CoreDash compared with other RUM tools

I built CoreDash, so I have a stake in every answer on this page. I wrote a comparison for each tool below anyway, because clients keep asking me about them and because most of them are good. Where another tool does something better I say so, here and on the detailed pages.

First: get field data at all

Lighthouse loads your page once on a simulated phone and never clicks anything. CrUX only counts opted-in Chrome users, and only once a page is popular enough (see not enough data in Search Console). Nine of the ten tools here measure your own visitors on your own pages. Without that you find out about a bad deploy weeks later, when the 28-day CrUX number finally moves. So having field data at all matters more than which tool you pick.

If you already use one of these and it answers your questions, keep it. Switching takes time, and your team already knows its way around the tool you have. This page is for people who are still choosing, or who are starting to run into the limits of what they have.

The overview

Checked September 2026. Prices are per month in USD, billed annually where there is a choice, unless noted. Each detailed page has the sources and the fine print.

Dedicated performance tools

DebugBear

DebugBear is the only tool here with scheduled Lighthouse tests, its own RUM and CrUX history in one interface. The lab tests throttle at the system level, keep the median of several runs and handle user flows and logged-in pages. On the RUM side LCP and INP are split into subparts, TTFB into seven parts, and INP is attributed to scripts with LoAF, down to the invoker type. Their blog and tools like the INP Debugger are some of the best free web performance material out there. The main difference with CoreDash is the bundle: RUM on DebugBear starts at $99 a month and includes lab test quota you might not need, while 500K pageviews on CoreDash cost $31. CoreDash vs DebugBear

SpeedCurve

SpeedCurve has been around since 2013, and its people helped shape the standards behind Core Web Vitals. It combines mature RUM with a proper synthetic suite: scheduled tests in Chrome and Firefox, scripted journeys, and filmstrips and comparison videos that make a slow page obvious to anyone. It reports LCP and INP from Firefox and Safari as well as Chrome, and has had LoAF attribution since May 2025. Embrace bought SpeedCurve in November 2025, and signup now goes to an Embrace demo request. No MCP server, and RUM history is 3 months on plans under $500 a month. CoreDash vs SpeedCurve

RUMvision

RUMvision is the most similar to CoreDash: a Dutch, EU-hosted RUM tool built on Google's web-vitals library, and Erwin Hofman and Jordy Scholing know LoAF better than most. You get LCP subparts, INP phases, TTFB and FCP breakdowns, LoAF down to the function and 45+ dimensions. Health Checks open and close to-dos by themselves, every plan comes with a setup call and monthly consultancy, and their MCP server has 13 tools (mine has 3). We differ on price and coverage. RUMvision starts at €150 a month for 2 domains and only tracks your homepage until you add URL rules. CoreDash starts at $16 and tracks every URL from the first pageview. CoreDash vs RUMvision

Request Metrics

Request Metrics comes from the TrackJS team: bootstrapped, with twenty years of monitoring experience. Its Long Animation Frames reporting is some of the best I've seen, with a dedicated dashboard, invocation versus evaluation, per-script reports and LoAFs drawn into the waterfall. There are also weekly Lighthouse scans and a CrUX dashboard that puts your origin next to any competitor. No cookies, unlimited users on the entry plan, and a free plan. Where it falls short is history and getting data out: retention is 30 days on Essentials and 90 on Professional, there's no public data API or MCP server, and pageviews above the per-minute limit get dropped. CoreDash vs Request Metrics

Observability platforms with a RUM module

Datadog RUM

If your company already runs Datadog, its RUM module can follow a slow LCP straight into the backend trace, log or profile behind it. No Core Web Vitals tool does that, mine included. The attribution is up to date too: LCP and INP subparts, LoAF mapped back to unminified source, and browser profiling down to the function. With RUM without Limits you can ingest every session cheaply and keep full detail only where you need it. For Core Web Vitals work the downsides are per-session pricing across several SKUs, raw events kept for only 15 to 30 days, and an SDK of about 66 KB gzipped. CoreDash vs Datadog RUM

