Blog / Give your CRO testing team the authority to be wrong

Marketing leadership can generate way more revenue this quarter by giving the testing team the authority to make decisions without a committee.

One of the biggest opportunities for almost every client we work with is improving the conversion rate of the traffic they already get. In 2026 there are many ways people get answers without ever reaching your site, including AI overviews, LLMs and a whole host of other things.

So getting people to ‘do what you want them to do’ when they hit your site matters more than ever, and the traffic you already have is the lever that is most inside of your control. If I think of everything we do as an agency, conversion rate optimization is probably the program with the highest return on investment of anything we run. It can deliver tangible results very quickly. The impact lands across every single channel you have, because every channel ends up at the same website.

What gets in the way of CRO testing authority

In big corporate organizations there is often a fear and hesitancy to do anything that might rock the boat, and too many stakeholders involved in every decision. The speed to test is so slow because of the amount of approvals needed. Every single change, even for a short two-week test, has to get through a full committee of people. It drives me crazy because it’s not that business-critical.

And it isn’t only me being impatient about this. McKinsey surveyed 1,259 people on how their organisations actually make decisions. At companies with 1 to 3 reporting layers, 70% thought the decisions coming out were good ones. At companies with 7 or more layers, that fell to 45%. The same survey found the people who described their decision making as fast were roughly twice as likely to say those decisions were high quality too. So the extra layers cost you speed, and on this evidence they don’t win you any quality back. (It’s a 2018 survey, and I’d be amazed if it’s gotten better since.)

The whole point of testing is to run something for a short period of time, on a small slice of your traffic, so you have a business case (did it make us more money?) for the bigger, wider conversation IF it wins. The risk is capped before anyone signs anything, because the time frame and the traffic slice are what keep it small. The people doing the approving often don’t understand that because nobody has ever explained to them. They don’t really need to know.

They experience the task of doing a test as a commitment to changing the product, and they treat it with the caution a product change deserves.

It just slows everything down.

A client we’re working with right now has exactly this issue. It’s taking months and months to get a test live (I wish I was exaggerating). The test itself then runs for two weeks. The site gets such huge volumes of traffic that two weeks is enough for an answer. Then the result lands and the bigger conversation can finally happen.

Why most marketing experimentation tests lose

Having run thousands of tests across hundreds of sites, I can tell you that what feels like a sure thing loses all the time. At Google and Bing, only about 10 to 20% of changes have a positive effect on the metrics they want to move, and at Microsoft around two-thirds of changes turned out to be negative or neutral (Kohavi, Tang and Xu published the figures in Trustworthy Online Controlled Experiments, 2020). So a committee that only lets through the sure things has essentially stopped running tests altogether.

A test you already know the answer to is actually just a launch.

We ran a test for an enterprise client where the change was as close to a sure thing as this work ever gets, a treatment that had won at another company, in the same industry the previous month. It lost painfully by 39% against control. But because it was a test it got canceled within two weeks. The client’s own team said it on the call, “this is why we test”. That was music to my ears, they were starting to get it.

Ship that same change without a test and it sits there draining conversions for months, while everybody hunts for the cause in the ad account.

Empowering marketing leadership in CRO testing

So if you’re a marketing leader and you have a team that’s testing, or an agency doing it for you, the job is to give them the authority to test things quickly, to change things, to see what works, and then to go to the wider organization and remove the levels of approval that shouldn’t be there. The expensive mistake is the sure thing that ships untested, and no committee meeting is looking at that one.

Improving your conversion rate can happen fast. The impact shows up across every channel you have. Doing it as quickly as possible, and getting out of the way so your team actually can, really can transform your business.

So if you take one question into your next meeting, make it this one. How many working days from a test being proposed to that test being live? Ask your team, or ask the agency doing it for you, and don’t settle for a range. The answer tells you how much of your conversion rate is being decided by your approval process rather than your traffic.

The bigger version of this problem is the number itself. The traffic you already have matters more every quarter, which is where this piece started, and the number you report on it to your board has stopped meaning what it used to mean. I’ve written about that one separately.