Server Software — Does It Actually Matter?
Origin TTFB comparison across 20,909 WooCommerce stores by detected server software
Server software matters less than you think. The gap between Apache and Nginx is 11%. Your hosting provider and caching strategy matter more.
2,121 additional sites had unidentifiable server software and are excluded
Most Common Server
Cloudflare
7,913 sites
Fastest Server (Median)
Kinsta
857ms median TTFB
Server Types Found
8
7 with 5+ sites
Sites Tested
20,909
WooCommerce with known server
Median Origin TTFB by Server Software
Sorted fastest to slowest. Only servers with 5+ sites shown.
TTFB Distribution by Server Software
Box shows p25–p75 range, whiskers extend to p10 and p90, line marks median
Detailed Performance Statistics
| Server | Count | % of Total | Median | p25 | p75 | p90 |
|---|---|---|---|---|---|---|
| Caddy (n<5) | 2 | 0.0% | 635ms | 635ms | 820ms | 820ms |
| Kinsta | 1,004 | 4.8% | 857ms | 624ms | 1,225ms | 1,650ms |
| Nginx | 5,879 | 28.1% | 976ms | 610ms | 1,615ms | 2,376ms |
| Cloudflare | 7,913 | 37.8% | 996ms | 656ms | 1,536ms | 2,319ms |
| LiteSpeed | 1,798 | 8.6% | 1,066ms | 637ms | 1,814ms | 2,858ms |
| Apache | 4,236 | 20.3% | 1,083ms | 622ms | 1,789ms | 2,813ms |
| Microsoft IIS | 34 | 0.2% | 1,180ms | 455ms | 2,630ms | 4,106ms |
| OpenResty | 43 | 0.2% | 2,346ms | 995ms | 3,407ms | 5,186ms |
TTFB Pass Rates by Server Software
| Server | Count | % Under 500ms | % Under 800ms |
|---|---|---|---|
| Kinsta | 1,004 | 13.4% | 44.6% |
| Nginx | 5,879 | 16.8% | 38.9% |
| Cloudflare | 7,913 | 13.4% | 36.4% |
| LiteSpeed | 1,798 | 15.7% | 35.9% |
| Apache | 4,236 | 18.0% | 35.7% |
| Microsoft IIS | 34 | 26.5% | 35.3% |
| OpenResty | 43 | 7.0% | 16.3% |
Server software is not the full picture
These numbers reflect the median site running each server software, not the server software itself in isolation. Many factors affect TTFB beyond the web server:
- Hosting hardware — CPU, RAM, disk speed, and network quality
- PHP version — PHP 8.x is significantly faster than 7.x
- Plugin count & quality — bloated plugins add server-side processing time
- Database optimization — unoptimized queries, missing indexes, large wp_options autoload
- Object caching — Redis/Memcached vs no persistent object cache
Managed platforms like WP Engine and Kinsta reflect their full stack optimization (custom caching layers, tuned PHP workers, managed databases), not just their choice of web server.
Caching Effectiveness by Server Software
Sites with a confirmed cache HIT on homepage. Shows how much caching reduces TTFB for each server type.
| Server | Sites w/ Cache | Origin Median | Cached Median | Improvement |
|---|---|---|---|---|
| Cloudflare | 2,098 | 976ms | 123ms | 87.0% |
| Apache | 42 | 998ms | 137ms | 84.2% |
| Kinsta | 352 | 883ms | 151ms | 82.9% |
| LiteSpeed | 537 | 1,143ms | 226ms | 81.4% |
| Nginx | 1,418 | 876ms | 215ms | 75.9% |
Methodology
Server detection is based on the Server HTTP response header, supplemented by
platform-specific headers (e.g., x-powered-by) for managed platforms like WP Engine and Kinsta.
LiteSpeed includes both LiteSpeed Enterprise and OpenLiteSpeed — these are
indistinguishable from the Server header alone.
Cloudflare appears as the server when it acts as a reverse proxy. The actual origin server behind Cloudflare is hidden, so Cloudflare results reflect the proxy layer, not the origin software.
Origin TTFB is measured with WooCommerce session cookies set
(woocommerce_items_in_cart, woocommerce_cart_hash, wp_woocommerce_session_*),
which forces the server to bypass page cache and generate a dynamic response. Range: 50–10,000ms.
Sample threshold: Servers with fewer than 5 sites are shown in the detailed table but excluded from charts and pass rate analysis to avoid misleading conclusions from small samples.
Copy this code to embed the report on any website:
