Did I just purchase a $100 phone? Absolutely! to see how badly many websites perform in the wild. If your site is only fast on a flagship device, you're excluding millions of users. Stop designing for luxury! Emulators lie, low-cost hardware shows the truth.
Riccardo Margiotta
@riccardomargiotta.co.uk
🖥️ Lead Engineer (Digital Product Ops) at Arnold Clark ✏️ HTML/CSS/UX/WebPerf obsessive 👦👧👧 Daddy 🍕 Vaguely Italian 🏴 Mostly Scottish 🎮 Mario/Zelda/Halo fan 🤷♂️ Emoji enthusiast
performance.now() 2026 will take place on 19th and 20th of November 2026. #perfnow.
Super talk from @tunetheweb.com on Speculation Rules. Been nervous about using them for fear of false positives on things like Google Analytics. The upcoming prerender_until_script logic sounds like exactly what I need - you'll benefit from fetching subresources, but won't execute scripts. #PerfNow
A wise man once said... "Simple to use, progressive enhancement, built in APIs are the best!" - @tunetheweb.com Yep, he's onto something. loading=lazy and fetchpriority=high are wonderfully simple 'cheat codes', and some of my favourite additions to the world of #webperf. #PerfNow
@mmocny.com talking about some super cool new measurements coming to Chrome - such as the equivalent of LCP but for 'soft' navigations like those in SPAs. That has the potential to start measuring a huge blind spot affecting a very significant portion of the web. #PerfNow #webperf
Be aware of "survivor bias" in your metrics. Users having a fast experience will view more pages and have longer sessions. Your RUM data is always inherently biased towards those with fast experiences. @andydavies.me #PerfNow #webperf
Cross-browser testing is an act of user advocacy. As much as Chrome has pushed the web forward, if that's all we test we could be missing big gotchas. Meet users wherever they are, no matter the scenario. @timkadlec.bsky.social is crushing it, as the kids are saying these days. #PerfNow #webperf
Be the interpreter for your non-technical stakeholders. Milliseconds don't mean much to most people, make the data relatable. "Now they get a smoother experience that's over 50% faster." @ethangardner.com on making performance allies. #PerfNow #webperf
So many of the optimisations we make for web performance make for a more sustainable web. So by improving your LCP or removing that unused legacy JavaScript, you're helping to save the planet. 🙂 @inesakrap.bsky.social #PerfNow #webperf
The coffee bar has started pre-brewing to reduce queuing time. That's real life speculation rules in practice! #PerfNow
An incredible live demo (very brave) from @michael-hladky.bsky.social showing the performance impact of the contain and content-visibility CSS properties on layout calculation and paint. These have been on my TODO list for a long time, I'm feeling braver to experiment now! #PerfNow #webperf
Just like web performance work, accessibility winds up being less expensive the earlier you implement it. Build accessibility in from the start, rather than bolting it on - make it part of the process. @marcysutton.com #webperf #PerfNow #a11y
From a paper in the 1980s and still relevant today: "What if we measured performance by the number of keyboard commands it took to complete a task?" I've spent plenty of time tabbing around my individual pages, but need to start considering full user journeys. @marcysutton.com #webperf #PerfNow
In #webperf, we over-focus on worst case scenarios, particularly cold starts. Don't test unlikely scenarios like a checkout page with no basket items, or as a landing page. Pasting URLs into perf tools isn't the same as a real user clicking around. @csswizardry.com #PerfNow
Be aware of synthetic tests using slow defaults. If you're fortunate enough to have users with decently powerful devices and decent connections, overly pessimistic lab results are useless. Pick settings matching at most your P75 - align lab data with real users. @csswizardry.com #webperf #PerfNow
The #webperf industry has settled on P75 as our benchmark. P75 isn't very ambitious. "It's fine for 25% of our customers to have a bad experience, we don't mind." Why not aim for p95? @csswizardry.com #PerfNow
Google's performance thresholds are not your thresholds. Remember that Core Web Vitals are certainly a good starting point, but not the end goal. Don't aim for somebody else's target, correlate with your own real user conversions and set your own goals. @tammyeverts.com #PerfNow #webperf
Our opening #PerfNow keynote is from @tammyeverts.com, as always giving out the best vibes in all of #webperf (and also Halloween candy). "Fast is magical." But how fast is fast enough? If you're not getting any return on your efforts, are you done? Pragmatic vs magic. #Pragmagical
@screenspan.net talks about approaches for making your alerts meaningful, reducing noise and false positives. And @programmingart.bsky.social really commits to the presentation theme. #PerfNow #webperf
"But Andrea, the synchronous tracking code is secretly hidden deep within Google Tag Manager!" @verlok.bsky.social is calling me out, and I'm not sure I like it. 😅 But some great tips on common INP mistakes. #PerfNow #webperf
In what seems to have become a ritual, I always stop at Black Gold Coffee on the way to my hotel after landing in Amsterdam for #PerfNow. This hipster coffee shop/record store is pushing it a bit even for my beardy face!