CoreDash vs CrUX and Search Console
Everyone should start with Google's free field data. Here's what CrUX, PageSpeed Insights and Search Console do well, and when it makes sense to pay for real-user monitoring.
This is a bit of a strange comparison, because CrUX isn't really a competitor. CrUX, PageSpeed Insights and Search Console are one free Google dataset shown in three places, and every site owner should look at it. It's the first thing I check on any site, also on sites that already run CoreDash.
The real question is whether the free data is enough for your site. I built CoreDash, so keep that in mind. But Google's data is good, so let's start there.
What Google's free field data does well
It's free and there's nothing to install. Chrome measures LCP, INP, CLS, FCP and TTFB for users who opted in, and Google publishes the result as the Chrome UX Report. PageSpeed Insights shows that data on top of a Lighthouse run. Search Console groups it into URL groups and has a 28-day validation flow for fixes. CrUX Vis shows the weekly trend for the last 40 weeks. It replaced the Looker Studio dashboard (which Google retired at the end of November 2025) and it's a lot easier to read.

developer.chrome.com, October 2026.
CrUX is also the dataset Google uses for Core Web Vitals in Search. So when a client asks "do we pass?", everyone looks there, their SEO agency included. No RUM tool replaces that, and CoreDash doesn't try to. (Google also says good scores don't guarantee rankings, which is worth telling that client.)
It works for any public site, your competitors included. You can pull the numbers for every shop in your niche in an afternoon. A RUM tool only measures sites where you've added the script, so for benchmarking and prospecting CrUX is the obvious choice.
BigQuery goes back to 2017, with monthly data per origin. The History API gives you 40 weekly periods. The August 2026 release covered 18.3 million origins.
And Google keeps adding things. Since January 2025 CrUX splits LCP into subparts (TTFB, resource load delay, resource load duration, element render delay) and tells you whether the LCP was an image or text. September 2026 added experimental ad metrics. The official Chrome DevTools MCP server now pulls CrUX field data into a performance trace by default, so an AI agent sees Google's p75 next to the lab trace. The CrUX API is free at 150 queries per minute per Google Cloud project, and lots of third-party tools are built on it.
Start here, then measure your own visitors
If you don't have any real-user monitoring, CrUX is your field data. That already puts you ahead of teams that only look at a Lighthouse score. I've seen teams celebrate a 98 in PageSpeed Insights while field INP on their product pages was poor. Lighthouse loads the page once and never clicks anything, so any real-user data tells you more than that.
And any RUM tool is better than none. If you already run one and it answers your questions, you don't need this page. It's for people who've outgrown what the free data can tell them, and for people whose site isn't in CrUX at all.
CoreDash and Google's free data side by side
Pricing and features checked September 2026.
| CoreDash | CrUX, PageSpeed Insights, Search Console | |
|---|---|---|
| Price | From $16 a month, billed annually (75K pageviews). 14-day trial, no credit card | Free. BigQuery is free up to its monthly free tier |
| Setup | One script tag or the WordPress plugin | Nothing to install. Search Console needs verified ownership |
| Who gets measured | Every visitor the script loads for, on every browser that supports the metric | Opted-in Chrome users on desktop and Android. No iOS, Safari, Firefox or Edge |
| Minimum traffic | None | "Sufficiently popular" pages and origins. The threshold is not published |
| Freshness | Pageviews show up within seconds | Daily 28-day rolling p75 (API, PSI). Weekly History API. Monthly BigQuery |
| Finest level | Every URL, and every single pageview | One URL if it has enough traffic, otherwise the origin. Search Console shows URL groups |
| LCP attribution | Element, element type, image URL, preload and priority state, four phases | Four subparts at p75, image LCPs on full page loads only. No element |
| INP attribution | Element, input type, load state, LoAF script URL, three phases | None |
| CLS attribution | Shifting element, cause and timing | None |
| Segments | 25+ dimensions, including page labels, logged-in state, A/B variant, country and device capability | Form factor (phone, desktop, tablet). Country only in monthly BigQuery tables, per origin |
| History | 6 months to 8 years, depending on plan | 40 weeks in the History API. Monthly origin data back to 2017 in BigQuery |
| Alerts | Custom alerts and AI Auto Alerts by email and Slack | Search Console emails site owners about new Core Web Vitals issues. No custom thresholds |
| API and MCP | API and read-only MCP server on your own data, one endpoint | Free CrUX API and PSI API. Chrome DevTools MCP adds CrUX p75s to lab traces |
| Other people's sites | No, only sites running the script | Yes, any public origin |
What CoreDash adds
The free data is very good at telling you if you pass. Once you start fixing things you want other answers: which element is slow, which script blocks the click, did yesterday's deploy make it worse. CrUX was never made for that. It's a public, anonymous set of p75 values, which is exactly why Google can publish it for millions of origins for free.
