Experimentation News: June 2026

This fortnight the bigger story is measurement plumbing rather than new test types. A Google consent change quietly alters how your conversion data reaches Ads, two platforms shipped EU data-residency features, and Statsig spelled out what its Amplitude tie-up actually delivers. Here is what changed and whether it touches your programme.

Google Consent Mode change rewires how experiment conversions reach Ads

From 15 June 2026, the ad_storage parameter in Consent Mode became the single control over advertising data flowing from Google Analytics to linked Google Ads accounts. The Google Signals setting, which some teams switched off as a privacy lever, is now narrowed to behavioural reporting inside GA4 and no longer gates that data flow. For anyone running experiments, the catch is downstream: if your consent banner misfires or ad_storage is set wrong on some pages, the conversions and audiences feeding your tests and your Ads campaigns can go dark without an obvious error. Audit how your consent management platform sets ad_storage across pages and regions before you trust post-15-June conversion numbers in a test readout. The detail is in UniConsent’s breakdown of the change, and the consent rules behind it are covered in our conversion rate optimisation guide.

Optimizely Web Experimentation adds EU events export and clearer bandit results

In its 18 June release notes, Optimizely expanded Experimentation Events Export to EU-based accounts, so European customers can now push raw experiment event data to Google Cloud Storage and keep it in-region. The same update improved the Contextual Bandit results page to show attribute importance per variation and relative improvement per variation against the holdback group, which makes it easier to see why a bandit is favouring one variant rather than just that it is. The EU export is the practical win: data-residency gaps have kept Optimizely off some European shortlists, and raw event access in-region removes a procurement objection. If you are comparing the wider suite, our Optimizely alternatives guide covers where it sits. The changes are in Optimizely’s Web Experimentation release notes.

Statsig spells out its Amplitude integration roadmap

On 17 June Statsig published the detail behind its Amplitude tie-up, moving past the ownership headlines into specifics. Over the coming months Amplitude events and cohorts will work inside Statsig, experiment results will surface in Amplitude dashboards, and Amplitude’s data pipeline is being wired into Statsig’s stats engine for speed. The team also flagged MCP governance updates already shipped, with release pipelines for experiments and AI experiment reports on the way. For current Statsig customers this is the reassurance the takeover left missing: a concrete plan rather than a question mark, though the value depends on those features landing as described. It is a useful counterpoint when you weigh a focused tool like VWO against Optimizely versus an analytics-led suite. Statsig’s Phase 1 post has the specifics.

PostHog tightens replay privacy and ships per-variant session summaries

PostHog’s late-June updates leaned on privacy and experiment analysis. A new disable_capture_url_hashes option strips URL fragments from captured URLs and ships on by default from 25 June, which is a sensible privacy step but a breaking change for hash-based single-page-app routing, so check your replay and analytics URLs still group correctly. Alongside that, PostHog AI can now summarise session replays directly from the Experiments interface, grouping recordings by variant and flagging behavioural differences between them so you can read the why behind a result without watching every recording. For a low-traffic programme, that qualitative layer often surfaces a fix faster than waiting for significance. The release detail is in PostHog’s June changelog.