Core Web Vitals, diagnosed and fixed
Diagnostic method, common causes ranked by frequency, measurable outcomes.
Request a proposalPain points
Where you lose money
Pain points
Where you lose money
Guessing instead of diagnosing
Teams chase random tips instead of measuring field data, so effort goes to problems that aren't the bottleneck while the real one persists.
Lab score, real-world fail
A green Lighthouse run masks poor INP and layout shift for actual users on mid-range phones and slow networks.
Third-party weight ignored
The biggest LCP and INP hits often come from tags and embeds, but those are treated as untouchable business requirements.
Fixes that don't stick
Without CI performance budgets, every new feature re-breaks the vitals a team just spent weeks fixing.
Channel mix
What we'd run
Channel mix
What we'd run
We measure real-user INP, LCP and CLS to find the actual bottleneck, instead of chasing tips that don't move it.
We tackle the tags and embeds that cause the biggest hits, treating them as negotiable rather than untouchable.
Causes fixed in order of user impact, with before/after field data to prove each one moved the number.
Budgets in the pipeline so the vitals you fixed stay fixed as new features ship.
FAQ
Do Core Web Vitals really affect rankings?
They're a real but secondary ranking signal, and a large direct conversion factor. Poor LCP/CLS costs you conversions today regardless of what it does to rank. We fix them for both reasons.
Which metric usually fails first?
LCP (largest paint) from heavy images and render-blocking scripts, and CLS (layout shift) from unsized media and injected content. We measure field data, not just a lab score, then fix the real offenders.
Lab score vs real users, which matters?
Field data (what real users experience) is what Google uses and what reflects revenue. A green lab score with red field data means real visitors still suffer, we optimise the field numbers.
How long until it improves?
On-page fixes land immediately for users; Google's field data updates over ~28 days as it collects real visits. Conversion impact shows first, ranking impact follows.
Get started