PageSpeed & Lighthouse Audit: Bulk Core Web Vitals, No API Key

PageSpeed & Lighthouse Audit runs Google's open-source Lighthouse 13 in headless Chrome on Apify's servers and returns, for every URL, the four Lighthouse scores (performance, accessibility, best practices, SEO, each 0–100) and the lab Core Web Vitals and speed metrics: Largest Contentful Paint, Cumulative Layout Shift, Total Blocking Time, First Contentful Paint, Speed Index and Time to Interactive. Audit mobile, desktop or both, and optionally get the top failing audits with their estimated savings.

No API key is needed. Many PageSpeed tools call Google's PageSpeed Insights API, whose shared keyless quota often runs out ("Queries per day" errors). This Actor runs Lighthouse itself, so there is no Google quota. If you also want real-user field data from the Chrome UX Report (CrUX), paste your own free PageSpeed Insights API key and each URL goes to Google's API instead.

Why this Actor

• No key, no quota errors. Lighthouse runs here, in a fresh headless Chrome for every audit.
• Bulk. Paste hundreds of URLs; each URL × device is one row, ready for CSV, Excel or an integration.
• The same engine as PageSpeed Insights. Lighthouse with its standard mobile settings (phone screen, simulated slow 4G, 4× CPU slowdown) or desktop settings.
• Field data when you want it. With your own key you get CrUX p75 LCP, INP and CLS for the URL (or its whole origin when the URL has too little traffic), next to Google's lab scores.
• No audit charge for failures. Invalid URLs, DNS failures, timeouts and pages Lighthouse can't load get a row with a plain-English error and no audit charge (the run start still applies, see pricing).
• Guards against risky URL lists. Every page and every request it makes go through a filter that refuses private, loopback, link-local and cloud-metadata addresses.

Use cases

• SEO and site audits: score every template of a site (home, category, product, blog) on mobile and desktop in one run.
• Client reporting: a scheduled run over your clients' key pages, sent to Google Sheets.
• Lead generation for agencies: find the companies in a scraped list whose sites score badly on mobile.
• Before and after a release: run the same URLs before and after a change and compare LCP, CLS and TBT.
• Accessibility triage: turn on Include top failing audits to see which accessibility checks fail on each page.

How to use it

1. Add URLs (full URLs or bare domains such as example.com, which become https://example.com/).
2. Pick the Device: mobile (the default, like PageSpeed Insights' mobile tab), desktop, or both.
3. Optionally pick fewer Categories, turn on Include top failing audits, or paste your own PageSpeed Insights API key for field data.
4. Run it. Each URL × device becomes one row. The run summary (counts and average scores) is saved as OUTPUT in the key-value store.

Input example

{
  "urls": ["https://www.python.org", "https://apify.com", "example.com/pricing"],
  "strategy": "both",
  "categories": ["performance", "accessibility", "best-practices", "seo"],
  "includeAudits": true,
  "maxConcurrency": 1,
  "timeoutSecsPerUrl": 90
}

Output example (shortened; values are illustrative)

{
  "url": "https://www.example.com/",
  "inputUrl": "example.com",
  "finalUrl": "https://www.example.com/",
  "strategy": "mobile",
  "engine": "lighthouse-local",
  "lighthouseVersion": "13.5.0",
  "scores": {"performance": 75, "accessibility": 100, "bestPractices": 96, "seo": 100},
  "metrics": {"lcpMs": 2374, "fcpMs": 943, "tbtMs": 1151, "cls": 0, "speedIndexMs": 943, "ttiMs": 2374},
  "fieldData": null,
  "topOpportunities": [
    {"id": "render-blocking-insight", "title": "Render blocking requests", "category": "performance", "score": 0,
     "displayValue": "Est savings of 250 ms", "savingsMs": 250, "savingsBytes": null},
    {"id": "errors-in-console", "title": "Browser errors were logged to the console", "category": "bestPractices", "score": 0,
     "displayValue": null, "savingsMs": null, "savingsBytes": null}
  ],
  "runWarnings": [],
  "benchmarkIndex": 2373,
  "durationMs": 7337,
  "fetchedAt": "2026-10-19T12:00:00.000Z",
  "error": null
}

With your own API key, engine is "pagespeed-api" and fieldData holds the real-user numbers, for example:

"fieldData": {"scope": "url", "lcpP75Ms": 2710, "inpP75Ms": 180, "clsP75": 0.05, "fcpP75Ms": 1450, "ttfbP75Ms": 640, "category": "AVERAGE"}

All columns

• url (normalised), inputUrl (as you typed it), finalUrl (after redirects), strategy (mobile or desktop), engine (lighthouse-local or pagespeed-api), lighthouseVersion.
• scores: performance, accessibility, bestPractices, seo, as 0–100 integers (null for a category you didn't pick).
• metrics (lab, with Performance): lcpMs, fcpMs, tbtMs, cls, speedIndexMs, ttiMs.
• fieldData (only with your API key, otherwise null): scope (url, or origin when Google only has data for the whole site), lcpP75Ms, inpP75Ms, clsP75, fcpP75Ms, ttfbP75Ms, category (FAST, AVERAGE or SLOW). These are the 75th percentile of real Chrome users over the last 28 days.
• topOpportunities (only with Include top failing audits, otherwise null): up to 10 failing audits with id, title, category, score, displayValue, savingsMs, savingsBytes, biggest savings first.
• runWarnings (Lighthouse's own warnings, such as a slow CPU or a page that redirected), benchmarkIndex (Lighthouse's measure of the CPU it ran on), durationMs, fetchedAt, error (null on success).

