What happened to the experimentation market, 2023–2026 What happened to the experimentation market, 2023–2026

Between September 2023 and July 2026, the A/B-testing market consolidated faster than its documentation did. Google killed Optimize and named three successors — AB Tasty, Optimizely and VWO. Since then, two of those three have agreed to become one company under one brand. Of the developer-era tools that rose to replace them, Statsig was acquired by OpenAI and its brand handed to Amplitude, and Eppo was absorbed into Datadog and renamed. Most pages that rank for these tools' names were written before any of this happened.

Here is the record, in order.

The timeline

Two of Google's three named successors are now one company. The third — Optimizely — is the only one still standing alone.

Google Optimize died, and Google never replaced it

Google's stated reason, in its help centre's current wording: Optimize "did not have many of the features and services that our customers request and need for experimentation testing."

What matters for anyone choosing a tool today is what came next: nothing. Google's own [GA4] A/B test help article, live as of 28 July 2026:

"To run an A/B test in Google Analytics, you must integrate with a third-party A/B test tool."

Nearly three years after the sunset, Google Analytics reports on experiments; it does not run them. Google's developer guide for that integration describes itself as "primarily intended for third-party A/B experiment solution providers." Nothing in Google's own documentation suggests this is changing. The deeper history — what the sunset meant for Optimize's users and where its refugees went — is covered on our Google Optimize alternatives page.

The three named successors, three years on

Optimizely remains independent, and remains the category's reference point for statistical method — its docs publish bandit mechanics down to the algorithm and update cadence, which no other vendor in this set does. It has never published a price: its plans page says every plan is "individually packaged," and the free tier, Rollouts, runs "one concurrent experiment at a time." Our full evaluation is here.

VWO and AB Tasty are becoming Wingify, per the timeline above. Both companies emphasise continuity — VWO's migration FAQ states "Your existing contract, subscription terms, and pricing remain unchanged," and no end-of-life has been announced for either platform. The integration of engineering, data models and workflows is, by their own description, a multi-quarter programme in progress. We've written the two sides separately, because the questions differ: the VWO side and the AB Tasty side.

An observation this page can make without arguing: all three vendors Google endorsed price by sales call. None of the three publishes a number.

The developer wave got acquired too

The tools built after Optimize's era — warehouse-native, SDK-first, priced in public — were the market's answer to enterprise quote-only selling. Two of the most prominent are now inside larger platforms:

• Eppo → Datadog (5 May 2025). Now branded Datadog Experiments — experimentation as a feature of an observability platform.
• Statsig → OpenAI (2 Sep 2025), with the brand and customers passing to Amplitude (5 May 2026) — the founding team building at OpenAI while Amplitude runs the platform. Experimentation as a feature of a product-analytics platform.

Neither product died; both are actively maintained by their new owners, per those owners' own announcements. But a buyer bookmarking a vendor in this category should assume the vendor's owner — and therefore its roadmap — can change between renewal cycles, because since 2023 it usually has.

The pricing picture, vendor by vendor

Read on 26 July 2026, each from the vendor's own pricing page. This table is the current state of the market Google's sunset created.

The pattern is visible without editorialising: the three vendors Google endorsed all price by sales call; the vendors that publish numbers are the newer, developer-shaped ones. A team that loved Optimize because it was free and self-serve is, three years later, choosing between enterprise proposals and engineering tools.

Why this page will change

Three of the events above are processes, not moments, and this page inherits their deadlines:

• The Wingify integration is scheduled "over several quarters" — product names, plan structures and documentation URLs are still moving.
• The Statsig-Amplitude connection was announced as a four-month data-layer plan in June 2026.
• Datadog Experiments continues to absorb Eppo's surface — geteppo.com already redirects its identity.

We re-read the primary sources and re-date this page when they move. If you're citing it, cite the review date at the top.

Where Mtrix sits in this — interest declared

Mtrix sells experimentation too, so read this section knowing that. The short version: Mtrix is an all-in-one platform for DTC brands — experiments, session replay, analytics, heatmaps, error tracking and the store itself in one product, with server-side bucketing on every install and no cap on concurrent or monthly experiments on any plan. It is not free — one month free, no credit card, then paid — and it is not in the tables above because this page is a record, not a shortlist.

If the consolidation above cost you a tool, or is about to: start your free month and run a real test before anyone gets you on a call.

FAQ