Changelog
Everything we ship lands here.
The comparison pages now show their working
- Every /vs page and the GTmetrix alternatives round-up were re-reviewed against each vendor’s own site and documentation, and re-dated to August 2026
- We stopped printing competitor prices. They move, several vendors block automated re-checks, and a stale number on a comparison page is the one error a reader can’t catch — so we compare capabilities and link you to each vendor’s own pricing page instead
- The tables gained a legend: ✅ and ❌ mean the vendor publishes it (or publishes that they don’t), and a dash means we could not confirm it from their own pages, so we make no claim
- The comparison rows now cover the monitoring stack that actually shipped — alert channels beyond email, percent-change and rolling-baseline conditions, TLS-certificate and domain-registration expiry alerts, real-user monitoring, and performance budgets that fail a CI build from the CLI
- Two of our own claims did not survive the review and were removed: a free-test-allowance multiple we could no longer verify, and “no SEO auditing” for a competitor that publishes a single-page SEO checker
- Plan cards now show the performance-budget allowance and say plainly that real-user monitoring is a paid-plan feature
A GitHub Action for performance budgets — built, not published yet
- A first-party RankVitals GitHub Action now exists: one workflow step starts the test, waits for it, enforces your budgets and fails the job on a breach — no GitHub App to install and no extra repository permissions, because a red check is only a non-zero exit code
- Scores, Core Web Vitals, a verdict per budget and a link to the full report are written to the GitHub job summary, so the numbers sit on the run page instead of scrolled off the end of a log
- A budget on a metric the run never measured counts as a breach rather than a silent pass — a gate you cannot evaluate must never report green
- The commit SHA is the default idempotency key, so re-running a workflow replays the stored result instead of spending another test credit
- It is not on the GitHub Marketplace yet, so there is no uses: line to copy — the CI docs say exactly that, and the CLI one-liner and the curl recipe remain the supported ways to gate a build until it lands
Per-engine AI crawler checks, and an AI-readiness score that explains itself
- The audit now reports, engine by engine, whether GPTBot, OAI-SearchBot, ClaudeBot, PerplexityBot and Google-Extended can reach your site — blocked everywhere is a warning, blocked from some pages is a notice
- noai / noimageai directives in the X-Robots-Tag header or the meta robots tag are picked up too, so an inherited CDN rule stops being invisible
- The AI readiness card on a report expands into a per-check pass/fail breakdown, so the percentage says what produced it instead of being an opaque number
- Because the new check joins the AI-readiness score, AI readiness percentages shift for sites that block an AI crawler — the SEO score is unaffected, and a site that blocks none of them scores exactly as before
- The free AI crawler access checker and the audit engine now share one robots.txt parser, so the tool and your report can no longer disagree about the same file
Every crawl now tells you what changed
- SEO reports are diffed against your previous crawl: new, fixed and still-broken issues are separated, so you can see progress instead of the same wall of red every week
- Issues are grouped into families (indexing, content, links, structured data, AI readiness), each with fix guidance written for that specific check
- Long URL lists inside an issue are paginated rather than dumping every affected page at once
- SEO score history is kept and charted on the report, alongside the existing performance history
- All four Lighthouse category scores — Performance, Accessibility, Best Practices and SEO — are stored and shown, not just Performance
Six new SEO and AI-readiness checks
- External links are now checked for 404s, not just internal ones
- Redirect loops are detected and reported separately from long redirect chains
- Temporary redirects being used as permanent ones — a 302 that never changes — are flagged
- hreflang is validated properly: reciprocal tags, x-default presence, and self-reference
- The audit engine now runs 47 checks in total, and every page on this site derives that number from the engine itself, so it can no longer go stale
Alerts you can tune, on the channels you use
- Microsoft Teams and PagerDuty joined email, Slack, Discord and generic webhooks as alert destinations
- A test-fire button on every alert sends a sample notification, so you find out a webhook is wrong before an incident does
- Alert rules can trigger on percent change or on a rolling baseline, not only a fixed threshold — catch "20% slower than usual" without having to guess a number
- A new "alert after N consecutive failures" setting stops one flaky run from paging you
- If a team runs out of credits, monitoring now pauses visibly — with a reason on the monitors page and a single email — instead of windows quietly doing nothing
Two-factor authentication, and a dashboard that keeps up
- TOTP two-factor authentication — enrol any authenticator app from your account settings
- A test history page with search, pagination and delete, so old runs are findable instead of endless scrolling
- A light / dark / system theme toggle that remembers your choice
