Core Web Vitals made our clients care about performance — finally
For years, our performance pitch went like this: slow websites lose users, slow websites hurt your SEO, slow websites cost you money. We had the data. We had the case studies. We had the screenshots from PageSpeed Insights showing load times that would embarrass anyone. Clients nodded, said they understood, and then asked us to add another plugin or a new hero section with a 4MB background video.
Then Google announced Core Web Vitals as a ranking factor and something shifted. We started getting inbound calls about "that Google speed thing." Clients who had said no to performance audits for two years suddenly wanted one immediately. The difference was not that our arguments got better. The difference was that Google said it.
What we actually did with the opportunity
We ran a performance audit sprint in early 2022 across six client sites. The findings were consistent and grim. Unoptimized images were the biggest culprit on every single site. Multiple render-blocking scripts from third-party tools, analytics, chat widgets, ad pixels. WooCommerce stores loading 3MB of JavaScript on every page. Themes built for visual demos, not actual users.
The fixes we prioritized, in rough order of impact: serve images in WebP with proper lazy loading, defer non-critical JavaScript, implement proper caching at the server layer, audit and remove unused plugins, move to a faster hosting stack where the client was on shared hosting.
We also built a process around this that we call our "Core Web Vitals baseline." Every new client engagement starts with a measurement pass. LCP, FID, CLS, TTFB, all recorded. We establish the baseline before we touch anything, and we re-measure after every significant change. That sounds obvious. It was not standard practice before this.
The part that surprised us
Improving performance had conversion effects we could now actually demonstrate. One retail client went from a 5.8-second LCP on mobile to 1.9 seconds. Their mobile conversion rate improved by 18 percent over the next two months. That is not a controlled experiment. There were other changes happening too. But the direction was clear and we had the data to show it.
The frustrating thing is that we had been right all along. Performance matters. But being right is not enough when you are talking to someone whose mental model of "website quality" is how it looks, not how it loads. Sometimes you need Google to say something before the client believes you. We have made peace with that.