apify-ai-market-research

name

apify-ai-market-research

description

Research how a company shows up in AI answer engines vs. its competitors and produce a Gauge-style "AI presence" report as a shareable Artifact. Use whenever the user gives a website URL (or company name) and wants to understand their competitive standing in AI search / GEO / AEO — their visibility in ChatGPT, Perplexity, Gemini, Copilot, and Google AI Overview, which domains AI cites, their traditional SEO strength (domain rating, keywords), market positioning, and the highest-value actions to close the gap. Triggers on "who are X's competitors", "how does X show up in AI / ChatGPT", "GEO/AEO report", "AI visibility audit", "competitive landscape", "why don't we appear in AI answers", or "run a Gauge report for [url]". Pulls live data via Apify actors and synthesizes it into a branded interactive report.

Company competitor research → AI presence report

Turn a company's website URL into a Gauge-style AI-presence report: how the brand shows up in AI answer engines versus its competitors, who AI cites, how its traditional SEO compares, where it sits on positioning, and the highest-value actions to close the visibility gap. The deliverable is a branded, shareable Artifact (interactive web page) modeled on the reference deck.

This skill orchestrates. The actual scraping is done by Apify actors — invoke the apify-ultimate-scraper skill for the mechanics (auth, CLI flags, dataset fetching). Your job is to decide what to gather, run the right actors, and synthesize the results into the report. The exact actor inputs/outputs are documented in references/actors.md — read it before running anything.

What the report contains

The report mirrors these sections (full structure + visual style in references/report-structure.md):

1. Cover — brand + category, the framing "Data we've found for [category]".
2. Brand Coverage — % of AI answers each brand appears in (visibility / share-of-voice) + sentiment, target vs. competitors.
3. Prompt Capture — the actual AI answers to category buying-intent prompts, per engine, showing where the brand does/doesn't appear.
4. Citations — the domains AI relies on to build answers, tagged by relationship (competitor / channel partner / owned).
5. Traditional SEO — domain rating & organic keyword strength vs. competitors.
6. Action Center — prioritized, high-impact recommendations to close the gap (content targets, keywords, topics).

Workflow

Step 1 — Scope

From the user's input establish (infer sensible defaults; ask only if genuinely blocked):

• Brand + website URL (required — this is the anchor).
• Category — the market label AI buyers would use (e.g. "AI automation agency", "CRM software"). Confirm this from the company's own site before anything else — never assume it from the name, a template, or a similar company. Read the homepage (WebFetch, or apify/website-content-crawler; if a JS-rendered SPA returns empty text with the cheerio crawler, use WebFetch or crawlerType: "playwright"). Getting the category wrong invalidates the whole report — competitors, queries, and positioning all derive from it, so state it back to the user and let them correct it before you spend on scans.
• Competitors — if the user named any, use them. Otherwise discover them in Step 2.
• Region/language — default US/English unless the site or user says otherwise.

State the scope back in one line before spending money on scraping.

Step 2 — Discover competitors & buying-intent prompts

Competitors: if not supplied, use apify-ultimate-scraper with apify/google-search-scraper on "best <category>", "<category> software", and "<brand> alternatives". Pull the recurring brand names. The visibility actor caps the tracked set at 5 competitors, so pick the 5 most relevant (by market presence / how often they co-occur in the SERPs) — they also become the share-of-voice denominator, so choose deliberately.

Prompts: the AI-visibility scan needs category buying-intent prompts — the questions a real buyer asks an AI when choosing a vendor. Generate 5–10 spanning the category's sub-topics (e.g. for VC software: "best venture capital CRM", "cap table management software", "LP reporting tools", "portfolio analytics for VC funds", " alternatives"). The visibility actor fans each into 3–5 variants automatically, so seed breadth matters more than count.

Confirm the competitor list and prompt set with the user before the scan — this is the main cost driver.

Step 3 — Gather data

Run these via apify-ultimate-scraper (see references/actors.md for exact inputs, outputs, and cost). Kick off the AI-visibility scan first (it's the slowest, ~3–15 min) and run the others while it works.

Report section	Actor	Notes
Brand Coverage, Prompt Capture, Citations	doesaiknow/ai-brand-visibility-tracker-chatgpt-perplexity-gemini	The core engine. One call returns visibility, sentiment, per-engine matrix, per-query gaps, and top-cited domains for the brand + all competitors. $0.30/query.
Traditional SEO	radeance/ahrefs-scraper + radeance/semrush-scraper (+ radeance/similarweb-scraper for traffic)	Run on the brand's domain + each competitor's domain. Gives domain rating, organic keywords, authority. ~$0.05–0.08/domain.
Positioning	apify/website-content-crawler	Crawl homepage + pricing/product pages (maxCrawlDepth: 1, restrict with includeUrlGlobs) for the brand + top 2–3 competitors. Feeds positioning narrative.

Ground every number in returned data. A figure in this report (a visibility %, a domain rating, a citation share) may end up in the user's board deck — if a data point is missing, mark it "Not found" rather than filling it from memory. The visibility actor is ground-truth validated for exactly this reason; preserve that integrity downstream.

Step 4 — Synthesize & build the Artifact

Read references/report-structure.md for the section-by-section structure and the Gauge visual style (palette, layout). Then:

1. Save the raw actor outputs (JSON) to a data/ folder so the numbers are traceable.
2. Load the artifact-design skill, then build the report as a self-contained HTML Artifact following that structure — charts inline (no external requests), theme-aware, Gauge's visual language.
3. Publish it with the Artifact tool and give the user the link.

Close with the source trail: which actors ran, dataset/run links (https://console.apify.com/storage/datasets/<id>), total Apify spend, and any signals that came back empty (the report's blind spots).

Notes

• The AI-visibility scan is the star. It's what makes this different from ordinary competitor research and what the reference deck is built around. Never skip it. The SEO and positioning data are supporting context.
• Cost transparency. The visibility actor is $0.30/query and the SEO actors bill per domain. Before a big run, state the estimate (queries × $0.30 + domains × ~$0.06) and let the user trim scope.
• Freshness over breadth. A rep or founder acts on this. Prefer a tight, current, correct report over a sprawling one.