- A responsive dashboard nav with active states and loading skeletons on every sub-route
- Proper modals, toasts and confirmation dialogs throughout — no more browser confirm() popups
- New About and Contact pages, with a spam-protected contact form
Tests that survive a bad day
- Transient test failures are retried automatically, and credits are refunded when a run fails for infrastructure reasons rather than something on your site
- Results are saved even when a screenshot or HAR upload fails — you get the report instead of losing the whole run
- Artifact uploads retry, orphaned files are cleaned up, and the report shows when artifacts finished uploading
- A scheduling bug that could leave monitored tests unclaimed has been fixed
The MCP server can run your monitoring, not just read it
- Eight new MCP tools — list monitored pages, create/list/enable/disable/delete monitors, and create/update/delete alert rules — bringing the total to 15
- You can now ask your AI assistant to "monitor my pricing page hourly and alert me if LCP goes 20% above baseline", and it can set that up end to end
- Every tool carries read-only and destructive annotations, so a well-behaved client asks before it deletes anything
- A published server manifest and a server README, ready for the MCP registries
Analytics only after you say yes
- Google Analytics loads only after you accept the cookie banner, and only on public marketing pages — never inside the dashboard
- Our self-hosted, cookieless analytics is unchanged; it never identified you in the first place
SEO scores stay meaningful on large crawls
- SEO and AI Readiness scores now cap how much each severity bucket can deduct (critical 40, warning 30, notice 10), so a large crawl no longer bottoms out at 0
- A site with 50 minor notices scores 90 instead of 50, and ten critical issues score 60 instead of a flat 0 — badly broken sites are finally distinguishable from each other
- Scores already stored on past reports are unchanged; only new tests use the capped scale, so an alert threshold set on the SEO score may be worth revisiting
Rewritten privacy policy
- Full sub-processor table with each provider’s purpose and processing location
- Exact retention windows per plan, matching what the cleanup job actually enforces
- New sections on real-user monitoring, MCP/API data access, and AI fix plans
- Honest analytics disclosure: self-hosted cookieless Umami plus Google Analytics on public marketing pages — and still no advertising or retargeting pixels
- A dedicated privacy contact address, and privacy links from the MCP docs and server README
Security and speed hardening
- Tests and monitors can no longer be pointed at internal or private network addresses, closing a request-forgery route
- Per-tool rate limiting on the MCP server, so one runaway assistant cannot burn through your credits in a loop
- Faster speed-test and free-tool pages, and a lower server response time across the site
- Behind the scenes: credentials removed from client code, in-memory caches bounded, and stricter typing on the scoring engine
Annual billing — two months free
- Every paid plan can now be billed yearly for the price of ten months
- A monthly / yearly toggle on the pricing page shows the saving before you commit
- Switching between monthly and yearly refills credits correctly instead of skipping a cycle
41 SEO checks, free tools hub, comparisons, and a blog
- Ten new SEO/AEO checks (favicon, Twitter cards, soft-404 probe, FAQ-schema opportunity, content freshness, and more) — 41 total
- Eight free tools: meta-tag checker, robots.txt tester, Core Web Vitals checker, structured-data validator, sitemap checker, redirect checker, llms.txt generator, broken-link checker
- Comparison pages for GTmetrix, PageSpeed Insights, Pingdom, WebPageTest, and DebugBear
- Platform speed guides expanded to nine platforms with real Lighthouse measurements
- Embeddable grade badges for your README
Teams, bulk testing, white-label PDF, GitHub App
- Team invites with roles, a team switcher, and member management
- Bulk testing: paste a URL list, get a live results grid with CSV export
- White-label branded PDF reports (Growth and Agency plans)
- GitHub App connect flow — no token pasting, auto-refreshing access
- Customer portal for plan switches, cancellation, and invoices
Code-aware AI fix plans + Create PR
- Connect a GitHub repository and fix plans cite your real files with ready-to-apply patches
- One-click pull requests with the fixes applied (opt-in write access)
- Content-matched diff application that skips anything that no longer applies
Compare runs, AI provider switching, deep links
- Side-by-side comparison of any two test runs with deltas
- Admin-selectable AI providers including Ollama Cloud models
- Deep-linkable report tabs and re-run from the report toolbar
Advanced test options, waterfall v2, filmstrip v2
- HTTP auth, custom cookies/headers, URL blocking, and throttle presets on every test
- Waterfall with DOMContentLoaded/onLoad markers, filters, and request detail
- Filmstrip with per-frame timings, FCP/LCP badges, and keyboard scrubbing
RankVitals launches
- Lighthouse speed tests with waterfalls, filmstrips, and letter grades
- Site-wide SEO + AI-readiness audits
- Scheduled monitoring with email and webhook alerts
- The first performance tool with a native MCP server for AI assistants
- REST API with credit-based freemium plans