November 11, 2026 · 4 mins read

The video that cost WeBoost 10% of its sales (and led to a 46% win)

You've done the research and you've got evidence from another page, so you tell the client you're "quietly optimistic”

Then the results land, and sales are down 10%.

The temptation is to bury it. Revert the change and mention "learnings" in the next status call, ideally while nobody is listening. We'd do the opposite.

A big loss is your users shouting at you. They care a lot about something you changed. Work out what, and you've got an insight you can use across the whole site. Here's how one losing test at WeBoost turned into a run of big wins.

The plan that couldn't fail

WeBoost sells mobile phone signal boosters. One of their paid landing pages was doing very nicely, and it opened with a big how-it-works video in the hero.

Their question was a fair one. If it works there, why not put it on the homepage?

Our user research agreed. Lots of users wanted to know how the product works. Which carriers is it compatible with? How hard is it to set up?

So we redesigned the homepage hero, with the video as a secondary CTA.

(Confession: we changed several things at once, so this wasn't the tidy, isolated A/B test we'd normally run.)

With a winning landing page and supportive research behind us, what could go wrong?

It went wrong

Lots of people watched the video, and views went up 150%. Then they left without buying, and sales fell 10%. Then they left without buying and sales fell 10%.

Not the email anyone wants to send a client. But a result that strong, even in the wrong direction, told us users cared about what we'd changed. A lot. So we stopped sulking and started asking why.

Following the clue

One detail kept bugging us: more video views went hand in hand with fewer sales. If that was true on the homepage, could it be true elsewhere?

Which raised an awkward question: where else was the video? Front and centre on every main product page, as it turned out.

So we flipped the experiment around. If more views meant fewer sales, maybe fewer views would mean more.

We demoted the video from the middle of the hero to the thumbnail strip on the left. Its old spot went to a plain value proposition: "32x stronger cell signals".

Video views fell 30%. Sales rose 35%.

Same video, opposite result. The client call went a lot better.

So was the research wrong?

At this point you could decide users didn't want product information after all. But the research said they did, and said it clearly. Either the research was wrong or we were missing something.

Our bet: users still wanted the information, and video was the wrong way to give it to them.

First we added static content showing which carriers the boosters work with. Sales rose 23%.

Encouraged, we added a traditional how-it-works section. Sales rose 46%.

The research was right all along: users wanted the detail.

How to make your losing tests earn their keep

When a test loses badly, resist the urge to bin it. That's a lead worth chasing.

Play detective on the cause and then go hunting for the same element elsewhere on your site. If it's hurting you in one place, it may be hurting you in others. Question the format before you ditch the idea. Sometimes users want exactly what you're offering, served a different way.

Not every story wraps up this neatly. Our product page test could easily have lost too. Then we'd have been scratching our heads over why the same change hurt one page and helped another, and we'd have kept digging until the answer turned up.

Winners are more fun. Losers, if you stay curious, can be more useful. If you want to get more out of your losing tests, get in touch.

• Conversion
• Tech & Telecoms
• Conversion
• Retail
• Tech & Telcoms

---