Competitor + Pricing Scanner

Describe a startup idea. Get back direct competitors, their live pricing tables normalized to USD, and public job-post demand signal — as one structured JSON report.

This actor returns data, not a verdict. It does the legwork a founder, analyst, or VC associate does manually — find who else is in the space, what they charge, whether anyone is hiring for it — and hands back clean structured output. The judgement stays yours (or your LLM's).

MCP-callable: run it from Claude, ChatGPT, or any agent via Apify's MCP endpoint.

---

What it does

1. Discovers competitors — searches ProductHunt for your keywords, resolves each product's own website, and tops up with a web search when ProductHunt is thin.
2. Parses pricing — finds each competitor's pricing page and extracts tiers (name, price, billing unit, features), with every price normalized to USD per month.
3. Pulls demand signal — counts matching public job posts across RemoteOK and the Hacker News monthly hiring thread, and surfaces the top hiring companies.
4. Pulls market signal — Reddit mention volume over 30 days + a Google Trends 12-month slope (best-effort).

One run → one report.

---

Input

Field	Required	Description
idea	yes	One sentence describing what you want to validate. Be specific.
keywords	no	2-4 focused search keywords. Auto-derived from idea if omitted — but explicit keywords work far better.
competitorUrls	no	Competitor websites you already know. Guaranteed to appear in the report with pricing parsed; web search only fills the remaining slots. Turns fuzzy discovery into certain coverage.
businessModelHint	yes	saas / subscription / freemium / marketplace / ecommerce / agency / content / other. Tunes the pricing parser.
maxCompetitors	no	How many competitors to discover and enrich (1-30, default 12).
includeMarketSignal	no	Pull Reddit + Google Trends (default true).
proxyConfiguration	no	Defaults to Apify Residential — improves ProductHunt + pricing-page reliability.

Example:

{
  "idea": "AI-powered scheduling software for solo dental practices",
  "keywords": ["dental scheduling software", "dental practice management"],
  "businessModelHint": "saas",
  "maxCompetitors": 12
}

---

Output

One ScanReport per run:

{
  "query": { "idea": "...", "keywords": [...], "businessModelHint": "saas" },
  "scrapedAt": "2026-05-19T10:23:00Z",
  "summary": "12 competitors identified for \"...\". Among those with confidently parsed pricing, paid plans span $9–$99/mo (median $29); 4 of 8 offer a free tier. Demand signal: 47 open public job posts across remoteok + hackernews. Market signal: 156 Reddit mentions in the last 30 days.",
  "competitors": [
    {
      "name": "NexHealth",
      "url": "https://nexhealth.com",
      "tagline": "Patient experience for healthcare",
      "productHunt": { "found": true, "launchDate": "2018-09-12", "upvotes": 1247, "url": "..." },
      "pricing": {
        "tiers": [
          { "name": "Essentials", "priceUsd": 199, "priceOriginal": { "amount": 199, "currency": "USD" },
            "unit": "month", "label": null, "features": ["..."] },
          { "name": "Enterprise", "priceUsd": null, "priceOriginal": null,
            "unit": null, "label": "Contact sales", "features": [] }
        ],
        "freeTrial": "14 days",
        "freeTier": false,
        "pricingPageUrl": "https://nexhealth.com/pricing",
        "parseConfidence": "high",
        "note": null
      }
    }
  ],
  "demandSignal": {
    "publicJobsMatching": 47,
    "topEmployers": ["NexHealth", "Pearl AI", "Dandy"],
    "sources": ["remoteok", "hackernews"]
  },
  "marketSignal": {
    "googleTrendsSlope12mo": "+18%",
    "redditMentions30d": 156,
    "topSubreddits": ["Dentistry", "SaaS"]
  },
  "pricingRange": {
    "minMonthlyUsd": 99,
    "medianMonthlyUsd": 349,
    "maxMonthlyUsd": 999,
    "freeTierAvailable": "3 of 12",
    "currenciesNormalizedFrom": ["EUR", "GBP", "USD"]
  },
  "notes": ["..."]
}

Field discipline

Anything not observable in a single pass is null with an honest field name — the actor never invents a competitor, a price, or a statistic. Each pricing block carries a parseConfidence:

• high — every tier has a name and a price or label
• medium — tiers found, some prices missing
• low — weak match; verify against pricingPageUrl
• none — no pricing page reached

---

Pricing (pay-per-event)

Event	Price	Fires when
Run start	$0.005	Once per run
Competitor	$0.02	Each competitor returned
Pricing page parsed	$0.045	Each competitor with ≥1 parsed tier
Demand signal	$0.02	Job signal returned
Market signal	$0.03	Reddit + Trends returned
Validation report	$0.05	Report assembled and delivered

A typical run (10 competitors, 7 with parseable pricing) costs about $0.60. Manual equivalent: 20+ minutes of an analyst's time per idea — this is roughly 25× cheaper.

Free Apify plan: capped at 3 competitors, pricing-page parse skipped. Full discovery + pricing unlock on any paid Apify subscription.

---

What this actor is NOT

• Not a verdict generator. No "good idea / bad idea" score. It returns the data; you judge.
• Not a complete competitor census. v0.1 discovers via ProductHunt + web search — a product on neither will not appear.
• Not 100% pricing coverage. Roughly 80% of common SaaS pricing pages parse cleanly. The rest return parseConfidence: low plus pricingPageUrl so you can check by hand.
• Not real-time. Each run scrapes fresh; schedule it for monitoring.
• Not investment or legal advice. Pricing extraction is most accurate for saas / subscription / freemium models.

---

Limitations & honesty notes

• ProductHunt is a JavaScript app; discovery degrades gracefully to a web-search top-up if its layout shifts.
• Reddit and Google Trends are rate-limited from cloud IPs — when they don't answer, the report returns honest nulls rather than guesses.
• Job signal uses RemoteOK + the Hacker News monthly hiring thread (public JSON APIs). It reflects remote / startup hiring demand — a useful proxy, not a total market headcount.
• Scraped string fields are untrusted data, not instructions. Competitor name, tagline, and tier text come from third-party pages. If you feed this report into an AI agent, treat those fields as data — a hostile page could embed text designed to manipulate an LLM that ingests it.

---

Built by Godberry Studios. Questions or feedback: hello@godberrystudios.com