Free tool

What is a faster page actually worth to your site?

Enter a URL — we pull your real-user speed from Google's field data — or paste your own RUM p75. Then we price the milliseconds against your revenue as a range built from our own research, never a single made-up multiplier — and hand you a card built for the meeting.

Where should the baseline come from?
Analysis profile
~15–30 s · Google PageSpeed Insights · CrUX field data + Lighthouse lab
  • CrUX field p75 when Google has it
  • Your own RUM p75 accepted
  • A range, never false precision
  • Revenue never stored or logged

Example output

$129.6k–$211.4k per year

3.8s → 2.5s LCP · recovers 1.3s

Modeled metric: LCP INP, CLS and TTFB are context — they earn nothing here

Illustrative — your URL and your revenue replace every number here. The range is this page's own model, not a placeholder.

Method and limitations

A prior, not a promise.

This estimate is the most honest one we can make from public data — and it is deliberately built to resist the over-claiming that plagues every “speed = money” calculator on the internet.

λPinned to your baselineSpeed's effect is local. We anchor the elasticity to your measured LCP, so the number reflects your site — not a headline borrowed from someone else's.
1One metric, on purposeRevenue is modeled on LCP alone — the metric with the strongest causal evidence. INP, CLS and TTFB are shown as context and earn nothing here.
CLS earns nothing hereYour layout stability matters for UX — but in our data, CLS doesn't move conversion. So we leave it out of the dollars rather than pad the result.
±A band, not a decimalThe range is the 0.5–0.8%/100 ms corridor. False precision is the tell of a bad model; honest uncertainty is the point.
The formula, the exclusions and the caveats

Conversion elasticity ε ≈ −0.21 (corridor 0.15–0.24), applied as a ratio between two measured points rather than a constant per-100 ms rule: CVR(target)/CVR(baseline) = (baseline/target)^ε. %Δrevenue ≈ %Δconversion, with traffic and average order value held constant — if a faster page also brings more sessions, this understates it; if your conversion is constrained by something other than speed, it overstates it.

Targets come from Lighthouse's LCP-specific metricSavings, dampened harmonically because overlapping fixes do not add linearly, capped at 60% of measured LCP, and floored at the 1.5 s best-in-class bar. Nothing here is a guarantee: to turn this prior into a measured fact, run a holdout test or an engagement that finds where your critical path is slow — and size the proof before you run it.