We built an independent hosting speed lab. Here is the first dataset.
For months we measured hosting speed with borrowed numbers. Now we run our own lab: identical WordPress sites, one on each host, measured the same way from the same place. This is the first published run, and we will refresh it every month.
Most hosting speed numbers you read are second-hand: a screenshot from a testing tool, a figure quoted from another review, a vendor’s own claim. We wanted our own. So we built the HostingDive Speed Index, a benchmark we run ourselves against a row of identical WordPress sites, one on every host we cover.
Each site is a bare WordPress install on the host’s entry plan, with the host’s own caching left exactly as it ships. We measure time to first byte — how quickly the server starts answering — from a single fixed US cloud location, the same place for every host, in one back-to-back run. Because the sites are identical, server response is the part that actually differs, and that is what we are measuring. The full method, including why we lead with time to first byte, is on the methodology page.
The inaugural results
First-party time to first byte, measured from a fixed US cloud vantage. Median is the typical response; p95 captures the slower end of the spread. Sorted fastest median first.
| # | Host | Median TTFB | p95 | As-shipped stack | |
|---|---|---|---|---|---|
| 1 | Kinsta | 68.3 ms | 89.8 ms | Edge Caching + Cloudflare | Visit → |
| 2 | Hostinger | 79.1 ms | 270.6 ms | LiteSpeed + LSCache | Visit → |
| 3 | WP Engine | 111.6 ms | 188.3 ms | EverCache + Cloudflare | Visit → |
| 4 | Cloudways | 120.5 ms | 183.6 ms | nginx + Varnish + Breeze | Visit → |
| 5 | SiteGround | 174.5 ms | 212.2 ms | nginx + SG Optimizer | Visit → |
| 6 | Namecheap (EasyWP) | 199.7 ms | 218.0 ms | nginx + EasyWP cache | Visit → |
| 7 | Bluehost | 314.3 ms | 355.6 ms | Oracle Cloud + WP | Visit → |
| 8 | Scala Hosting | 329.2 ms | 335.9 ms | SPanel | Visit → |
HostingDive Speed Index — inaugural run, June 2026. First-party measured, all eight hosts. Refreshed monthly.
How to read this (and how not to)
This is the start of a series, not a final scoreboard. A single run is a snapshot, and snapshots move around. We are not crowning a fastest WordPress host on the strength of a single run, and neither should you.
What the table does show is real: where each host’s server response landed when we measured them side by side, on the same day, from the same place, with nothing on the sites but WordPress. As the runs stack up month over month, the order that holds is the order worth trusting. Watch the trend, not the first row.
Why these numbers come from a datacenter, not a desk
Here is a result that taught us something. When we first measured our Hostinger probe site from a home internet connection, time to first byte came back around 498 ms. We moved the exact same test to a fixed US cloud vantage and measured the same site again: 79 ms.
Same server. Same page. Same WordPress install. The only thing that changed was where we measured from. A residential connection added hundreds of milliseconds of its own, and none of it was the host’s doing. If we published numbers from a home or office line, we would mostly be ranking our own broadband.
That is why every figure in the Speed Index is taken from one fixed cloud location, identical for all eight hosts. It is the difference between measuring the host and measuring the distance to your couch.
What happens next
We refresh these numbers every month and keep the history. The roster grows as we add probe sites, and the labels stay clear about it: hosts we measure ourselves are marked first-party, and anywhere we lean on outside field data instead, we say so on the spot. We never blend the two into one number.
Bookmark this page if speed is what you are buying for. Come back in a month, and the month after that, and you will see which hosts hold their response time and which ones drift. If you want the full method first, start with how we test hosting speed.
By the HostingDive Team · HostingDive Speed Index, inaugural dataset published June 2026, refreshed monthly.