For years, we've applied a proven methodology to optimize conversion rates:
- Identify a friction point (through data analysis)
- Form a hypothesis
- Design an A/B test
- Wait two weeks
- Measure performance
… and start over :)
It works very well. It's also long-term, foundational work, and sometimes hard to scale.
Open any media outlet, from LinkedIn to Le Monde, and you'll be reminded that AI is upending everything in its path.
CRO is no exception. For some time now, it's been possible to generate variants for our tests, analyze traffic dynamically and run continuous experimentation loops, often faster than a new team ever could.
In short: AI.
The question every e-commerce leader should ask isn't whether AI will change CRO (spoiler: it will!). It's which part of the process to hand to the machine, and what to keep in human hands.
The problem AI solves
One of the biggest barriers to CRO experimentation has always been cost, not just in tools but also in time and people.
Client-side testing requires dedicated UX designers, developers and analysts to interpret the results. For many mid-sized e-commerce companies, that means running a handful of tests per quarter instead of per month.
Unsurprisingly, AI lowers that barrier considerably. The most telling example for e-merchants is probably what's new in the latest Winter '26 Edition:
Sidekick
The AI assistant built into the Shopify admin has gone from a simple chatbot to a true sparring partner. It continuously analyzes your store data and market trends to make personalized recommendations… and edits your theme directly from plain-language instructions:
round this button's corners, change the color of this banner
Rollouts
The missing piece is finally built into Shopify, removing the need for third-party tools for simple tests (though we're still team Intelligems). You can now schedule and run A/B tests to turn your hunches into data-backed decisions.
SimGym
This tool goes even further: you can simulate different scenarios using your store's real data, to project results before you launch a test.
Young brands now get optimization insights that used to be out of reach. CRO consultants, meanwhile… 🙀
Editorial photograph for L'Officiel by creative studio LTI Creative
What AI handles well and what it doesn't
Scared-face emoji, yes, but with some nuance. AI excels at time-consuming optimizations: testing button colors, tweaking headlines, moving images around, optimizing form fields.
These "small bets" matter in the race for cumulative gains, but they will never lead to major discoveries.
Experienced CRO teams need to focus on strategy more than ever, not on these micro-optimizations.
High-impact work, like spotting a failing UX, rethinking a journey from start to finish or clarifying reassurance messaging, takes creativity, sensitivity and an understanding of context.
🤖 → 🙀
Once a team has validated a strategic direction, say rebuilding the product page around UGC, AI can take over the iterative optimization of that new design.
The risk of losing insights
A legitimate concern comes up every time automation enters the picture: if the machine runs the tests, do we lose the learnings?
A/B testing doesn't just produce winners and losers. It's first and foremost a set of insights into why users behave the way they do.
A failed test showing that customers don't trust a new layout teaches you something valuable about your audience.
An AI that quietly runs variants and picks the best performer might never surface that signal.
That doesn't mean avoiding AI-driven testing. The solution is feedback: CRO teams need visibility into what the machine changes, so they can understand what works and which patterns emerge.
Tomorrow, the best setups will combine AI execution with human review cycles, turning automation into a source of learning rather than a magic black box.
But do humans still visit your site?
While we debate whether AI will replace our expertise, a more fundamental shift is underway: AI agents are starting to shop on users' behalf, potentially bypassing your website altogether.
All the major players have launched experiments in agentic commerce, and some predictions say that by 2030, one in two buyers will actually be… a robot!
For e-commerce leaders, the question is almost provocative: if an AI agent compares my prices, scans my reviews and completes purchases without ever loading my product page… what exactly am I optimizing?
If an AI agent compares my prices, scans my reviews and completes purchases without ever loading my product page… what exactly am I optimizing?
Answer Engine Optimization (AEO), which means structuring your product data so AI agents can find, evaluate and recommend it, is becoming as important as traditional on-site optimization.
Your conversion funnel may increasingly start with a machine reading your structured data, not a human landing on your homepage.
Learn more:
Jérémie Werner
Don't miss the opportunity (what to do right now)
If you run an e-commerce CRO strategy, here are four things I'd try soon, in my humble opinion:
- Hand off the iteration: let AI handle ongoing optimization of basic elements. The ROI is immediate and the risk is low.
- Protect time for strategy: use the freed-up hours to invest in research, UX mapping and competitive analysis. You need to understand what your competitors haven't solved yet.
- Build in feedback: don't let AI become a black box. Schedule review sessions to identify what was done and which patterns emerged.
- Get ready for agentic commerce: I tell you everything (really everything) in this article!
- Make room for experimentation: try AI tools and fail. The companies that adapt fastest won't be the ones with the best tech. They'll be the ones with curious teams and leaders willing to trust.
The question isn't whether AI will take over CRO. In part, it already has.
The real question is whether your team is ready to let go of the small details and focus on the work that truly moves things forward.
Well played 💐
