: everything is deterministic.

Data sources and official API scope

• Export mode (no setup): paste the CSV; columns such as Publish time, Description, Post type, Views, Reach, Likes, Comments, Shares, Saves are matched flexibly.
• API mode: Instagram Graph API media endpoint plus media insights (views, reach, saved, shares, total_interactions). Needs a professional account and the instagram_manage_insights permission. For other people's accounts your Meta app needs App Review. Insight data can lag up to 48 hours and counts organic activity only; the Actor therefore ignores posts younger than 3 days by default.

Input

Field	Meaning
dataCsv / dataUrl / dataRows	Your performance export.
accessToken, accountId	Optional API mode (token is a secret, used only for graph.facebook.com).
reelsOnly	Default on.
additionalAccounts	More accounts as your own data, for comparison.
demoMode	Fabricated account, labelled SYNTHETIC.

API example

curl -X POST "https://api.apify.com/v2/acts/YOUR_USER~instagram-reels-intelligence/run-sync-get-dataset-items?token=$APIFY_TOKEN" \
  -H "Content-Type: application/json" -d '{"dataUrl":"https://example.com/insights.csv"}'