The dataset has three views: Scores, Real-user data and Top failing audits.

error values include Invalid URL: …, Blocked: private or internal address …, DNS lookup failed: …, Timed out after 90 seconds …, Lighthouse couldn't audit the page: … (for example the page returned an error status, never painted, or has a bad certificate) and PageSpeed Insights API error ….

How much does it cost?

Pay per event: $0.02 per audited URL (one URL on one device, one output row), plus the run start. Apify platform usage is included.

Apify Store discounts: the price above is for Apify's Free plan. Paid plans pay less per audit: $0.018 on Starter, $0.016 on Scale, $0.014 on Business. The "from" price at the top of the Store page is the Business price.

The run start is charged by the platform per GB of Actor memory: $0.02 at the default 4 GB. Rows with an error (invalid or blocked URLs, DNS failures, timeouts, pages Lighthouse couldn't load, API errors) have no audit charge. The run start applies to every run, even one where every audit fails. Both devices means two audits, so two charges, per URL.

Worked examples:

Examples use the Free-plan rate. Where an example doesn't add it, each run also has its run-start charge, counted per GB of Actor memory (minimum one; four at the default 4 GB, which is $0.02).

• 100 URLs on mobile: 100 × $0.02 = $2.00 + start.
• 100 URLs on mobile and desktop: 200 × $0.02 = $4.00 + start.

If you set a maximum cost per run, the Actor stops cleanly at that limit and doesn't start audits it can't charge for. If a run is restarted (for example after a migration), it skips the URL × device pairs already in its dataset.

FAQ and limits

Why are my scores different from PageSpeed Insights?

This is lab data measured on Apify's servers, not on Google's machines: the CPU, network and location differ, and PageSpeed Insights may run a different Lighthouse version. Performance scores also vary from run to run on any machine, often by several points. Accessibility, best-practices and SEO scores are much steadier. For important pages, run the audit more than once and compare the median. benchmarkIndex and runWarnings show when the CPU was slower than Lighthouse expects.

How much memory should I give it?

The default is 4 GB because on Apify a run's CPU share grows with its memory, and Lighthouse needs about one full CPU core: in our tests a heavy page (apify.com) timed out after 90 s at 1 GB but finished in 36 s at 4 GB. Each parallel audit needs about 900 MB, and the Actor lowers Parallel audits to what fits. Running with less memory costs less to start but makes timeouts and lower performance scores more likely. Parallel audits share the CPU and lower performance scores, so keep Parallel audits at 1 when you compare performance; for accessibility, best practices and SEO only, parallel audits are fine (up to 3).

How long does it take?

Each audit loads the page in a fresh Chrome and waits for it to settle, so expect tens of seconds per URL per device on real pages; heavy pages take longer. For long lists, raise the run timeout (default 1 hour).

How do I get real-user (CrUX) data?

Create an API key in Google Cloud Console, enable the PageSpeed Insights API for its project, and paste the key into Your PageSpeed Insights API key. Each URL is then audited by Google's own PageSpeed Insights servers (their lab scores plus CrUX field data), and Google's quota for your key applies. Without a key, fieldData is always null: real-user data only comes from Google. Sites with little Chrome traffic have no CrUX data at all; the row then says so in runWarnings. If Google refuses your key, the run stops with a clear message instead of filling the dataset with errors. The key is stored encrypted by Apify and is never written to the output or logs.

Why is there no INP in the lab metrics?

Interaction to Next Paint needs a real user clicking and typing, so Lighthouse can't measure it in a page load. Total Blocking Time (tbtMs) is the lab metric that tracks it best. Real INP comes with fieldData when you use your own key.

Can it audit pages behind a login?

No. Each audit starts a fresh browser with no cookies, and there are no options for logins, custom headers or cookies. Pages that block bots or data-centre IP addresses may score oddly or fail with NO_FCP or an error status.

What does it load, and is it safe?

Each page is loaded in a real headless browser: it runs the page's JavaScript and loads its images, scripts and third-party tags, as a visitor's browser would. Requests to private, loopback, link-local and cloud-metadata addresses are refused, including redirects and subresources, so internal or localhost URLs can't be audited (they get a free Blocked row). Every audit uses a new, empty browser profile that is deleted afterwards.

Other limits

• Only http and https URLs. URLs with a user name or password aren't supported.
• Each audit gets Timeout per URL (default 90 seconds); slower audits are stopped and get a free Timed out row.
• Lighthouse 12 removed the PWA category, so it isn't offered.
• The same URL listed twice is audited once.

Related Actors

Actor	Use it to
Broken Link Checker	Crawl a whole site to find 404s and redirects, then audit its key pages here.
Bulk URL Status Checker	Check status codes, redirect chains and SSL expiry for thousands of URLs, fast and cheap.
Sitemap URL Extractor & New Page Monitor	Get every page URL of a site from its sitemaps to audit here.
Website Tech Stack Detector	See which CMS, shop platform and tags are behind a slow site.
Dataset Filter & Transform	Keep only pages under a score, flatten scores and metrics into columns, or rename fields before export.