As we start 2026, a lot of teams are still running experimentation programs that look busy but do not move the business in any meaningful way.

• They are shipping tests.
• They are reporting results.
• They are optimizing metrics.

And yet, very little actually changes.

The problem is not effort.
The problem is what teams are optimizing for and how they decide what to test in the first place.

This article outlines an experimentation strategy for 2026 that focuses on intent, learning, and real business impact.

Here is what I believe teams should leave behind, and what they should focus on instead.

Leave Behind “Best Practices” Testing

Most experimentation programs are quietly powered by recycled advice.

• Someone posts, “This worked for us.”
• It gets shared.
• It becomes a rule.

Suddenly teams are testing the same things everywhere, without understanding why they were tested in the first place.

• Button colors.
• Badge placements.
• UI tweaks with no real hypothesis behind them.

None of these are inherently bad. The issue is context.

• Why was that change tested?
• What problem was it meant to solve?
• What metric actually mattered to that business?

Those questions rarely get answered.

If you take one thing away, it should be this:
You cannot borrow learning. You have to earn it inside the realities of your own business.

Many teams still rely on so-called “best practices,” even though best practices can be a misnomer— they’re often usurped by better, emerging approaches that fit unique business contexts better than rigid rules.

Stop Worshipping Conversion Rate

Conversion rate still dominates how many teams measure success.

I understand why. It is simple. It is familiar. It is easy to report.

But conversion rate is a diagnostic metric, not a strategy.

• It tells you something happened.
• It does not tell you what to do next.

You can create a 100 percent conversion rate tomorrow by making everything free. You will also put yourself out of business immediately.

Revenue per visitor is a better lens.
Profit per visitor would be even better, though harder.

Conversion rate should be treated like taking your temperature. Useful, but not actionable on its own.

If experimentation is meant to help the business grow, the goal should be learning that leads to revenue, not small percentage movements in isolation.

Move From Testing to Experimentation

Testing implies A versus B.

Experimentation implies learning.

That distinction matters.

A test often answers a narrow question.
An experiment opens up a line of inquiry.

The most valuable experiments rarely deliver a clean “winner.” They surface something unexpected. A pattern in behavior. A drop-off that points to a deeper issue. A moment of hesitation you did not know existed.

Those are the insights that compound.

When teams shift their mindset from “What can we test?” to “What are we trying to learn?” the quality of ideas changes immediately. This shift is foundational to a modern experimentation strategy in 2026.

Prioritize Intent Over Identity in Experimentation

Personalization is important.
One-to-one personalization is not where most teams should start.

In retail especially, most visitors are anonymous. They do not log in. They may purchase once or twice a year. You often do not know who they are until checkout.

That means teams spend enormous effort personalizing for a small percentage of traffic.

Meanwhile, the other 90 plus percent is sending signals constantly.

• Where they came from.
• What they viewed.
• How deeply they explored.
• What category they focused on.

That is intent.

• Intent is observable.
• Intent is actionable.
• Intent scales.

One-to-one personalization can add value later, but it should not be the foundation.

Design for How People Buy, Not Just What You Sell

One of the simplest and most effective shifts we made was starting with behavior.

We looked at how many pages someone viewed. Nothing fancy.

When we mapped page depth to conversion, clear patterns emerged.

Low page depth correlated with low intent.
Higher page depth correlated with readiness to decide.

From there, the experience changed.

Low-intent visitors needed help engaging.
High-intent visitors needed help deciding.

• Same site.
• Same pages.
• Different goals.

Layering in product intent made this even clearer. Someone shopping for an engagement ring behaves very differently from someone buying a necklace. The timeline, emotion, and decision process are not the same.

Once you design around how people buy, experimentation opportunities appear everywhere.

The Real Shift for Experimentation Strategy in 2026

If teams want to make experimentation matter again, the shift is not about tools or tactics.

It is about focus.

• Leave behind borrowed best practices.
• Stop optimizing single metrics in isolation.
• Use experimentation to learn, not just to win tests.
• Design around intent and behavior.

When teams do that, experimentation stops being a reporting exercise and starts becoming a growth engine.

The strongest experimentation strategy for 2026 prioritizes clarity, intent, and compounding insight over test volume.

This is the same approach we apply across our experimentation work with eCommerce teams.