Competitor Review Gap Analyzer

Compare customer-feedback themes supported by at least two explicitly labeled brands. Findings are hypotheses to check against actual product capabilities, not verified competitor weaknesses.

First run

Preview the output: run the input below with no records, dataset IDs or app IDs. It uses labeled synthetic records with no report-event fee. After the run, open Markdown report for the readable result or JSON report for integration.

{
  "demoMode": true
}

The preview uses example.com source URLs and fictional feedback. It is a format demonstration, not customer or competitor research.

Analyze your own data

Include brand on every record and supply evidence about the same topic for at least two brands. Two brands discussing unrelated topics do not create a comparison. Include text and original sourceUrl so every comparison remains reviewable.

For an existing dataset, select it with the dataset picker in the input form. The API equivalent is below; replace the placeholder before running:

{
  "demoMode": false,
  "sourceType": "dataset",
  "datasetIds": [
    "YOUR_LABELED_COMPETITOR_REVIEW_DATASET_ID"
  ],
  "maxRecords": 1000,
  "maxInsights": 20
}

You can instead paste objects into records. Each record should contain text and its original sourceUrl; optional fields include id, platform, brand, location, rating, date, likes and title. Supplied sources override synthetic demo mode. Dataset reads do not rerun the collector. Any separate upstream collection workflow is outside this report price.

What comes back

One report row in the default dataset contains an insights array, analyzed-record count and finding count. The output links also provide REPORT.md and REPORT.json in the key-value store. Use JSON to preserve nested evidence; use the Markdown report for review.

Comparisons include per-brand sample counts, excerpts and complaint-supported hypotheses within shared topics. Unlabeled records or topics without at least two brands cannot establish a cross-brand comparison.

Record limits and sample comparison

maxRecords caps raw current records examined at 1–1,000 before deduplication and filtering. maxInsights caps findings at 1–30. A small or unmatched batch can return fewer findings, including zero. The fee is per delivered report, not per finding.

previousDatasetId remains in the input schema for compatibility. Prior-sample comparisons are implemented only in App Store Review Feature Roadmap; leave that field blank in this workflow.

Pricing

$0.25 per delivered report, covering at most 1,000 current raw records. This includes analysis of inline records, existing datasets or native Apple reviews. The actor uses one report-delivered event and no additional dataset-row event. See the Pricing tab for the current platform price.

Synthetic previews on the imported-data workflows have no report-event fee. The App Store live sample uses actual reviews and the normal $0.25 report price.

Method and recurring use

Analysis uses deterministic English phrase/rating rules and text grouping; no generative AI model is called. Scores and draft actions are research aids. Sarcasm, negation, uncommon phrasing and multilingual text can be misclassified. Inspect supporting evidence before acting. Topics can overlap, so their percentages must not be added together. Samples do not establish market-wide demand, product capabilities or future business results.

Use the Actor API or Apify MCP with the same input schema. Imported datasets are snapshots; provide fresh source data for repeated analysis. This actor does not directly scrape YouTube, Reddit or Google Maps or start third-party collectors. No external model key is required.

Local run

Requires Node.js 22 or newer. Run npm ci, then npm start. The Dockerfile uses apify/actor-node:22.