Page Speed and Conversions: The Numbers That Actually Matter
Slow pages cost money. This isn't a vague claim — there are specific, documented numbers from large-scale studies that make the business case precisely.
Google's research on mobile page speed found that 53% of mobile site visitors abandon a page that takes longer than three seconds to load. Deloitte's research commissioned by Google across five retail and travel brands found that a 0.1-second improvement in page load time improved conversion rates by 8% on average.
Pinch those numbers together: half your mobile visitors are leaving before your page loads, and the ones who stay convert measurably worse for every additional second. The compounding effect on revenue is significant.
Where page speed actually affects SEO
Google made page speed an explicit ranking factor for mobile search in 2018 and codified it through the Core Web Vitals update in 2021. The three metrics that matter:
LCP (Largest Contentful Paint) — how long until the largest visible element loads. Target: under 2.5 seconds. This is the metric most directly correlated with perceived load time from a user perspective.
FID (First Input Delay) / INP (Interaction to Next Paint) — how quickly the page responds to user interaction. FID was replaced by INP in March 2024 as the official Core Web Vital. Target: under 200ms. Poor INP usually indicates heavy JavaScript execution blocking the main thread.
CLS (Cumulative Layout Shift) — how much the page layout shifts after initial load. Target: under 0.1. The frustrating experience of clicking a button and having something else jump in front of it — that's CLS.
Google's Search Console Performance report shows your Core Web Vitals status across real user data, categorized as Good, Needs Improvement, or Poor. This is the right place to start — it shows actual user experience, not just lab measurements.
Where the biggest gains come from
Most page speed improvements fall into a few categories, and not all categories are equal in impact:
| Optimization | Typical impact | Difficulty |
|---|---|---|
| Image optimization (format, compression, sizing) | High | Low |
| Removing render-blocking resources | High | Medium |
| Enabling browser caching | Medium | Low |
| Using a CDN | Medium | Low |
| Reducing JavaScript bundle size | High | Medium–High |
| Improving server response time (TTFB) | High if slow | Depends on hosting |
| Fixing layout shift (CLS) | Medium for SEO | Medium |
Images are the fastest win. Most sites have images that are 3–5x larger than they need to be for their display size, in formats that compress worse than modern alternatives. Converting images to WebP or AVIF, serving correctly sized versions for different devices, and lazy loading images below the fold typically produces the largest LCP improvement for the least development time.
Cloudinary's image analysis data consistently shows that images account for 50–75% of total page weight on average. Solving the image problem solves most of the load time problem.
JavaScript is the complex win. Render-blocking scripts, large bundle sizes, and poorly timed script execution are the most common causes of poor INP and slow Time to Interactive. The fix requires auditing what's loading, when, and whether it's necessary — which often requires a developer and a prioritization conversation about which third-party scripts are actually producing ROI. For the full prioritized fix sequence, see Website Load Time Optimization.
How to measure before you optimize
Optimizing without measuring first produces random results. The tooling to use:
| Tool | Best for | Access |
|---|---|---|
| Google Search Console | Real user Core Web Vitals data by page | Free, requires site ownership |
| PageSpeed Insights | Lab + real-world data for any URL | Free, no login required |
| WebPageTest | Detailed waterfall charts, multi-location testing | Free |
| Lighthouse | Full audit in Chrome DevTools | Built into Chrome |
| GTmetrix | Consolidated performance report with recommendations | Free tier available |
Start with Search Console for the aggregate picture, then use PageSpeed Insights or WebPageTest on your highest-traffic pages to identify specific issues. Fix the highest-impact issues on the pages that matter most — your homepage, product pages, checkout flow — rather than trying to optimize everything at once.
The ongoing maintenance problem
Page speed isn't a one-time fix. Every new marketing tag, every new JavaScript library, every new image uploaded by a non-technical team member can degrade performance over time. Sites that score well today without monitoring will drift over months.
Practical maintenance approaches:
- Add Lighthouse CI or a similar tool to your deployment pipeline to flag speed regressions before they reach production
- Audit third-party scripts quarterly — each marketing tool added to the site adds load time, and tools accumulate over time
- Set a performance budget: a defined maximum acceptable page weight, and a process for evaluating additions against it
Web.dev's performance measurement guide is the most current reference for testing methodology and interpreting results.
Page speed improvements tend to have a compounding effect — better scores → better rankings → more traffic → more conversions.
But the relationship between specific changes and specific outcomes needs to be measured, not assumed.
If you're not sure where your biggest gains are, we can audit it →
