TLS 1.0 hangover by edge provider

Per-customer toggles make edge identity an imperfect predictor.

web average = 34.9%cloudflare: 60.2; +25.3pp vs web average = 34.9% (worse); n=419551 cloudflare +25.3ppgoogle: 57.1; +22.2pp vs web average = 34.9% (worse); n=6802 google +22.2ppother-cdn: 26.2; −8.7pp vs web average = 34.9% (better); n=31530 other-cdn −8.7ppcloudfront: 18.5; −16.4pp vs web average = 34.9% (better); n=5920 cloudfront −16.4ppnone: 15.6; −19.3pp vs web average = 34.9% (better); n=463955 none −19.3ppazure-front-door: 6.6; −28.3pp vs web average = 34.9% (better); n=14773 azure-front-door −28.3ppakamai: 4.1; −30.8pp vs web average = 34.9% (better); n=15946 akamai −30.8ppfastly: 0.5; −34.4pp vs web average = 34.9% (better); n=14964 fastly −34.4pp
Better than benchmarkWorse than benchmark

Denominator: deep-scanned websites · n per cohort in the table

fastly has the least TLS 1.0 at 0.5%; cloudflare the most at 60.2%.

View as table
SliceSharevs benchmarkn
cloudflare60.2%+25.3pp419,551
google57.1%+22.2pp6,802
other-cdn26.2%−8.7pp31,530
cloudfront18.5%−16.4pp5,920
none15.6%−19.3pp463,955
azure-front-door6.6%−28.3pp14,773
akamai4.1%−30.8pp15,946
fastly0.5%−34.4pp14,964

Source: GeoTrust Research — The State of SSL/TLS, October 2026