New Relic Browser

New Relic Browser captures more Core Web Vitals detail than people tend to think. The agent source shows all four LCP subparts, the three INP phases with target and load state, and the largest CLS shift. Session Replay is hooked into the Core Web Vitals tables with poor-LCP sessions first, distributed tracing takes you into the backend, and the free tier (100 GB a month and one full user) is enough for lots of small sites. It has no LoAF though, so it can't name the script behind a slow interaction. Raw browser events are kept for 8 days by default, and the MCP server has no Core Web Vitals tools, so an agent has to write NRQL. CoreDash vs New Relic Browser

Akamai mPulse

mPulse is enterprise RUM, built on boomerang.js, one of the original open-source RUM libraries. On the Akamai CDN you switch it on at the edge without touching a template, which in some organisations saves months of approvals. It's strongest on the business side, with revenue and conversion next to performance and what-if forecasting. For Core Web Vitals it records the element behind LCP, INP and CLS, but no LCP phases, no INP phases and no LoAF. There's no MCP server, no public pricing, and only a handful of product releases since 2024. CoreDash vs Akamai mPulse

RUM that comes with your platform

Vercel Speed Insights

On Vercel, Speed Insights is as easy as RUM gets. Add one component, deploy, and the data comes in. It groups dynamic routes like /blog/[slug] automatically, keeps preview and production apart, and since August 2026 there's a free tier on every plan. The vercel metrics CLI gives a coding agent decent query options. But it's made for sites hosted on Vercel, and it only measures hard navigations (in Next.js, the first page view of a session), while CoreDash also measures soft navigations in single page apps. Attribution stops at a CSS selector. Speed Insights Plus costs $10 per project plus events, on top of the Pro plan, which adds up quickly for an agency with a few dozen client sites. CoreDash vs Vercel Speed Insights

Cloudflare Web Analytics

Cloudflare Web Analytics is free on every plan, sets no cookies and is often already on. It uses a current version of Google's web-vitals library, handles soft navigations, and the Debug View lists the top 5 elements behind each metric with P50 to P99. The GraphQL API even has LCP subparts and INP phases, which is a lot for a free tool. It was built as privacy-first traffic analytics though, performance came second. No LoAF, no custom segments, no Core Web Vitals alerts, heavy sampling after 7 days, and it skips EU visitors by default when Cloudflare turns it on automatically for a free zone. CoreDash vs Cloudflare Web Analytics

Google's free data

CrUX, PageSpeed Insights and Search Console

Everyone should start with Google's own field data, and it's the first thing I look at on any site. It's free, needs no install, works for any public site (competitors too), and it's the dataset Google uses for Core Web Vitals in Search. Since 2025 it even splits LCP into subparts. But it only counts opted-in Chrome users on pages popular enough to qualify, every number is a 28-day p75, and there's no element or INP attribution. CoreDash has a CrUX cross-reference, so you can see Google's number next to your own. CoreDash vs CrUX and Search Console

Why I think CoreDash is the better choice long term

In the first month most of these tools tell you the same thing: LCP is slow on mobile. After a few months you want to know which element and which script, whether last Tuesday's deploy did it, and whether someone (or an agent) can go fix it. Those later questions are what I built CoreDash for.

Attribution you can fix things with

Every LCP comes with its four phases (TTFB, load delay, load time, render delay), the element, the element type, the image URL and its LCP priority: preloaded, fetchpriority high, not preloaded or lazy loaded. Every INP comes with its three phases, the input type, the load state and the Long Animation Frame script that held up the main thread. CLS comes with the shifting element, the cause and the timing. The Diagnosis view checks for 14 LCP anti-patterns and gives fixes for Shopify, WordPress, Next.js and eight more stacks, and the CrUX cross-reference puts Google's p75 next to yours.

DebugBear, RUMvision, SpeedCurve, Request Metrics and Datadog are close on this, and RUMvision has more dimensions. New Relic has the phases but no LoAF. Compared with mPulse, Vercel, Cloudflare and CrUX the difference is big. The column I'd miss most in any other tool is LCP priority. It shows you when the hero image has loading="lazy" on it, which I still find on a lot of sites.

