Page Speed and Revenue:
Why Every 100ms Costs Uzbek E-Commerce Real Money
The relationship between page load time and e-commerce revenue is one of the most consistently replicated findings in web performance research. A Google and Deloitte study found that each 100ms reduction in mobile load time correlates with an 8.4% increase in conversions for retail and 10.1% for travel. For a typical Uzbek e-commerce site hosted abroad, this is not an abstract statistic — it translates directly to a calculable monthly revenue gap.
The Geography Problem for Uzbek E-Commerce
Most Uzbek e-commerce sites are hosted on European or Asian cloud infrastructure — AWS Frankfurt and Singapore are common choices. From a server in Frankfurt, Time to First Byte (TTFB) to a mobile user in Tashkent is 150–250ms under normal network conditions. By the time you add asset loading (product images, CSS, JavaScript), a typical product page takes 3–5 seconds to fully load on a mobile connection. This is the baseline most Uzbek online retailers are operating from, often without realising it because their development and testing happens on fast local connections.
The same application on Hyper App infrastructure in Tashkent produces a TTFB of 8–25ms. Total page load with optimised assets: 0.8–1.5 seconds. That difference — 1,000ms or more of extra load time — has a direct, measurable revenue impact.
Calculating the Revenue Impact
An Akamai study (2017, widely cited and confirmed by subsequent research) quantified the relationship at approximately 1% reduction in conversion rate per 100ms of additional delay. Using conservative inputs for a mid-size Uzbek e-commerce operation: 10,000 monthly orders, $45 average order value, 3% baseline conversion rate. A site with 1,000ms of extra latency (10 × 100ms penalty) experiences approximately 10% lower conversion rate — meaning roughly 1,000 orders per month that do not complete. At $45 average order value, that is $45,000 per month in missed revenue. Even applying a conservative 3% impact instead of 10%, the number is $13,500/month — $162,000/year.
Performance Improvements in Order of Impact
-
1. Move your server to Tashkent — The single highest-impact change. Reduces TTFB from 150-250ms to 8-25ms. Everything else is optimisation on top of this foundation.
-
2. Optimise images — Convert to WebP format (30–35% smaller than JPEG at equivalent quality). Serve images at the actual display dimensions. Use srcset for responsive loading. Images are typically 50–70% of total page weight on e-commerce sites.
-
3. Enable HTTP/2 and GZIP compression — HTTP/2 multiplexing eliminates head-of-line blocking for concurrent asset loading. GZIP reduces text asset sizes by 60–80%. Both are Nginx config changes, not code changes.
-
4. CDN for static assets — Serve CSS, JavaScript, and images from a CDN with edge nodes closer to users. Reduces asset load time and offloads traffic from your origin server.
-
5. Lazy-load below-the-fold content — Images and components not visible in the initial viewport do not need to block page load. The HTML loading="lazy" attribute handles this natively in modern browsers.
How to Measure Your Current Performance
Use WebPageTest.org to measure your site's actual load performance from a user perspective. Set the test location to the nearest available location (Almaty or another Central Asian option if available), and set the connection profile to mobile 4G to simulate Uzbek mobile conditions — which represent the majority of e-commerce traffic in the country. The test gives you TTFB, First Contentful Paint, Largest Contentful Paint, and a waterfall diagram showing exactly which assets are slowest to load.
Run the test from your current infrastructure and record the baseline numbers. Then estimate what each improvement step above would contribute to reducing load time. For most Uzbek e-commerce sites hosted on Frankfurt infrastructure, the infrastructure move alone — step 1 — reduces total load time by 40–60%, before any code or configuration changes.