A/B Testing and CRO News: June 2026

The theme this fortnight is AI moving from a marketing badge to actual workflow plumbing. Optimizely shipped agents that decide what to test next and connectors that let you run experiments from a chat window, while the Amplitude and Statsig handover is starting to worry buyers about who owns the roadmap. Here is what changed and whether it should change your plans.

Optimizely Opal adds agents that prioritise your backlog

In its 4 June update, Optimizely added three experimentation agents to Opal, its built-in AI assistant. Experiment Backlog Prioritisation ranks your queued ideas so the team stops arguing about what to test next, the Experiment Value Estimator projects the annualised revenue impact of rolling out a winner before you commit, and the Experiment Program Health Review generates a dashboard summarising programme performance over any timeframe you pick. For a CRO lead, the value estimator is the one to watch: it targets the recurring problem of proving programme worth to a finance team that wants numbers, not test counts. These are aimed at existing Optimizely customers, so they are a retention feature rather than a reason to switch from your current A/B testing tool. The full list is in Optimizely’s June Opal update.

Optimizely’s MCP server brings experimentation into Claude and ChatGPT

Optimizely also released a remote Model Context Protocol server that connects its Web and Feature Experimentation data to AI clients including Claude, ChatGPT and Cursor. The practical effect: a product manager can create flags, configure tests and pull results in plain language without touching an API or a code editor, and the server inherits your existing Optimizely permissions so people can only do what the UI already lets them do. It is available to all Experimentation customers at no extra charge and authenticates over OAuth with your Opti ID. If you have been weighing the wider Optimizely suite against rivals, this kind of AI-native access is becoming a differentiator worth factoring into an Optimizely alternatives shortlist. CMSWire has the launch write-up.

Amplitude’s Statsig takeover leaves roadmap questions for buyers

The dust is settling on the Statsig handover, and the picture is clearer than the headlines suggested: OpenAI acquired Statsig and kept much of the experimentation engineering and product leadership in-house, while Amplitude inherited the platform, brand and customer relationships. For anyone running or shortlisting Statsig, that split is the catch. With the people who built the roadmap now sitting at OpenAI, current customers are right to ask who drives feature direction and how fast innovation continues under new ownership. The broader signal is that standalone experimentation tools keep getting absorbed into larger analytics and product platforms, which is worth remembering when you compare a focused tool like VWO against Optimizely versus a wider suite. VWO’s team published a useful breakdown of what the deal means.