Visitors and pages CrUX skips
CrUX only counts Chrome users on desktop and Android who have usage statistics on, sync their history and don't have a sync passphrase. Chrome on the iPhone doesn't count, and neither do Safari, Firefox or Edge. On top of that the page has to be public, indexable and popular enough. Pages behind a login never qualify, and Search Console falls back to the origin score for URL groups that don't have enough data themselves. For a lot of sites that means "not enough usage data" (I wrote about why that happens).

CrUX methodology on developer.chrome.com, October 2026.
CoreDash measures every pageview where the script runs. A page with 50 visits a month has data, and so do your iPhone visitors and your logged-in customers. It sets no cookies by default and collects no personal data, so you don't need a consent banner and you're not just measuring the people who clicked accept.
Which element to fix
This is what I need most in my day-to-day work. CrUX tells you LCP on a URL is 3.4 seconds at p75, and since 2025 it also tells you most of that is resource load delay. Useful. But not which image, on which template, or whether someone put loading="lazy" on the hero again (I still run into that every few weeks).
For INP, CrUX gives you one number and that's it. Which button, which script, did the click happen while the page was still loading? No idea. That's where most INP work gets stuck: you know it's bad but not where to start.
CoreDash records the element behind every LCP, INP and CLS. For LCP you also get the element type, the image URL and the LCP priority (preloaded, fetchpriority high, not preloaded or lazy loaded). For INP you get the input type, the load state and the Long Animation Frame script that blocked the main thread. You can split all of that by page label, logged-in state, A/B variant, country or device capability. The Diagnosis view checks for 14 LCP anti-patterns and gives fixes for your stack.
You see a change the day it goes live
CrUX data is only a day or two old (the 28-day delay is a myth). But it's a 28-day p75, so a fix moves the number a little each day and only shows fully after four weeks. A regression on one template gets averaged in with the rest of the site. Search Console does email you about new issues, but only once the 28-day number has moved far enough to change the status of a URL group.
In CoreDash a deploy shows up in the time series within hours. Release tagging from your CI puts the p75 of each release next to the previous one. AI Auto Alerts check hundreds of segments every day and only flag a regression when it affects at least 2% of your traffic, with the metric, the segment and the number of users hit.
Your AI agent reads your own field data
The CrUX API is great for scripts, and the Chrome DevTools MCP server gives an agent Google's p75 next to a lab trace. But that's still a page-level or origin-level number from opted-in Chrome users.
The CoreDash MCP server is the same endpoint as the API and it's read-only. Claude, Cursor or any other agent can ask for any metric at p50 to p95, filtered and grouped by any dimension. "Which LCP element is slowest on mobile product pages in Germany" is a single call. CWV Superpowers is a free, MIT-licensed agent that starts from that field data, records a Chrome trace with mobile emulation, Fast 3G and a 4x CPU slowdown, and ends with a code fix (file, line and diff) or an HTML report. No Lighthouse scores. Fixing INP with an AI agent shows a real run.
Agencies get an Agency API with one master key that creates client projects and reads all of them with the same three tools. The price only goes up with traffic, every plan has every dashboard, and the agency Studio plan covers 10 client sites for $71 a month, billed annually.
When to stick with the free data
- You want to know how Google sees your site. That's CrUX by definition, and you should keep checking it no matter what else you run.
- You're benchmarking competitors or prospects. CrUX is the only field data you can get for sites you don't own.
- You're doing research across lots of sites or years. BigQuery back to 2017 is hard to beat, and free up to the free tier.
- Your site passes on mobile and desktop, the pages you care about have their own data, and the frontend rarely changes.
- You can't add a script to the site.
- There's no budget. CrUX plus a Lighthouse run on the page you're fixing is a sensible setup, and much better than guessing.
Use both
I'd run both. CoreDash has a CrUX cross-reference that puts Google's numbers next to your own RUM data, so you see what Google sees and what causes it. You can also check any site's CrUX history for free with the CrUX History tool on corewebvitals.io.
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 see all comparisons.
Sources: CrUX methodology, CrUX metrics, CrUX API, CrUX History API, CrUX on BigQuery, CrUX release notes, CrUX Dashboard deprecation, CrUX ad metrics, About PageSpeed Insights, Search Console Core Web Vitals report, Google Search and Core Web Vitals, Chrome DevTools MCP server.