Stop Chasing a Green Lighthouse Score

Run Lighthouse on a heavily customised Shopify store with a real app stack, real reviews, real upsells, a real chat widget, and you'll rarely see green. That's not a sign something's broken. It's a sign the store is doing its job.

Every script a merchant installs to make more money, on-site search, personalisation, loyalty, live chat, also blocks the main thread for a few extra milliseconds, and Lighthouse punishes that hard.

"Chasing 90+ on a store built the way most $1M-$20M brands actually run their business is an expensive, often pointless exercise. The number worth watching isn't the score. It's whether the specific things a shopper feels are fast."

What the score is actually measuring

Lighthouse's performance score is a weighted blend of five lab metrics: Total Blocking Time carries the most weight at 30%, Largest Contentful Paint and Cumulative Layout Shift are tied at 25% each, and First Contentful Paint and Speed Index take 10% each. TBT alone is close to a third of your score, and it's the metric third-party scripts hit hardest, because it measures how long the main thread is tied up in long tasks during load. Stack five apps that each fire their own script on page load and TBT climbs fast, dragging the whole score down even if nothing about the actual shopping experience got worse.

"None of this is what Google uses to rank you. Google's ranking signal comes from the Chrome UX Report, real field data collected from actual visitors over the previous 28 days, not from a single simulated Lighthouse run."

Search Console's Core Web Vitals report shows you that field data, scored against three metrics: Largest Contentful Paint, Interaction to Next Paint (which replaced First Input Delay as a Core Web Vital in March 2024), and Cumulative Layout Shift. A store can sit at a 60 in Lighthouse and pass Core Web Vitals cleanly in Search Console, or the reverse. They're measuring different things, on different visits, with different weightings, and only one of them touches your rankings.

The two numbers that map to revenue

If you're only going to watch two things, watch Largest Contentful Paint and how long it takes before a tap actually does something. LCP is close to "time to main image," the point your hero product shot, banner or above-the-fold content finishes rendering. Under 2.5 seconds is good, over 4 seconds is poor, and every extra second past that is a shopper looking at a half-loaded page deciding whether to wait. That's the moment a browsing decision gets made.

The second is responsiveness once the page has loaded, effectively "time to a usable buy button." Google's field metric for this is Interaction to Next Paint, and the target is under 200 milliseconds; over 500ms is poor. Lighthouse doesn't measure INP directly in a lab run, it uses Total Blocking Time as a proxy, but the underlying question is the same: when someone taps Add to Cart, does the button respond straight away, or is the main thread still busy running a script from an app that finished rendering minutes ago? A customer who taps and gets nothing taps again, then leaves.

What's vanity, and what buying the last points actually costs

Speed Index and First Contentful Paint move your overall number around a lot without telling you much a shopper would notice. Speed Index in particular rewards visually quiet pages, sparse ones with little going on above the fold, which is the opposite of what most retail homepages are trying to do. Treat a 10-point swing in either as noise unless it comes with a specific, named fix attached.

The real cost shows up when an agency promises to get a customised theme with a full app stack into the 90s. Getting there usually means stripping out the reviews widget, the personalisation engine, the loyalty program or the live chat, the very tools installed to lift conversion and average order value, because those are what's hammering TBT. You end up trading a working revenue lever for a lab score nobody outside your own dev team will ever check. That's a bad trade nine times out of ten.

"The better question isn't 'how do we hit 90,' it's 'which specific app on this page is adding the most blocking time for the least commercial return,' and that's a much shorter, much cheaper list to work through."

Chrome DevTools makes that list easy to find. Open the Performance panel, record a page load, and look at the main thread track for long yellow blocks, each one is a script holding up everything else, and hovering over it names the file. Nine times out of ten it's the same handful of categories: a chat widget polling for updates before anyone's opened it, a personalisation pixel firing on every product image, or a reviews app that loads its full JavaScript bundle even on pages with zero reviews to show. Fixing three specific scripts usually buys back more real responsiveness than a blanket "optimise everything" pass, and it doesn't touch the apps actually doing commercial work.

Measure it yourself

1. Open Search Console's Core Web Vitals report. See what real visitors are experiencing over the last 28 days, that's the number tied to your rankings.
2. On your own phone, on your own mobile data, load a product page. Time how long after it looks finished before a tap on Add to Cart actually does something.

If that gap is under half a second, you're in better shape than your Lighthouse score probably says you are.

At beCommerce, we separate real performance issues from Lighthouse vanity metrics as part of every technical review. If you want a hand finding which scripts are actually costing you, book your free audit →.