How to choose a Lighthouse/Core Web Vitals monitoring tool in 2026
Before you evaluate any tool, get clear on what you actually need. Paying for enterprise RUM when all they needed was scheduled Lighthouse on 8pages.
Lab vs field is the first decision. Lab tests (scheduled Lighthouse runs) catch regressions the moment they're deployed - deterministic, repeatable, and fast to act on. Field data (CrUX or RUM) shows real-user outcomes by device and network, which is what Google actually uses for ranking signals. Ideally you want both.
Diagnostics depth separates tools when you need to fix something fast. Request waterfalls, filmstrip/video capture, INP attribution, and third-party request breakdowns tell you why a metric regressed, not just that it did. If your tool only shows a score trend line, you'll spend extra time reproducing issues locally.
Budgets and alerting close the loop. Being able to set thresholds on specific metrics, LCP over 2.5s, TBT over 200ms, overall score below 80 and route alerts to Slack, email, webhooks, or whatever you're using is what turns monitoring into an actual workflow rather than a dashboard you check occasionally or even forget.
Coverage and control means: how many locations, what devices and throttling profiles, can you script multi-step journeys, do you have a CI/API integration, and how long is data retained? These parameters directly affect how much a plan costs and whether it fits your architecture.
Pricing model fit is often the deciding factor. Per-run quotas, per-URL limits, per-seat pricing, and free tiers interact with your cadence and page count in ways that aren't obvious until you do the math. A tool that looks affordable at 5 pages and hourly monitoring can become expensive fast at 50 pages.
Security and governance: if you're handling client data or operating in a regulated environment, self-hosting or open-source options matter. White-label reporting matters too if you're an agency packaging these results for clients.
I built Vigilant with that last point as a first-class requirement, scheduled Lighthouse, self-hostable under AGPL, with white-label client pages and branded PDF reports built in. It fits alongside or instead of the tools below depending on your priorities.
SpeedCurve
SpeedCurve is what you get when a team has been thinking seriously about web performance monitoring for over a decade. It combines synthetic testing, Lighthouse scoring, and its own RUM product (LUX) into one platform.
Lab and field coverage is genuinely strong. Synthetic tests run from global locations with configurable devices and network profiles, producing Lighthouse scores and CWV metrics. LUX gives you real-user CWV at p75 with device and network segmentation, so you can see if your LCP regression only affects mobile users on 4G.
Diagnostics are mature. Waterfalls, filmstrips, SPA support, custom User Timings, and third-party request tracking are all there. The coaching UI surfaces actionable recommendations tied to specific audit failures, which is useful for teams building a performance culture rather than just reacting to alerts.
Alerting and budgets cover the essentials: score and metric thresholds, baseline comparisons, email and webhook alerts (Slack included via webhook). It's not the flashiest alerting setup, but it works reliably.
Pricing is the main friction point. SpeedCurve doesn't publish exact dollar figures publicly, you size your plan using an in-app calculator based on synthetic check volume and RUM pageview count. For teams that want predictable monthly costs, the lack of a public price list requires a conversation before you can budget.
Best for: Mid-market and enterprise teams investing in performance as a practice, where the combined lab+field view and coaching UI justify the premium.
If you only need scheduled Lighthouse with agency-ready client pages and predictable pricing - without RUM - Vigilant is simpler and the cost is easier to forecast.
Calibre
Calibre takes a clean, developer-friendly approach to the same problem. You get scheduled Lighthouse, CrUX data, and first-party RUM in one place with transparent, published pricing.
The data coverage story is solid. Lighthouse handles your lab baseline. CrUX pulls Google's field dataset for trend context. Calibre RUM adds first-party telemetry so you can see real-user CWV alongside your synthetic results.
Diagnostics are Lighthouse-standard plus request detail views and run-to-run comparison. It's not as deep as WebPageTest for frame-by-frame analysis, but for most teams diagnosing regressions it's sufficient.
Alerting works through Slack and Zapier, with a CLI and API for CI integration. The Zapier connector makes routing alerts to less common channels straightforward without writing custom webhook handlers.
Pricing Tiers are sized by monthly synthetic test count and seats, check the current Calibre site for exact figures since they adjust periodically. The 15-day trial covers full functionality. One thing to note: if you exhaust your monthly test quota, monitoring pauses rather than overage billing. That's a trade-off worth understanding before you set your cadence.
Best for: Product and SEO teams that want a clean all-in-one CWV stack with clear costs and good developer ergonomics.
Agencies that need white-label reporting and a broader site health suite - SSL expiry, DNS changes, broken link scanning - alongside Lighthouse may find Vigilant a better organizational fit.
DebugBear
DebugBear is built for developers who want to understand why a metric changed, not just that it did. The experiment feature, where you can compare two versions of a page side by side with real Lighthouse runs, is genuinely useful during optimization work.
