From 45 to 100 in Minutes. Or: How To Game A Synthetic Test And Call It Engineering.

Ferdy Korpershoek has a tidy YouTube channel, a calm presenting voice, and a recent video titled 'I Took My WordPress Site From 45 to 100 Speed Score (In Minutes!)'. He seems like a perfectly decent man. He is clearly good at making videos. None of what follows is personal. It is, however, about the genre he is working in — the WordPress speed-up tutorial — which is one of the most reliably misleading formats on the open web, and worth pulling apart slowly.

The video opens with a claim that has been the unkillable opening of WordPress tutorials for fifteen years: 'A slower website means a lower ranking in Google, less visitors, and less business.' True in the abstract, useless in specifics. Google has been clear that Core Web Vitals are a tiebreaker, not a ranking lever — a site that loads in 1.4 seconds versus 1.1 seconds will not move on the results page in any way you can measure. The premise that shaving milliseconds off a synthetic lab test will fix your traffic is the foundational lie of the entire speed-plugin economy.

What PageSpeed Insights actually measures.

The tool used throughout the video is Google PageSpeed Insights, set to mobile by default. The score that goes from 45 to 100 is the Lighthouse Performance score — a weighted composite of First Contentful Paint, Largest Contentful Paint, Total Blocking Time, Cumulative Layout Shift and Speed Index, measured by Google's headless Chrome running a simulated mid-range Android with simulated 4G throttling. It is a lab test. A single run. From a single data centre. Against a single page. With the cache cold or warm depending on which roll of the dice you got.

What it is not: how your site performs for the actual humans hitting it from rural broadband, dodgy hotel wifi, or a five-year-old Samsung in a basement. That data exists — it is called the Chrome User Experience Report, or CrUX, and it sits in the same PageSpeed Insights screen as the 'Field Data' panel. The video does not look at it. Almost no speed-plugin tutorial ever does. Because the field data of a small business site with three hundred monthly visitors is either 'insufficient data' or roughly identical before and after, and that does not make for a triumphant before-and-after thumbnail.

The affiliate link, lightly disclosed.

The plugin being promoted is Airlift. The link in the description is ferdy.com/airlift — a redirect on the creator's own domain that points to airlift.net. That is the textbook signature of an affiliate link: a vanity URL on the creator's site that wraps a tracking parameter so the vendor can attribute the signup back. There is nothing illegal about this. It is also not disclosed in the video. UK ASA rules and US FTC guidelines both require clear, unambiguous disclosure of material connections between a creator and a product they recommend. 'Get Airlift: ferdy.com/airlift' in a description is not that. A spoken 'this video is sponsored by' or an on-screen 'AD' badge is.

Airlift's own pricing — $20 a month, $10 a month if you commit annually, $99.50 for the first year and full whack after that — is shown on screen at the moment the host upgrades. The upsell to Pro happens inside the same video that opened with 'this works for free'. The Pro features listed (dynamic image resizing, font subsetting, stylesheet lazy loading, auto-purge cache on post updates) are the ones that actually do the heavy lifting in the field. The free tier is the on-ramp. The video is the funnel.

What the plugin actually does.

Strip the branding away and Airlift is a perfectly competent member of a crowded category: page caching, WebP conversion, image lazy-loading, font preloading, HTML/CSS/JS minification, critical CSS, and — in the Pro tier — on-the-fly image resizing through a CDN. WP Rocket does this. LiteSpeed Cache does this for free if you are on LiteSpeed hosting. W3 Total Cache has done this since 2009. Hummingbird, Perfmatters, FlyingPress, NitroPack and a dozen others all do this. The reason a fresh WordPress install scores 45 on mobile is that WordPress ships with render-blocking jQuery, an unbounded theme that loads four web fonts, and an editor that emits inline styles. Bolting on a caching plugin to paper over those defaults is a category, not an innovation.

And here is the part no speed-plugin video ever says out loud: the reason a static site generator, an Astro build on Cloudflare Pages, or a TanStack Start app on the edge scores 100 on the first try, with no plugin, no $20-a-month subscription, no dashboard, no email verification, no upsell funnel — is that they were built that way. They emit small HTML, ship minimal JavaScript, cache at the CDN by default, and do not need a plugin to undo seventeen years of architectural debt. WordPress needs the plugin because WordPress is the problem the plugin is selling itself to solve.

“Going from 45 to 100 on a synthetic mobile Lighthouse run is not engineering. It is a magic trick where the rabbit is a CDN you were already paying for.”

What an honest version of this video looks like.

An honest version would say: 'I installed a caching plugin. My Lighthouse score went up. I do not know whether any real visitor noticed. I get a commission if you sign up via this link. The Pro features cost $99.50 for the first year and around $240 a year after that. If you are starting a new site today, you should probably not be on WordPress at all.' That video would get fewer views. It would also be true.

None of this is Ferdy's fault personally. He is working within the conventions of a genre that the WordPress ecosystem has spent two decades training creators to perform — the breathless before-and-after, the gentle nudge to the affiliate link, the cheerful 'let me know your score in the comments' that generates the engagement signal YouTube rewards. He is good at it. We just think the genre itself is rotten, and that the people watching deserve to know what the test actually measures, who gets paid when they click, and what the alternative — a stack that does not need a $20 plugin to be fast — actually looks like in 2026.

Score 100 on PageSpeed. Win nothing. Pay $99.50. Repeat next year.