DEV Community

Mustafa Aydemir
Mustafa Aydemir

Posted on Fully Autonomous

5 Core Web Vitals Tools: One-Off Checks or Ongoing Monitoring?

Start with a one-off check when you need to investigate a slow page. Pay for ongoing monitoring when you need repeated evidence, a record of changes or someone to get notified when performance slips.

The important detail is what gets measured. A scheduled lab test, Google's historical field data and your own real-user monitoring answer different questions.

These five Core Web Vitals tools cover those different jobs. This is a documentation-based comparison, with sources checked on November 12, 2026, rather than a hands-on benchmark. I'm a co-founder of Screpy, which is included below.

Decide what evidence you need

Google's current Core Web Vitals are LCP for loading, INP for responsiveness and CLS for visual stability. The recommended good thresholds are 2.5 seconds, 200 milliseconds and 0.1 respectively, assessed at the 75th percentile and separated by device type.

Lab tests help reproduce a condition and investigate a change. Field measurements describe actual visits. Running a lab test every day doesn't turn it into field data.

Tool Main job in this comparison Important boundary
PageSpeed Insights An on-demand lab check with available CrUX field context Check whether field data covers the URL or the origin
Lighthouse Local investigation and tests you automate yourself You operate the recurring workflow
DebugBear Managed synthetic tests, CrUX and real-user investigation Choose the measurement coverage and subscription you need
SpeedCurve Synthetic/RUM monitoring with performance budgets Test volume and sampled traffic affect the quote
Screpy Scheduled lab checks for selected URLs alongside SEO work Current results, without a multi-point performance history or alerts in this view

A green lab score is useful diagnostic evidence. It doesn't certify that every visitor, page or interaction has a good experience.

1. PageSpeed Insights

PageSpeed Insights is a useful free starting point for a public URL. You get a fresh Lighthouse lab run and, where coverage exists, historical real-user information from the Chrome UX Report.

Read the two parts separately. The field data covers a trailing 28-day period, so a deployment today cannot instantly replace that entire window. Check whether the report describes the exact URL or the broader origin. If neither has enough samples, field data remains unavailable.

I'd use it to answer an initial question such as, “Is this article slow to display its main content, and does the available visitor evidence point in the same direction?” Then inspect the diagnostics rather than stopping at the overall score.

The standard web report is an on-demand check. If your real need is retained reports, automatic scheduling and notifications, plan that workflow separately. Repeated manual screenshots can document a baseline, but somebody still has to run and compare them.

2. Lighthouse

Lighthouse makes sense when you are already working on the page. It is open source and runs through Chrome DevTools, the command line or a Node module. DevTools also lets you audit pages that require authentication.

It shares the lab engine used by PageSpeed Insights, but gives you a different workflow: investigate a local change, inspect the report and repeat under comparable conditions before releasing it.

For example, you could compare the same landing page before and after changing its image delivery. Keep the page version and test conditions attached to each report. A single improved score doesn't establish a stable improvement across visitors.

You can build recurring checks around Lighthouse, including Lighthouse CI. The software is free; operating a reliable workflow still involves runner capacity, storage, configuration and someone responsible for failures.

Choose this route when your team wants control and can maintain it. Choose a hosted service when the value you need is less work managing tests, reports and notifications.

3. DebugBear

DebugBear homepage social preview

DebugBear combines scheduled synthetic tests, CrUX reporting and real-user monitoring. Its reports, comparisons and alerts are relevant when performance needs an ongoing owner rather than an occasional check.

The useful distinction is between detecting a lab regression and understanding a visitor's slow experience. Its real-user monitoring includes page-view details and investigation of slow interactions, including the elements and scripts associated with poor INP.

That can matter when the problem appears after someone opens a menu or uses a filter, rather than during the initial page load. A loading diagnostic alone may miss the interaction you care about.

Before subscribing, decide whether you need synthetic testing, RUM or both. Ask for a quote against your actual URLs, test cadence and traffic coverage, rather than assuming a headline price covers the whole setup.

The current pricing page returned a JavaScript-only shell during this research, so I couldn't verify a subscription amount. I wouldn't use an older comparison's price to make that decision.

4. SpeedCurve

SpeedCurve homepage social preview

SpeedCurve is another option for a team combining synthetic testing and real-user monitoring. Its performance-budget workflow is especially relevant if you want a threshold crossing to reach someone who can investigate it.

The budget and alert features support absolute limits and rate-of-change thresholds, with notifications through email or webhooks. The deploy API connects release activity with performance investigations.

For a frequently updated product, that gives the team a clearer handoff: review the affected metric, compare the release context and decide whether the change needs investigation. An alert establishes that a threshold was crossed; it doesn't prove which code change caused it.

The pricing page lists plans starting at $90 USD per month, with the quote sized around RUM page views and synthetic checks. It also displays a 20% annual-billing saving. Check the selected billing period and included volumes before treating the starting figure as your budget.

I'd compare it when retained monitoring and release ownership are part of the requirement, rather than paying for that workflow just to investigate one slow page.

5. Screpy

Screpy homepage social preview reading AI SEO Analyzer and Website Audit Tool

Screpy Core Web Vitals fits a different job: keeping selected public URLs under scheduled lab checks while you work on technical SEO in the same workspace.

The URL table reports FCP, LCP, CLS, TTFB and TBT. Active eligible pages become due for another check after one day, subject to availability and plan access. That isn't a guarantee of a result at the same clock time every day.

There is an important boundary: TBT is not INP, and this feature doesn't report INP. Its lab PASS badge also doesn't certify Google's field-based Core Web Vitals assessment. Missing measurements need attention even when the available subset looks good.

The PageSpeed workflow shows current results, without a multi-point performance history or performance alerts in that view. Save your baseline if you want a meaningful comparison after a change. If automatic regression notifications and historical reports are essential, compare the dedicated monitoring services above.

The Pro plan includes five projects and 25 Core Web Vitals URLs per project. Its regular annual price is $192, equivalent to $16/month paid upfront. Welcome discounts are conditional; check your offer and checkout total. The seven-day trial becomes paid unless cancelled, and subscriptions renew automatically. AI Visibility is not included in Pro.

You can also use Screpy MCP to ask an assistant about the saved homepage lab overview. That tool reads available measurements with their URL and check time; it doesn't run a new performance test, return every monitored URL or fix the website. Access and plan eligibility still apply.

Size the monitoring job before subscribing

Before comparing prices, write down the pages, conditions and frequency you actually need. Google's Core Web Vitals workflow guide is a useful reference for combining diagnosis with field evidence instead of expecting one report to answer everything.

Here's an illustrative sizing calculation, not a recommended schedule or a measured result:

6 URLs × 2 test profiles × 2 runs per day × 30 days = 720 scheduled checks.

If you repeat each combination three times, that becomes 2,160 runs. Vendors may count checks differently, so map that workload to their billing definition. For RUM, also decide what traffic to collect or sample. A URL allowance and a monthly run allowance aren't interchangeable.

Keep a small baseline record with the URL, device/profile, measurement source, timestamp and release version. When a result changes, first check that you're comparing the same conditions. Then investigate the page; don't assign the cause from the score alone.

For an isolated investigation, start with PageSpeed Insights or Lighthouse. For a maintained regression workflow, compare DebugBear and SpeedCurve against your coverage and alerting needs. For scheduled current lab checks alongside SEO work, compare Screpy with its limits in mind.

The useful purchase is the one that helps someone take the next action. More reports won't help much if nobody owns that action.

Top comments (0)