Instagram Reels Creative & Trend Intelligence

Analyze Instagram Reel datasets to detect breakout creatives, engagement momentum, viral velocity and recurring competitor trend signals.

Why use this Actor

Brands and agencies can extract Reels but still need repeatable evidence of which creatives are accelerating, breaking out or losing momentum. This Actor is deliberately separated from source extraction. It accepts supplied public-data rows or an Apify Dataset and turns them into recurring decision intelligence. That architecture reduces dependence on source-site markup, login flows and proxy behavior while letting upstream collection tools change without rebuilding the intelligence layer.

Key features

• Computes viralVelocityScore from plays and engagement changes
• Compares current and previous Reel snapshots
• Flags breakout, rising and declining creatives deterministically
• Avoids mandatory direct Instagram scraping and its proxy cost
• Reads inline rows or an Apify Dataset with read-only permissions.
• Compares an optional previous snapshot and emits deterministic agentAction output.
• Runs at 256 MB with no mandatory paid LLM or browser dependency.

Example

{
  "currentItems": [
    {
      "id": "r1",
      "author": "brandA",
      "caption": "Launch",
      "plays": 120000,
      "likes": 8400,
      "comments": 350,
      "shares": 900,
      "url": "https://instagram.example/r1"
    },
    {
      "id": "r2",
      "author": "brandA",
      "caption": "Tutorial",
      "plays": 85000,
      "likes": 6700,
      "comments": 240,
      "shares": 520,
      "url": "https://instagram.example/r2"
    },
    {
      "id": "r3",
      "author": "creatorB",
      "caption": "Trend",
      "plays": 310000,
      "likes": 29000,
      "comments": 1200,
      "shares": 4100,
      "url": "https://instagram.example/r3"
    }
  ],
  "previousItems": [
    {
      "id": "r1",
      "author": "brandA",
      "caption": "Launch",
      "plays": 70000,
      "likes": 5100,
      "comments": 220,
      "shares": 400,
      "url": "https://instagram.example/r1"
    }
  ]
}

The run writes one structured report to the default Dataset and stores the same object in INTELLIGENCE_REPORT. Downstream automations can route the report by agentAction, while the current source Dataset can be preserved as the baseline for the next scheduled run.

Use cases

• Creative strategy
• Brand monitoring
• Creator intelligence
• Agency reporting
• Social commerce research
• AI agents

A common workflow is: collect authorized public data, store it in an Apify Dataset, run this Actor with current and previous Dataset IDs, route the resulting action to a workflow or agent, and preserve the current Dataset for the next comparison.

Pricing

The product uses one primary pay-per-event unit: One decision-ready intelligence report written to the default dataset. The configured price is $0.008 per report before any Marketplace tiering or future approved changes. There is no separate start-fee design in the QuanMatrix product model.

Limitations

• Analyzes supplied public-data datasets; it does not bypass authentication or private-data controls.

• Signals are deterministic heuristics based on observed snapshot fields.

• Input field aliases are normalized conservatively.

• The Actor does not bypass authentication, private-account controls, robots restrictions, or source-platform access rules.

• Decision scores are prioritization aids based on observed fields and snapshot differences, not guarantees about future outcomes.

Output and automation

Every result contains ok, mode and agentAction, plus source counts and product-specific score fields. The output schema exposes both the default Dataset and the stored intelligence report so the Actor can be used from Tasks, schedules, API calls, agents and other Apify workflows.

Gen2 decision intelligence

This Actor preserves its original analysis and adds a decision layer with baseline awareness, regression detection, confidence, GO/WARN/BLOCK executive output, and an optional economic-impact estimate. Economic estimates are produced only when the user supplies valuePerImpactUnitUsd; the result states the calculation basis instead of inventing monetary value.