SERP Providers and A Parser

SERP Providers and A-Parser

Rank tracking, clustering, outlines, content analysis, plagiarism checks, site: estimates and geo-grid scans all need search results. You choose where they come from in Settings → SERP & scraping.

Providers

Provider	What it's good for	Pricing
Serper.dev	Cheap, simple Google SERPs with geolocation. Best place to start	~$0.30–1 per 1,000
DataForSEO	Google + Bing SERPs; the same credential unlocks Demand, AI Overviews, LLM Mentions, brand citations and backlinks. Paste the Base64 token or login:password	pay-as-you-go
ScrapingRobot	Rank tracking and SERP analysis	free tier 5,000 scrapes/month
GoAnyAPI	Credits for domain traffic, cached SERPs with DR/traffic per result, KD, backlinks	pay-as-you-go
A-Parser	Your own self-hosted scraper: no per-request cost, you pay for proxies	license + proxies
XMLRiver	Google SERP API used by the index checker and the backlinks indexed-status check	per check
Firecrawl	Not a SERP: a scraping fallback that gets past anti-bot pages and returns clean Markdown, used when a direct fetch of a competitor page fails	per page

On the same tab, choose:

• the active SERP provider for content tools (clustering, outlines, analysis);
• the Rank Tracker provider, which can differ (e.g. something cheap for daily checks, something richer for occasional research). Empty means "use the active one";
• the Rank Tracker fallback provider for checks the first one couldn't answer (captcha, 503, "try again later").

Link-purchase tokens for Magic links are on this tab too.

A-Parser

A-Parser is a commercial scraper you run on your own machine. Connected to OpenGSC it becomes a SERP provider, a rank-tracker backend, the source for SERP Monitor, an AI Overviews source and a GEO Audit engine. Requests cost nothing per call; result quality depends on your proxy pool.

Connecting

1. In A-Parser: Settings → API: enable the API server and copy the password. Note the host:port it listens on (default 9091).
2. In OpenGSC: Settings → SERP & scraping → A-Parser: base URL (e.g. http://127.0.0.1:9091 when A-Parser runs on the same server; keep the external port closed) and the API password. Optionally a thread config name (empty = default). Press Test.
3. Or set it on the server, where it takes precedence over the UI:

   OPENGSC_APARSER_BASE_URL=http://127.0.0.1:9091
   OPENGSC_APARSER_PASSWORD=...

   An env-configured instance never fetches a URL typed into a browser, which is the safer deployment.
4. Docker: 127.0.0.1 inside the container is the container. Use the host's LAN IP or host.docker.internal.

Once connected, Research → A-Parser appears in the top menu: a console showing whether the instance is up, what it's running, which parsers this build has, whether proxies are behind it, and a test query.

If you change the API password in A-Parser, update it in OpenGSC too (or in .env), otherwise every A-Parser feature reports it as unconfigured or failing.

Rank Tracker on A-Parser

Uses SE::Google::Position, which stops at the results page where your site appears. A keyword in the top 10 costs one page, position 35 costs four, and only "not in the top 100" costs all ten.

1. Connect A-Parser (above). The instance needs SE::Google::Position and proxies Google accepts.
2. Optional sanity check on the server:

   cd /root/opengsc
   npx tsx scripts/aparser-position-probe.ts your-site.com "a keyword you rank for" gr el

   If an option reads NOT in preset, or the Position and full SE::Google numbers differ, don't switch yet.
3. Settings → SERP & scraping → Rank Tracker provider → A-Parser; optionally set a fallback.

Details worth knowing:

• Your site is matched as exact domains (site.gr and www.site.gr). A tracked apex ranking on blog.site.gr reads as "not found": a miss, never a false hit.
• Captcha solving is inherited from your SE::Google presets automatically.
• "Not found" is only stored if the parse went through all ten pages. A short parse is an error (aparser_partial_serp), so a dead proxy can't fake a rank drop.
• Local (city) keywords run as a normal SE::Google scan because Position doesn't accept geo.

GEO Audit on A-Parser

Runs GEO Audit stage 1 through FreeAI::ChatGPT on your instance instead of a paid API. Geolocation follows the proxy's exit IP, and citations come typed with snippets.

1. Make sure FreeAI::ChatGPT is in your build (the A-Parser console lists parsers).
2. SEO Tools → GEO Audit → engine: A-Parser.

Web search is forced on for every request. The report names the model the free session actually served. Stage 2 (analysis) still runs on your Utility AI task. Without it, the audit ships with the deterministic half only. Driving a free consumer service through automation is your call to make against that service's terms.

AI Overviews and SERP Monitor

• AI Overviews in AI Visibility uses DataForSEO when present, otherwise your A-Parser (SE::Google).
• SERP Monitor v1 runs only on A-Parser. See SERP Monitor.

Deep reference: RANK-TRACKER-APARSER.md, GEO-APARSER.md.