Your AI agent reads the data directly

This is what I'm most interested in right now. The CoreDash MCP server is included in every plan, Starter too. It's the same endpoint as the API: one key, JSON over HTTPS, and read-only, so there's no write path an agent could misuse. The three tools are made for Core Web Vitals. get_metrics returns the value with its good, needs improvement or poor rating, at p50 to p95, filtered and grouped by any dimension. get_timeseries tells you if a metric is improving or regressing. get_histogram shows the long tail. "Which LCP element is slowest on mobile product pages in Germany" is one call from Claude, Cursor, Windsurf, VS Code, Gemini CLI or a script in your CI.

Having MCP isn't rare anymore. DebugBear, RUMvision, Datadog and New Relic all have a server now, and RUMvision's has more tools than mine. Two things about the CoreDash setup are still unusual. CWV Superpowers is a free, MIT-licensed agent that starts from your field data, records a Chrome trace to find what causes the problem and ends with a code fix (file, line and diff) or an HTML report. And with the Agency API that same agent can read every client project with one master key. Fixing INP with an AI agent shows a real run.

If you just need an API to pull Core Web Vitals data out of a RUM platform, most tools here have one: CoreDash, DebugBear, RUMvision, SpeedCurve, Datadog, New Relic and Cloudflare, plus mPulse outside the Lite tier. Request Metrics has CSV export, Vercel a CLI and Drains. Google's CrUX API is free, but it only returns CrUX data. Your own visitors aren't in it.

Priced by pageviews, every feature on every plan

Every CoreDash plan has every dashboard, the MCP server and unlimited tracked pages per site. You pay for pageviews: $16 a month for 75K, $31 for 500K, billed annually. Retention is counted in years (3 on Studio, 8 on Enterprise), so you can put this Black Friday next to the last two.

Datadog charges per session and New Relic per GB plus seats, and both keep raw data for days or weeks. DebugBear and SpeedCurve bundle RUM with lab testing, which is good value if you actually run the lab tests. SpeedCurve keeps RUM for 3 months below $500 a month, Request Metrics for 30 or 90 days.

Agencies

The agency plans pool pageviews across clients. Studio covers 10 client sites and 1M pageviews for $71 a month, Growth 25 sites and 3M pageviews for $159, both with unlimited users. One master key creates, lists and removes client projects over REST. CoreDash isn't white-label: you keep your own reporting and CoreDash supplies the data. If you send clients branded PDFs, DebugBear and Request Metrics are better at that.

When another tool is the better pick

• You want scheduled lab tests and field data in one tool. Look at DebugBear or SpeedCurve.
• Your slow pages are mostly slow servers and you already run Datadog or New Relic. Tracing from the browser into the backend beats anything else on this page.
• You want to watch recordings of slow sessions. New Relic and Datadog have session replay, CoreDash doesn't.
• You want to forecast what a faster site would earn. mPulse has what-if forecasting built in.
• You want a consultant with the tool. RUMvision includes monthly consultancy on every plan.
• Your budget is zero. Cloudflare Web Analytics, the free tiers of Vercel, New Relic and Request Metrics, or CrUX all give you real field data for free.
• You need to benchmark competitors. Only CrUX has field data for sites you don't own.

Or run two. The CoreDash script is under 2 KB and async, so it runs fine next to Datadog, New Relic or mPulse. The APM keeps the traces and you use CoreDash for Core Web Vitals.

Try it on your own site

The trial is 14 days without a credit card, and data comes in from the first pageview. Have a look at the demo first if you want to see the dashboards, check pricing for your traffic, or go straight to installation.

Sources: DebugBear pricing, SpeedCurve pricing, RUMvision pricing, Request Metrics pricing, Datadog pricing, New Relic pricing, Akamai mPulse, Vercel Speed Insights pricing, Cloudflare Web Analytics, CrUX methodology. Each comparison page lists its full sources.