Website speed optimization for publishers

We make publisher websites measurably faster — Core Web Vitals in the green — and put the budgets and monitoring in place to keep them there.

For publishers, speed is revenue

A slow news site bleeds money three ways at once. Search rankings: Google's page-experience signals directly penalise slow article pages, and news SEO is won on the margins. Advertising: every second of load time cuts ad viewability — the metric your CPMs are priced on. Subscriptions: a reader who waits four seconds for an article does not start a checkout. Speed is not a technical nicety for publishers; it is a line item in three P&Ls.

What we measure — and what we ignore

We optimise the metrics that map to money: LCP under 2.0 seconds on 4G, INP under 200 milliseconds, CLS near zero, all measured on real-user data, not lab scores alone. Lighthouse scores are a debugging tool, not a KPI — a perfect lab score on a fast MacBook tells you nothing about a reader on a three-year-old Android phone on a train.

How an optimization engagement runs

We start with a two-week audit: real-user monitoring data, a render-profile of your article and front pages, and a third-party script inventory. You get a ranked list of fixes with estimated impact — typically a mix of critical-CSS and font-loading fixes, image pipeline work, third-party script governance, and edge caching. Then we fix, in order of impact per day of work. Most engagements land the largest wins in the first month.

The third-party script problem

The single biggest performance problem at every publisher we audit: tag managers that became a junk drawer. Consent tools, A/B testers, analytics, ad tech, social embeds — each added by a different team, none ever removed, collectively adding seconds. We implement script governance: a documented owner, a performance budget, and a removal process for anything that cannot justify its milliseconds.

Staying fast after we leave

Speed regressions happen at 3pm on a Tuesday when somebody ships a new module. That is why every engagement ends with a performance budget wired into CI and real-user alerts — a regression becomes a failed build or a page to the team, not a quarterly surprise. Our operations tier keeps that watch running permanently.

Want to know where your site stands? Send us your URL — we will tell you the three fixes that matter most.