How Fast Does Your Website Actually Need to Be?
Your website's main content should appear in under 2.5 seconds on a mid-range mobile phone. That's Google's threshold for a "good" Largest Contentful Paint, the moment the biggest piece of content on the page finishes loading. Pages slower than that start losing visitors, and the cost is well documented.
How fast is fast enough?
Google tracks real-world speed with three metrics. Largest Contentful Paint (LCP) matters most for first impressions. It measures how long your main content takes to appear:
- Good: 2.5 seconds or less
- Needs improvement: 2.5 to 4 seconds
- Poor: over 4 seconds
Interaction to Next Paint (INP) should be 200 milliseconds or less so buttons and menus respond without delay. Cumulative Layout Shift (CLS) should stay at 0.1 or less so the page doesn't jump around while it loads (Google, Core Web Vitals).
Measure them on a phone. A page that feels quick on your office laptop can miss these thresholds on a mid-range phone, and that's where most sites fall short.
What does a slow site cost?
Google and Deloitte analysed 37 brands across more than 30 million mobile sessions. A 0.1-second improvement in mobile load speed lifted retail conversions by 8.4% and average order value by 9.2%. Travel conversions rose by 10.1% (Deloitte/Google, 2020).
A 0.1-second mobile speed gain lifted retail conversions by 8.4% and average order value by 9.2% (Deloitte/Google, 2020).
Load time also drives bounce. As load time goes from 1 second to 3 seconds, the probability that a visitor leaves without doing anything rises by 32%. From 1 second to 5 seconds, it rises by 90% (Google/SOASTA mobile benchmarks). Five seconds may not feel like a long wait, yet it nearly doubles the chance the visitor is gone.
Portent analysed more than 100 million page views. Sites that loaded in 1 second had conversion rates around 3 times higher for B2B and 2.5 times higher for e-commerce than sites that took 5 seconds.
Is this just correlation, or does speed really cause it?
Most speed studies only show that fast sites and high conversion rates go together. The strongest causal evidence comes from Vodafone's controlled A/B test on its live site. Vodafone improved LCP by 31%, from 8.3 seconds to 5.7, and changed nothing else. Sales rose 8% and the lead-to-visit rate rose 15% (Vodafone, via web.dev). With the same site and traffic, the difference in sales comes down to speed.
The effect isn't linear. Cutting load time from 8 seconds to 6 is worth much more than cutting it from 2 seconds to 1.5. The steep losses happen above roughly 3 seconds, so the 2.5-second target leaves some headroom.
Why is "fast on my laptop" a trap?
Because your own device on office Wi-Fi is nowhere near what a typical visitor has. The realistic global baseline is a mid-range phone with about 7.2 Mbps download speed and 94ms of network latency. That phone runs roughly 4 to 5 times slower than a flagship (Alex Russell, "The Performance Inequality Gap," 2024). What feels instant to you can be slow for the person you're trying to win.
Most of the damage comes from what gets added to the page. Third-party resources appear on 92% of web pages, and the median page pulls in dozens of them (HTTP Archive Web Almanac, 2024). Chat widgets, analytics tags, font services, tracking pixels and embedded videos each add weight and delay. The same report found that the median mobile site's main thread is blocked for 1,209 milliseconds, six times the recommended limit, mostly by third-party scripts. Keeping that critical path clear is part of how we build websites.
Does speed help you get found, too?
Yes, in two ways. Google has used Core Web Vitals as a ranking signal since 2021, and they work mainly as a tie-breaker between otherwise comparable pages (Google Search Central).
The second way is newer. AI answer engines such as ChatGPT, Perplexity and Google's AI Overviews increasingly send people to websites. They favour pages that load quickly and are easy to crawl. Content structure matters too: peer-reviewed work from Princeton and the Allen Institute for AI found that well-structured, citable content can lift a source's visibility in AI-generated answers by up to 40% (Aggarwal et al., KDD 2024).
FAQ
What is a good website load time in 2026?
Aim for your main content to appear in 2.5 seconds or less on a mid-range mobile phone, measured on a real-world connection (Google, Core Web Vitals). Under 2 seconds is better. In Portent's study, the best-converting sites loaded in around 1 second.
How much does a one-second delay cost?
It depends on where you start, but the effect is large. Bounce probability rises 90% as load time goes from 1 to 5 seconds (Google/SOASTA). A 0.1-second improvement lifted retail conversions by 8.4% in a study of more than 30 million sessions (Deloitte/Google, 2020).
Why is my site fast for me but slow for visitors?
You're probably testing on a fast device with a strong connection. The realistic baseline is a mid-range phone, about 4 to 5 times slower than a flagship, on a 7.2 Mbps mobile network (Alex Russell, 2024). Test under those conditions.
What slows a website down the most?
Usually the extras. Heavy images, web fonts, chat widgets and tracking scripts do most of the damage. Third-party scripts block the median mobile site's main thread for over 1.2 seconds (HTTP Archive, 2024).
We measure every site we build on real phones against the LCP, INP and CLS thresholds above. Book a call to see how your current site does on the same test.