Prompt experimentation is (finally) fulfilling the promise of web experimentation
More than a decade ago, tools like Optimizely and VWO promised a revolution.
Visual editors for A/B testing. No code. No developers. Just results.
Their big idea: if you could edit a site visually, you could optimize it.
And for a moment, it looked like they might change everything.
But the web evolved. SPAs, dynamic personalization, and complex frontend frameworks made those visual editors fragile and unreliable.
Tests broke. Targeting was limited. Measurement got fuzzy.
And trust? It disappeared.
Many teams gave up on experimentation.
Or ran the safest, simplest tests they could manage.
As graphic editors proved too brittle for modern websites, tests that should have been launched easily on the client side were pushed to already-overloaded frontend or fullstack developers.
This shift created tension and helped fuel the rise of server-side testing and feature experimentation.
Teams embedded tests inside feature flags and backend workflows not because it was the best approach, but because the WYSIWYG editors failed them.
It kept testing alive but made it slower, more technical, and less accessible.
Experimentation was no longer available to the people closest to the customer.
It was locked inside sprint cycles and developer queues.
But the belief in test and learn never went away.
Experimentation-Led Growth (ELG) continued, because companies that experiment learn faster.
And companies that learn faster, grow faster.
Today, 9 out of 10 companies that call themselves category leaders and plan to significantly grow in 2025 run experiments.
The promise wasn’t wrong.
The tools were.
The next generation of testing tools won’t be visual editors or bloated workflows.
They will be prompt-based, flexible, and aligned to how modern teams actually work.
Get invited to the trial and see it for yourself
Begin your free trial for Prompt-Based Experimentation. Try it on your own site. Build experiments by chatting with AI. Explore a new way to interact with an experimentation platform.