Lab and field coverage follows a similar pattern to Calibre: strong Lighthouse-based lab monitoring with optional RUM and CrUX overlays for field validation. The CrUX integration surfaces Google's data directly in the metric trend view.
Diagnostics are where DebugBear earns its reputation. Request waterfall breakdowns, granular metric attribution, and the experiments panel make it a real debugging tool rather than just a monitoring dashboard. If your LCP regressed and you want to know whether it was a new render-blocking script or a CDN config change, DebugBear gives you the data to answer that quickly.
Budgets and alerts are straightforward to configure. Slack, Teams, webhooks, and email are supported. The alert setup is clear enough that you can have meaningful thresholds running within an hour of onboarding.
Pricing: Startup around $99/mo, Team around $249/mo. There's a 14-day trial. For a single developer or very small team, the entry price is higher than lighter alternatives, so right-sizing your cadence and page count before committing matters.
Best for: Developer and SEO teams who want deep, actionable diagnostic detail and controlled experiments for ongoing Lighthouse regression work.
If your priority shifts to client reporting and unified website health checks beyond performance, Vigilant's white-label dashboards and broader monitor suite can complement DebugBear nicely.
WebPageTest Pro
WebPageTest is the reference implementation for web performance diagnostics. The Pro tier, now operated by Catchpoint, adds scheduling, budgets, and monitoring on top of the diagnostic engine the community has trusted for years.
Diagnostics are unmatched. Waterfalls with request-level timing, filmstrip captures, video comparison, visual diffs between runs - WPT produces more signal per test than any other tool on this list. For root cause analysis of complex LCP or CLS regressions, it's the first place I go.
Lab and field: WPT Pro is primarily a lab platform. The Expert tier via Catchpoint adds RUM starting at 10M pageviews/month alongside 30k synthetic runs/month - that's a significant jump from the Professional tier. Most teams using WPT Pro are using it for lab.
Pricing is public and starts accessibly. The free Starter tier gives 150 runs/month with 60-day data retention - good for ad-hoc investigation. Professional from $180/year adds scheduling, API access, and 13-month retention. Expert from $999/month is where Catchpoint RUM enters the picture.
Best for: Teams that need definitive lab diagnostics over time - and a clear growth path to enterprise RUM if the business case develops.
For agencies who need shareable, branded client-facing reporting alongside WPT's deep diagnostics, Vigilant can run as the always-on client dashboard while WPT handles deep-dive investigations.
GTmetrix
GTmetrix is the entry point most developers have already used for ad-hoc testing. The monitored slots feature extends that familiar interface into basic scheduled monitoring.
Data coverage is lab-only. GTmetrix runs Lighthouse-powered reports, captures CWV metrics, and generates waterfall and filmstrip views. CrUX integration isn't a focus, particularly on lower plans.
Diagnostics are standard: waterfall, filmstrip, video, audit recommendations, and historical trend graphs per monitored slot. Not as deep as WPT or DebugBear, but recognizable and usable.
Alerting is email-based on most plans. Integration with external channels requires the higher tiers or manual webhook configuration.
Pricing starts low around $5.99/month (billed annually) for a single monitored slot on the Solo plan. Higher tiers add more slots, longer retention, priority testing, and API access. The free plan for ad-hoc tests has tightened over time, with more features gated to paid plans.
Best for: Individuals and small teams who want the most affordable path to scheduled Lighthouse monitoring with a familiar UI.
If you outgrow basic email alerts and need multi-channel notifications, team collaboration, or agency-grade reporting, it's worth evaluating tools built specifically for ongoing monitoring workflows.
sitespeed.io (open source)
sitespeed.io is what you reach for when you want full control and are comfortable running your own infrastructure. It's a mature OSS toolkit that runs real browsers via Browsertime, integrates Lighthouse, captures CWV, and stores results in Graphite or InfluxDB for Grafana dashboards.
The data model is entirely lab-based. You run it on a schedule (cron, CI, or a dedicated runner), and results flow into your chosen time-series store. There's no built-in CrUX or RUM - pairing with a separate field data source requires additional wiring.
Diagnostics are rich if you invest in the setup. HAR files, video captures, timeline traces, and Lighthouse JSON artifacts are all available. Grafana dashboards give you metric trends over time with the flexibility to query and visualize however you want.
Alerting requires assembly: Grafana alerting rules, CI build gates using the budget feature, or custom scripts posting to Slack. It works, but it's not turnkey.
Cost is effectively infrastructure plus engineer time. At high page counts and frequent cadence, the per-run economics are hard to beat. For a team already running a dedicated monitoring server, adding sitespeed.io is relatively low overhead.
Best for: DevOps and performance engineers comfortable with Docker and Grafana who want maximum control, zero SaaS lock-in, and efficient cost at scale.
