PhantomBuster alternative for web scraping
If you need page data inside your own pipeline rather than a browser-driven automation, an API-first approach can be simpler to debug and scale.
This page explains how a single URL request can return HTML, structured JSON, or Markdown, which controls matter, and how to size credits and concurrency before you commit.
One URL, several output shapes
The choice of output format is really a choice about where parsing happens. Return raw HTML when you already own a parser and want full DOM control. Use selector rules when the fields you need are known and stable, so the API returns just those values instead of a whole document. Reach for AI extraction when you know which fields matter but the page structure keeps shifting. You describe the target fields once (ai_query / ai_extract_rules) and get back structured JSON that holds up as the markup evolves, which cuts the selector-maintenance loop when you track many sites at once. Choose Markdown or plain text when the goal is readable content for indexing, knowledge bases or language-model pipelines, where DOM detail is noise rather than signal.
A practical decision check: if your bottleneck is maintaining headless browsers, proxies and brittle selectors yourself, move retrieval to a managed API. If your bottleneck is a logged-in social workflow that needs clicks and account actions, evaluate whether a page-fetch API actually covers that task before switching.
Search data and page data are different jobs
For search monitoring, a dedicated Google Search endpoint returns realtime SERP data for rank tracking, audits and competitor research without parsing a results page yourself. For general pages, combine rendering controls with extraction: enable JavaScript when content loads client-side, use premium or stealth proxies for harder targets, keep sessions where continuity matters, wait for specific elements, and capture screenshots when visual evidence helps debugging.
Per owner-confirmed terms ({{OWNER_SUCCESSFUL_REQUEST_BILLING}}), billing follows the outcome of the request: responses that come back as HTTP 200 or 404 are charged, while transient failures are not, so a retry loop does not quietly drain your budget. That distinction matters when you design retry and budget logic around flaky targets.
Developer workflow and a concrete use case
You can call the API with cURL or use official Python and Node.js SDKs, so the same request pattern works across languages. Parameter-driven control keeps behaviour explicit in code: rendering, proxies, sessions, waits and screenshots are options on the call rather than hidden browser state. That makes runs easier to reproduce, log and scale.
Illustrative scenario: a listing-aggregation pipeline needs fields from dozens of sites. Rather than operating browsers internally, the team sends each URL with extraction rules, receives structured records, and spends engineering time on its own product. The same pattern fits SEO monitoring: SERP endpoint for rankings, page extraction for content changes, Markdown for archives. This is a neutral example of the workflow, not a customer outcome.
How to size credits and concurrency
Size a plan by three numbers: monthly credits, simultaneous requests, and the credit cost of your heaviest request type. Credits are described as monthly and non-rolling, requests start at one credit, and rendering or premium proxies raise the per-request cost — confirm current costs at {{OWNER_CREDIT_COSTS}}. Concurrency is simply how many requests may run at once, so a high-volume batch job needs a higher tier even if its credit total looks modest.
Use the table below as a sizing checklist, then confirm the current commercial terms at {{OWNER_PLAN_PRICES_CREDITS_CONCURRENCY}} and {{OWNER_FREE_CREDITS_AND_CANCELLATION}} before choosing. Higher tiers add priority support, a dedicated account manager and team management with custom quotas — confirm inclusion at {{OWNER_SUPPORT_AND_ACCOUNT_TERMS}}.
| Decision | Choose this when | Watch out for |
|---|---|---|
| Raw HTML | You have a parser and need full control | Markup changes break your parser |
| CSS/XPath rules | Fields are stable and known in advance | Selectors need maintenance per site |
| AI extraction | Fields are known but markup is unstable | Validate JSON against a schema |
| Markdown/text | Feeding search, docs or AI workflows | You lose DOM detail |
| Google Search API | You need SERP/rank data | It does not replace page extraction |
General guidance; commercial terms are owner-confirmed variables, not asserted prices.
| Plan tier | Credits | Concurrency | Support/team features |
|---|---|---|---|
| Hobby | {{OWNER_HOBBY_CREDITS}} | {{OWNER_HOBBY_CONCURRENCY}} | {{OWNER_HOBBY_SUPPORT}} |
| Freelance | {{OWNER_FREELANCE_CREDITS}} | {{OWNER_FREELANCE_CONCURRENCY}} | {{OWNER_FREELANCE_SUPPORT}} |
| Startup | {{OWNER_STARTUP_CREDITS}} | {{OWNER_STARTUP_CONCURRENCY}} | {{OWNER_STARTUP_SUPPORT}} |
| Business / Business + | {{OWNER_BUSINESS_CREDITS}} | {{OWNER_BUSINESS_CONCURRENCY}} | {{OWNER_BUSINESS_SUPPORT}} |
| Enterprise tiers | {{OWNER_ENTERPRISE_CREDITS}} | {{OWNER_ENTERPRISE_CONCURRENCY}} | {{OWNER_ENTERPRISE_SUPPORT}} |
Values are owner-confirmed variables: {{OWNER_PLAN_PRICES_CREDITS_CONCURRENCY}} and {{OWNER_SUPPORT_AND_ACCOUNT_TERMS}}; do not publish until confirmed.