From free CrUX to enterprise mPulse — real user monitoring products compared by data depth, retention, pricing shape and how well they serve delivery decisions.
Winner depends on your workload.
Winner depends on: what the data must decide. Free CrUX answers 'how does Google see us' and nothing per-session; performance-first tools (SpeedCurve from ~$90/mo, DebugBear from ~$49) serve web-perf teams; observability suites (Datadog, New Relic) win when RUM must join traces; delivery-grade RUM (mPulse, Catchpoint) is what CDN and multi-CDN decisions actually require, because it keeps the network dimensions the others aggregate away.
Same acronym, four products
Everything sold as RUM ships a JavaScript beacon and a dashboard of Core Web Vitals; the divergence is what else survives collection. For delivery work — the audience of this site — the decisive question is whether the tool retains the network dimensions: DNS and connect timings, protocol, POP or ASN, per-resource waterfalls. Aggregate LCP by country is a marketing chart; LCP by ASN and CDN POP is an engineering instrument, and the market splits cleanly on which one it sells.
| Tier | Products | What you get |
|---|---|---|
| Free baseline | Chrome UX Report (CrUX) | Google's own field data, origin/URL-level, Chrome-only, 28-day aggregates; the SEO ground truth, useless for sessions or non-Chrome reality |
| Performance-first | SpeedCurve (RUM+synthetic, from ~$90/mo), DebugBear (from ~$49), RUMvision et al. | Web-vitals depth, budgets, deploy tracking; frontend-shaped, lighter on network dimensions |
| Observability suites | Datadog RUM, New Relic Browser, Sentry, Grafana Faro (OSS) | Sessions joined to traces, replay, error context; retention and cost tiered — volume pricing bites at scale |
| Delivery-grade | Akamai mPulse, Catchpoint RUM | Network-dimension retention, ISP/path analysis, synthetic integration; enterprise-priced because CDN teams are the buyer |
The dimensions test
Before any demo, ask three questions of the data model. Can you segment a vital by ASN and by CDN POP or edge identifier? Can you see per-resource timing (the waterfall) for real sessions, not just page aggregates? And how long does raw, unaggregated data survive — because the incident you will investigate is three weeks old, and tiered retention that rolls sessions into daily averages has already destroyed the evidence. Tools fail these questions in predictable order: CrUX fails all three by design; performance-first tools pass the vitals half; the suites pass with per-GB or per-session bills attached; the delivery-grade tier passes because those dimensions are the product.
RUM as the steering fuel
For multi-CDN estates the stakes are higher than dashboards: real-user measurements are what the steering layer eats. The platforms we covered in the managed DNS roundup ingest RUM feeds — NS1's Pulsar consumes private beacons or aggregated cross-CDN benchmarks — which turns your monitoring choice into a routing dependency: sampling policy, beacon reliability and dimension fidelity now decide where traffic goes. If RUM will steer, treat it as tier-zero infrastructure, run the beacon first-party, and validate its numbers against a second source before automation trusts it.
The honest stack
Most estates should run two layers, not one perfect tool: CrUX (free, unarguable, what Google ranks you on) as the floor, plus one tool matched to the decisions — performance-first if the work is frontend, a suite if RUM must join traces, delivery-grade if the work is CDN engineering or steering. And whatever the choice, pair it with the synthetic layer for the reasons the two-sided argument in RUM vs synthetic lays out: real users tell you what happened; only robots tell you before it happens to someone who matters. Facts and prices verified against provider documentation, July 2026.
Choosing RUM that has to feed a steering layer — or just prove the CDN money worked? The assessment matches the tool to the decision.