If you want self-hosted without the DIY assembly, Vigilant is AGPL and self-hostable with scheduling, alerts, and client reporting built in, less infrastructure work than a full sitespeed.io + Grafana stack.
PageVitals
PageVitals sits in a pragmatic middle ground: Lighthouse monitoring plus CrUX plus lightweight first-party RUM at a price that works for freelancers and small agencies.
Lab and field coverage is complete at its scale. Lighthouse runs give you synthetic baselines. CrUX integration surfaces Google's historical field data. Privacy-first RUM rounds out the picture with actual visitor telemetry - all three in one tool at a price tier most SaaS tools don't match.
Diagnostics are lighter than WPT or DebugBear, simplified waterfalls, Lighthouse audit listings, and historical metric graphs. Enough to identify and track regressions, less equipped for deep forensic analysis.
Budgets and alerts work through Slack/email and a REST API with webhook support. Configuration is straightforward and covers the common cases.
Pricing (annual billing): Starter around $12/mo, Growth around $39/mo, Pro around $79/mo - scaled by website count, synthetic run volume, and RUM pageview allocation. There's a 14-day trial. Like Calibre, exhausting your monthly quota pauses monitoring rather than triggering overages - plan your cadence accordingly.
Best for: Freelancers, SMBs, and agencies who want cost-effective combined lab+field coverage without enterprise pricing.
Agencies that also need white-label dashboards, branded PDF reports, and broader website checks - DNS, SSL, broken links, CVE tracking, can pair PageVitals with Vigilant or consolidate into Vigilant as the client-facing layer.
Vigilant
I should be transparent: I built Vigilant. So take this section with appropriate context, but I'll try to be honest about where it fits and where it doesn't.
What Vigilant monitors goes well beyond Lighthouse. Uptime, latency, SSL expiry, DNS changes, broken links, CVE advisories, healthchecks, sitemap validation, and CRON monitoring all live alongside scheduled Lighthouse runs in one platform. The idea is that agencies managing client sites need one coherent view of site health, not five different tools and dashboards.
Lighthouse monitoring in Vigilant is lab-based: scheduled runs with configurable thresholds that trigger alerts when scores or metrics cross your defined limits. There's no native CrUX or RUM integration currently, if field data is critical to your workflow, you'll want to pair Vigilant with a CrUX-aware tool or wait for that roadmap item.
Agency fit is where Vigilant is most differentiated. White-label client pages and branded PDF reports mean you can package all of this, Lighthouse scores, uptime history, SSL status, security advisories - into a professional deliverable with your agency's branding, not Vigilant's. Shareable status pages give clients visibility without raw access to your monitoring account.
AI Flows extend Vigilant beyond page speed into synthetic browser automation using plain-English step definitions. You can script a checkout flow, a login sequence, or any multi-step user journey without writing Playwright code directly. When those journeys break, you get alerted the same way you would for an uptime event.
Pricing: hosted plans start with a 14-day trial. You can check the pricing and features here.
Best for: Agencies, developers, and technical site owners who want branded client reporting, a self-hostable option, and broad website health monitoring anchored by Lighthouse, rather than a pure performance analytics platform.
Summary: which Lighthouse/Core Web Vitals tool should you choose?
Here's how I'd cut through the decision:
If you need both lab and field in one platform: Calibre, SpeedCurve, DebugBear, and PageVitals all combine Lighthouse with CrUX and/or RUM. PageVitals is most affordable; SpeedCurve is most mature; DebugBear has the best diagnostic depth. WebPageTest Pro adds RUM via Catchpoint at the Expert tier if you're already invested in that ecosystem.
If best-in-class lab diagnostics matter most: WebPageTest Pro for waterfalls and visual filmstrips, DebugBear for actionable metric attribution, sitespeed.io if you want OSS and CI pipeline integration. These are your debugging tools.
If cost is the primary constraint: GTmetrix and PageVitals have the lowest entry points for SaaS. sitespeed.io is free if you can run Docker and wire up your own alerting.
If you're an agency: Vigilant's white-label dashboards, PDF reports, and broad site health checks are purpose-built for client-facing work. Calibre also suits agencies that need CrUX and RUM data as part of their reporting.
If data sovereignty or governance is non-negotiable: sitespeed.io (MIT) or Vigilant (AGPL) keep all data on your infrastructure. WPT can be run with private agents but enterprise features come at significant cost.
Practical next steps: List your critical pages and decide on monitoring cadence. Determine whether you need CrUX, first-party RUM, or both. Multiply pages by monthly runs to size quotas correctly. Then shortlist two or three tools, run them in parallel for two weeks, and validate that budget alerts actually fire in your Slack or incident tooling - not just on a dashboard you'll forget to check.
Agencies and Laravel/PHP shops who want open-source control with white-label client reporting can start a 14-day Vigilant trial or deploy the open source edition on their own infrastructure. Either way, the goal is the same: your clients shouldn't find out their site is broken before you do.