The short answer: Core Web Vitals are three numbers — loading (LCP), responsiveness (INP), and visual stability (CLS) — and Google publishes the exact thresholds a page should meet. For an HVAC website the stakes are concrete: the visitor most sensitive to a slow page is the one with a dead system and three competitor tabs open. This note explains each metric in contractor terms and where the weight actually hides.
What are the three metrics, and what counts as good?
Google’s documentation defines all three, measured at the 75th percentile of real page loads. Largest Contentful Paint — how long the main content takes to appear — is good at 2.5 seconds or less (web.dev, LCP). Interaction to Next Paint — how quickly the page responds when someone taps — is good at 200 milliseconds or less (web.dev, INP). Cumulative Layout Shift — how much the page jumps around while loading — is good at 0.1 or less (web.dev, CLS). Source check: official web.dev documentation reviewed September 8, 2026. Those three thresholds are the whole scoreboard; everything else in this note is about earning them.
Why should a contractor care about lab numbers at all?
Two reasons, one commercial and one honest. Commercially, the metrics describe the exact experience of your highest-intent visitor: the emergency caller on a phone, on cellular, in a hot house. A page that paints in under 2.5 seconds and responds to the first tap without stalling converts that visitor; one that doesn’t hands them to the next tab. The honest reason: vitals are one of the few marketing claims you can verify yourself, free, in minutes — no vendor report required. Search-ranking effects exist but are modest and contested; treat them as a tiebreaker, not the pitch. The conversion argument stands on its own.
Where does LCP actually go wrong on contractor sites?
Almost always in the same three places: a hero image uploaded at camera resolution and painted last, a page-builder theme that loads a dozen scripts before content, and fonts fetched from third parties that block text. The fixes are subtraction, not wizardry: size and compress the hero to what the layout displays, mark it as the priority resource, self-host the one or two fonts that matter with sensible fallbacks, and defer every script that is not needed to paint the first screen. The conversion-paths note treats the 2.5-second budget as an engineering constraint on the emergency page specifically — the page where the visitor’s patience is shortest and the payoff for meeting the budget is largest.
What makes INP fail, and why does it hide in testing?
INP fails when the browser’s main thread is busy doing something else at the moment the visitor taps — usually parsing a pile of scripts, running a chat widget’s startup, or handling an analytics burst. It hides in testing because lab tools load the page and touch nothing, while real visitors tap immediately; a page can score perfectly in a passive test and still stall on the first real interaction. The practical defense is to keep the first tap cheap: load heavy scripts after the page is interactive, split long tasks, and let the tap-to-call button be a plain link rather than something a framework has to hydrate before it works. Then test like a visitor — load the page on a mid-range phone and tap the phone number instantly.
What shifts layouts, and why is CLS the cheapest fix?
Layout shift comes from elements arriving without reserved space: images without width and height attributes, banners and badges injected above content, fonts swapping late and reflowing text. It is the cheapest vital to fix because the causes are visible in one slow reload — watch what jumps, then reserve its space. Give every image its dimensions, keep late-arriving content below the fold or inside a fixed-size slot, and preload the primary font. On a contractor site the worst offender is usually a call-now banner that pushes the page down right as the visitor aims at a link — which converts a ranking metric into a missed tap on the exact button the site exists for.
How do the tools contractors already use eat the budget?
Most contractor sites are assembled from vendor parts, and each part bills the vitals: the page builder ships its own framework, the chat widget starts a conversation nobody asked for during load, the review-badge script phones home before the hero paints, and two generations of tracking pixels outlive the campaigns that installed them. No single part looks expensive; the sum is the six-second page. The audit method is removal, not tuning — stage the site, strip every third-party script, measure, then add them back one at a time and watch which ones move the numbers. Keep what earns its cost, load it after the page is interactive, and delete the rest. Two rules of thumb survive most audits: one analytics tool is enough, and a chat widget that must exist should arrive only after first interaction. The point is not purity; it is knowing the price of each passenger.
What does staying fast look like after the project ends?
A habit, not a milestone. Sites regress the same way they slow down in the first place — one plugin, one banner, one uncompressed photo at a time — so the defense is a boring cadence: re-measure the money pages monthly and after any theme, plugin, or widget change, on mobile settings, and write the three numbers into a dated log next to the month’s call counts. When a number moves, the changelog is short enough to find the culprit; when a vendor proposes adding something, the log is the before-picture that keeps the decision honest. This is the same record-keeping discipline the measurement note applies to leads, pointed at the website itself — and it is what makes the redesign checklist’s before/after comparison possible when a rebuild eventually comes.
How do you actually measure your site this week?
Use both kinds of data. PageSpeed Insights gives you lab numbers plus — if your site has enough traffic — field data from real Chrome users, which is what the thresholds formally apply to. Test the money pages, not just the homepage: the emergency page, the quote request, the maintenance plan. Test on mobile settings first, because that is where the emergency visitor lives. Record the three numbers per page, fix the worst offender, and re-measure — one variable at a time, so you know what worked. If a redesign is on the table, capture the numbers before the switch; they are the baseline the new site has to beat, and the launch-week checklist should include re-measuring them.
When are the vitals not the problem?
Often. A fast page with a buried phone number still loses to a slower page that answers the visitor’s question above the fold. Vitals are a floor, not a strategy: they keep the door from slamming before the visitor sees what you offer, and nothing more. If the three numbers are already green, the next dollar of effort belongs in verifiable proof and clearer paths, not in chasing a higher score. The order of operations is the one the design practice runs: make the page answer fast, then make it answer well — and never let a performance audit talk you into deleting the content that does the convincing.