Experiment types & delivery
Optimizely
The plans page lists A/B, multi-page, feature and server-side testing with real-time personalisation and audience targeting, which is more than client-side splits. But every claim sits on a feature grid: I found no public information on documented SDKs, multivariate tests, mutually exclusive groups or holdouts, so the delivery mechanics cannot be verified from what was captured.
VWO
The captured pricing page shows A/B/n, split and multivariate testing with a visual and code editor, multi-armed bandit campaigns, mutually exclusive groups and SPA support, plus personalisation and feature management as product suites. What decides the score against it: the evidence is a feature grid, and we found no public information on server-side SDKs, edge delivery, holdouts or personalisation tested against a control.