Microsoft Copilot Responses Scraper — AI Search, GEO & Citation Tracking

Collect Microsoft Copilot responses, citation URLs, inline references, comparison tables, and product recommendations in a structured dataset. Turn the questions your customers ask into evidence for Generative Engine Optimization (GEO), answer engine optimization (AEO), competitor research, and AI search visibility reporting.

Enter your questions, choose a country and language, and optionally add brands to track. No Microsoft account, Copilot subscription, or external API key is required. An Apify account is needed to run the Actor.

What can you do with Microsoft Copilot response data?

• Monitor brand visibility in AI search. Check whether your brand appears in answers to high-intent customer questions.
• Research competitors. Compare brand mentions across the same set of questions and identify which alternatives receive attention.
• Discover cited websites. See the publishers, reviews, comparison pages, and domains supporting an answer.
• Plan GEO and AEO content. Find questions, topics, and sources worth investigating in your content strategy.
• Track product discovery. Review products named in shopping answers and product details when provided.
• Create agency reports. Deliver answer snapshots, citation evidence, and brand mention counts to clients.
• Monitor changes over time. Schedule repeated runs and compare timestamped datasets in your preferred analytics tool.

This Actor is designed for SEO and GEO agencies, content marketers, e-commerce teams, SaaS businesses, brand researchers, and developers building AI visibility dashboards.

Quick start

1. Add one or more questions, one per line.
2. Choose your preferred country and language.
3. Optionally enter your brand and competitor names.
4. Start the Actor and review Overview, Citations, Products, Brand visibility, or Full response.
5. Export the dataset or connect it to your workflow.

Example input

{
  "queries": [
    "What are the best CRM platforms for small businesses?",
    "What is the best iPhone cover to buy in New York?"
  ],
  "market": "en-US",
  "brands": ["HubSpot", "Salesforce", "OtterBox"]
}

Input	Required	Description
queries	Yes	1–1,000 questions, each up to 2,000 characters. Exact duplicate questions are collected once per run.
market	No	Preferred country and language. Default: en-US.
brands	No	Up to 50 brand or competitor names for literal mention tracking.

Available markets: United States, United Kingdom, Canada, Australia, Germany, France, Italy, Spain, Brazil, Japan, and India. Include a city or region in your question for local research. Market preference does not guarantee a particular answer language or an exact physical location.

Structured Microsoft Copilot output

Each unique question produces one dataset record. The output keeps answer content, citations, products, and measurement fields together so you can compare questions without joining separate datasets. Records can finish in a different order from your input.

Field	What you receive
query, market	Your question and requested audience
status, responseComplete	Whether the response is complete, partial, or unavailable
text	Answer content in Markdown, including lists and tables
rawResponse	The same unrewritten answer content as text; citation metadata is supplied separately
tables	Comparison tables with headers and rows
sources	Cited URLs, titles, domains, and available snippets
inlineCitations	Source references connected to their surrounding answer text
products	Available product cards and explicitly labelled product mentions
recommendations	Named list entries and section headings for further analysis; these may include features and categories
links	Other links appearing in the answer
researchSources	Additional research sources, kept separate from cited sources
relatedQuestions	Follow-up questions, when supplied
brandMentions	Literal mention counts, first character position, and short context snippets
sourceDomains	Unique cited domains
citationCount, inlineCitationCount, productCount, wordCount	Convenient summary counts
additionalSections	Additional answer sections, when present
collectedAt, durationMs, requestId	Timestamp, elapsed processing time, and stable question identifier
warnings, error	Availability or completeness notes

Example output

The following shortened example is illustrative. Actual answers, citations, and products vary by question and collection time.

{
  "query": "What are the best CRM platforms for small businesses?",
  "market": "en-US",
  "status": "success",
  "responseComplete": true,
  "text": "Compare HubSpot, Zoho CRM, and Pipedrive against your budget and sales workflow.",
  "rawResponse": "Compare HubSpot, Zoho CRM, and Pipedrive against your budget and sales workflow.",
  "tables": [],
  "sources": [
    {
      "id": 1,
      "url": "https://www.forbes.com/advisor/business/software/best-crm-software/",
      "title": "Best CRM Software",
      "domain": "www.forbes.com",
      "section": "citations",
      "snippet": null
    }
  ],
  "inlineCitations": [
    {
      "id": 1,
      "url": "https://www.forbes.com/advisor/business/software/best-crm-software/",
      "title": "Best CRM Software",
      "textAnchor": "Compare HubSpot, Zoho CRM, and Pipedrive against your budget and sales workflow."
    }
  ],
  "products": [],
  "brandMentions": [
    {
      "brand": "HubSpot",
      "mentioned": true,
      "mentionCount": 1,
      "firstPosition": 8,
      "snippets": ["Compare HubSpot, Zoho CRM, and Pipedrive against your budget and sales workflow."]
    }
  ],
  "citationCount": 1,
  "productCount": 0,
  "warnings": [],
  "error": null
}

Product records include name, position, evidenceType, and available url, imageUrl, priceText, currency, seller, rating, and description. productCard indicates structured product details; answerMention identifies a named product in the answer. An answer mention is not a verified merchant listing. Missing details remain null; no prices or ratings are invented. Narrative product mentions may not all appear in products; the full answer remains available in text.

Microsoft Copilot GEO reporting workflows

n8n

Use the Apify integration to run a saved question set, retrieve dataset items, and append the results to Google Sheets, Airtable, or a database. Filter on status and brandMentions before creating a visibility report.

Make

Connect an Apify run to your reporting scenario. Send citation domains to a spreadsheet and route questions mentioning competitors into your content research backlog.

Zapier

Trigger a workflow after a completed Apify run, then create spreadsheet rows or notifications for the questions and brands you track.

APIs and dashboards

Use the Apify API to start runs and retrieve datasets. JSON preserves nested citations, tables, and product details. CSV and Excel work well for Overview; dedicated citation and product views provide convenient rows for spreadsheet analysis. Schedule runs in Apify to create daily, weekly, or monthly snapshots.

Frequently asked questions

Does this require a Microsoft login?

No. End users do not need to authenticate with Microsoft or provide an external platform key.

Can I track competitor mentions?

Yes. Add brand names to brands. Counts are case-insensitive literal matches with word boundaries. They measure mentions, not sentiment, recommendation strength, market share, or ranking.

Are all research sources citations?

No. sources and inlineCitations describe citations attached to the answer. Additional research URLs are in researchSources and should not be counted as citation wins.

Will every question return products or tables?

No. These fields depend on the response. Empty arrays mean no matching items were available. Prices and other product details reflect the answer snapshot and may change.

Can answers change between runs?

Yes. AI answers and source selections can vary with time, region, and service availability. Repeated observations are more useful than treating a single answer as a permanent ranking. AI-generated statements may also be inaccurate; keep citations with your reports.

What happens when a response is unavailable?

The question remains in the dataset with an explicit status and error. Partial responses are labelled separately. If every question is unavailable, the run is marked failed and its dataset remains available for inspection. Filter to status: "success" when building reports that require complete responses.

Is this an official Microsoft product?

No. This is an independent Lofomachines Actor and is not affiliated with or endorsed by Microsoft.

More Lofomachines Actors for SEO and competitive research

• GEO Audit — AI Search Optimization Checker — assess website readiness for AI search visibility.
• Google Search Results Scraper — compare traditional search visibility with AI answer visibility.
• Website URL Extractor — build website content inventories for audits and research.
• Google News Scraper — follow coverage of brands, products, and competitors.
• Shopify Competitor Product Intelligence — extend product research with competitor catalog data.

Explore the Lofomachines Actor collection for more research and automation tools.