Core Web Vitals checklist for LCP, INP, and CLS

You open Google Ads on a Tuesday and CPC is fine. CTR is fine. Then you check PageSpeed and field data for the same landing URL and LCP is still sitting north of 4 seconds on mobile. That is not a “nice to have” UX issue. That is paid demand landing on a page that fails the handoff before the form ever loads.

As of the 2024 Web Almanac, around 43% of mobile origins pass all three Core Web Vitals - up from 29% in 2022. The rest are burning attention, scroll patience, and often ad spend on pages that feel broken on real phones in real networks. A tactics list (“compress images,” “use a CDN”) does not tell you which leak is costing you pipeline this week. A scored audit does.

This checklist is built for marketing leaders who own traffic quality and conversion cost in the US market: heads of growth, CMOs, founders running the marketing number. You will score LCP, INP, and CLS the way an operator audits a campaign: inspect, mark pass / weak / fail, then fix in priority order so you stop sending expensive clicks into slow, jumpy, unresponsive pages.

Why a scored Core Web Vitals audit beats another speed tips list

Speed tips assume the problem is “the site is slow.” Your problem is usually more specific. One hero video blocks LCP. A third-party chat widget tanks INP on mobile. A late-loading banner shifts the CTA and spikes CLS right when someone tries to tap “Book a demo.” Aggregate lab scores hide that. Field data and a line-item scorecard do not.

Marketing cares because mobile still drives the majority of paid search clicks, while conversion rates on mobile routinely lag desktop. When your money page fails LCP or INP on phone, you do not just lose a bounce metric. You raise effective cost per qualified lead without touching bid strategy. Search CPC pressure already sits in the multi-dollar range for competitive terms; every second of dead time multiplies that waste.

Page experience is also entangled with how Google surfaces and ranks URLs, but you do not need a ranking debate to justify the work. If Meta or Google is sending traffic and the first contentful paint is late, creative and keyword quality never get a fair test. Fix measurement of the page the same way you fix conversion tracking: verify the signal before you scale spend.

Use field data (CrUX / Search Console experience reports / real-user monitoring) as the source of truth for pass or fail. Lab tools catch regressions in staging. Lab alone is not your scorecard for live money pages.

If paid or organic traffic already hits a template that feels sluggish, pair this audit with focused website performance work rather than another creative refresh that cannot outrun a 5-second LCP.

Core Web Vitals scorecard: LCP, INP, CLS, and the systems around them

Score each item against your primary landing templates (home, paid landing, top organic commercial URLs). “Good” means field data in the recommended thresholds for a meaningful share of visits, not a green lab badge on one desktop run. Mark pass, weak, or fail as you go.

Audit scorecard

1. 0 JavaScript long tasks and hydration cost If you ship a heavy SPA or marketing site framework, measure hydration and long tasks after load. Good: progressive enhancement on landers; minimal client JS for forms. Fail: full app bootstrap before the lead form is usable. Cost of fail: INP tanks on mid-tier Androids common in US paid traffic mixes even when desktop lab looks fine.

2. 1 Template parity: ads vs SEO vs homepage Compare the URL Google Ads or Meta hits versus the SEO money page versus the homepage. Good: dedicated landers with a smaller script and media budget than the global chrome-heavy homepage. Fail: “we just send traffic to the homepage.” Cost of fail: you pay for intent and inherit every global tag and mega-menu cost. That pattern shows up constantly in buyer complaints about agencies and internal teams alike.

3. 2 Field monitoring and regression alerts Good: CrUX/Search Console reviewed on a cadence, RUM or synthetic checks on top landers after deploys, owners named for performance regressions. Fail: a one-time PageSpeed screenshot from last quarter. Cost of fail: a “small” tag or hero swap undoes months of work and only shows up when CPL drifts.

4. 3 Post-click behaviour vs pure lab greens Even with improving vitals, forms can still fail. Use session-level evidence to see rage clicks, dead taps on shifting CTAs, and scroll abandonment above the fold. Good: performance fixes validated against real behaviour on the same URLs. Fail: green Lighthouse with a form nobody completes. Pair vitals work with HeyLead Insights so you see whether CLS or INP issues show up as mis-taps and drop-off, not only as a lab metric.

DIY tools for this audit

Free checks you can run on production URLs

• Core Web Vitals checker

• H1 checker

• Title tag checker

• Meta description checker

• Open Graph preview

• Schema markup validator

Run the Core Web Vitals checker on every URL that spent meaningful ad budget last month. Title, H1, and meta checks keep you honest that the “fast” template is still the commercial page you meant to send traffic to, not an orphan experiment.

How to score pass, weak, and fail - and what to fix first

Keep the rubric simple so marketing and engineering share one language.

• Pass: Field data in the “good” band for that vital on the URL group, and the underlying mechanism (LCP element, script budget, space reservation) is intentional.

• Weak: Borderline field data, or good averages with a bad mobile p75; mechanism is known but not controlled.

• Fail: Field data in “poor,” unknown LCP element, unbounded third parties, or layout shifts on the primary CTA path.

Priority order matters more than the length of the backlog:

• Know the LCP element and fix delivery (image priority, weight, TTFB) before redesign debates.

• Cut main-thread and third-party cost on conversion paths before adding another tag “for attribution.”

• Kill CLS on forms and sticky UI before new creative tests that depend on a stable tap target.

• Only then scale paid into the template. Budget on a failing lander trains algorithms on bad post-click outcomes.

