Businesses generally use metrics as an indication of how well something is doing. In the computer business the implication is often that we can turn this the other way round. If we can improve our score on a metric, then whatever the metric is derived from must be doing better!

Lighthouse

Google's Lighthouse tool is a compound metric for how well a web site is built (there are others). The easiest way to see the Lighthouse score for a page is directly from the Chrome developer tools, where a Lighthouse tab enables the current web page to be scored from within the dev tools window. Pages are scored for various targets out of 100 for desktop and mobile, with 100 being best.

To get a higher score, you need less total bytes, a faster server response, fewer files, simpler rendering, less unused bytes, better caching, accessibility and more. The Lighthouse reports provides some helpful notes on where a page can be improved. The premise is that a better score is good for SEO. All sounds good.

A higher Lighthouse score = A better web page and better SEO

Diminishing returns

Well, only to an extent. Whilst a lot of the recommendations Lighthouse makes are good common sense, at some point chasing a better score battles against other factors, where a better score takes a lot of work and is often for the sake of the score while showing no improvement from a visitor's point of view.

For desktop, my experience is that this limiting speed score is somewhere between 80 and 95. Exactly where depends on the nature of the web site and how much you really care about the numbers. Getting to 100 usually takes a fast server and hand crafted pages.

So check your Lighthouse scores and look for the quick wins. This is where a few simple things such as image size, format, lazy loading, and a few .htaccess settings provide an easy gain. After that, it can become a trade off.

Mobile Lighthouse scores are generally lower, so you also need to consider whether your target audience is primarily desktop or primarily mobile. How much and where do you care? After all the quick gains, just breaking content up into smaller pages may provide the next step to a good mobile score.

Before getting further into the technical treadmill, take a step back, open a new incognito window, and get a feel for how your website renders on whatever desktop and mobile devices you have. If its doing its job, perhaps you should stop here.

A site is more than a home page

Don't limit your evaluation to just your home page. Most web sites have many pages. Now we get into a trade off between local optimization and global optimization. For example, consolidating all your CSS in one file which is then cached is excellent for global optimization. But that file results in unused CSS for any individual page and hence becomes a limiting factor for the Lighthouse score and local optimization.

For some components of the Lighthouse metric, beware that you could end up with:

A higher Lighthouse score = A poorer overall web site experience.

Pots and kettles

Finally, to console yourself about a ceiling, run Lighthouse on some of Google's own pages, such as the various documentation pages you can find beneath the Lighthouse home page. Even the self appointed authority on scoring a web page doesn't achieve the best scores in all situations.