Do not prioritize micro-optimizations on blog templates while the paid demo page fails INP. Score by business URL value: revenue landers first, then high-impression SEO commercials, then the rest.

When lab and field disagree, trust field for go/no-go and use lab to debug. Plenty of teams celebrate a local Lighthouse 95 while CrUX still shows poor LCP on mobile because real devices, cache states, and third parties differ.

Soft check: if your scorecard is mostly red on landers that carry Google or Meta spend, pause net-new tests that add weight (video backgrounds, extra pixels) until the top fails clear. Creative cannot fix a 4-second LCP.

A 30-day repair plan for LCP, INP, and CLS

This is ordered repair, not a new brand strategy. Assign an owner for each block.

30-day fix plan

1. Days 1-3 - Inventory. List top 10 URLs by paid spend and top 10 by organic commercial sessions. Pull field CWV where available. Name LCP elements. Export the GTM/tag list on those templates.

2. Days 4-10 - LCP sprint. Compress and correctly size heroes; set priority on LCP media; stop lazy-loading LCP; fix cache/CDN for HTML/assets; remove autoplay video as the LCP candidate. Re-check lab, then watch field over the following weeks.

3. Days 11-17 - INP sprint. Defer or remove non-essential third parties on landers; break up long tasks; simplify form widgets; delay chat until after first interaction or intent. Validate tap responsiveness on mid-tier mobile hardware, not only flagship phones.

4. Days 18-24 - CLS sprint. Reserve image/embed space; stabilize webfonts; lock cookie/chat/promo bar behaviour so they do not shove the CTA. Re-test multi-step forms on small viewports.

5. Days 25-30 - Guardrails. Add a performance check to release process for money templates; document the script allowlist; compare conversion rate and CPL on the repaired URLs against the prior 30 days (directional, same campaigns where possible). Wire behaviour review so you confirm fewer rage taps and form abandons, not only better vitals.

Example pattern we see on US B2B landers: a Phoenix SaaS team cut a full-screen hero MP4 and two personalization scripts from the demo page, reserved space for the logo bar, and moved chat to post-scroll. LCP dropped from a poor band into a competitive one on mobile field data within a few weeks, and form starts stopped falling off a cliff between first paint and first input. The mechanism was not “more brand.” It was fewer bytes and fewer main-thread competitors on the conversion path.

Another common fail: an agency-built homepage chrome copied onto every paid URL. Chicago contractor lead-gen traffic hit a mega-menu, review carousel, and three pixels before the quote form. Splitting a bare lander template with the same offer usually beats another bid tweak.

What marketing leaders are seeing

A pattern we see repeatedly on B2B SaaS landers: teams scaling Search while mobile LCP sits over four seconds on the demo URL. CPL looks like a bidding problem until someone names the LCP image and strips the tag pile - same keywords, different page physics.

Another pattern: INP only becomes real when sales reports that prospects swear the form button did nothing. Usually it is a chat widget and a heavy calendar embed fighting the main thread on Android, not the offer copy.

FAQs

Do Core Web Vitals still matter if we mostly buy ads?

Yes. Ads buy the click; the page decides whether the click becomes a lead. Poor LCP and INP raise effective CPL and muddy creative tests. You also still want field experience healthy on SEO URLs that assist the journey.

Should we chase a perfect 100 on Lighthouse?

No. Chase field thresholds on URLs that carry revenue, with clear owners and regression checks. A 100 lab score with poor CrUX on mobile is a false win.

What is a reasonable order if engineering time is limited?

LCP element and media on the top paid lander, then third-party reduction for INP, then CLS on the form and sticky UI. Leave blog polish for later.

How often should we re-score?

After any deploy that touches global JS/CSS, tags, or heroes on money templates, and at least monthly on the URL set that spends. Treat performance like conversion tracking: it drifts when nobody watches. Set up automated Core Web Vitals monitoring - even a simple CrUX API pull on a weekly cron - so regressions surface before they show up in CPL.

Can content or CRO wait until vitals are green?

If the URL’s LCP is in the poor band (>4s on mobile field data), any A/B test on that page is statistically compromised - slow pages suppress conversion rates unevenly across devices and time of day. Fix LCP first, then test copy or offer. CRO tests on CLS or INP issues are fine to run in parallel since they are non-destructive layout changes.

Putting it to work

This week

• Export the 10 URLs with the highest ad spend and pull field CWV where available.

• Name the LCP element on each; screenshot the failing mobile trace.

• List every third-party script on those templates and mark must-keep vs kill/defer.

• Run the Core Web Vitals checker and note pass/weak/fail per URL.

• File one engineering ticket for the single worst LCP fail and one for the worst INP fail (not a vague “make site faster” epic).

• Watch 5-10 session replays on the worst lander for rage taps and layout shift at the CTA.

Next 30 days

• Ship the LCP media and caching fixes on the top landers.

• Enforce a script allowlist on paid templates.

• Reserve space and stabilize fonts/banners for CLS.

• Re-score field data and compare CPL/CVR directionally on the same campaigns.

• Add a release checklist so new tags cannot silently undo the work.

When the backlog is clear but your team is buried in tag debt, hero swaps, and lander variants, HeyLead can own the operational loop on Core Web Vitals for revenue URLs - LCP element work, script discipline, CLS on forms, and proof that post-click behaviour actually improved - so you are not guessing from a single lab screenshot. Start a conversation at [email protected].