When a product is out of stock, don’t make the customer start again.

Around 30% of visits to some landing pages for a B2B Ecommerce client were reaching products that were out of stock or not sold any more.

That did not always mean the customer’s need could not be met.

Some products had been discontinued. Some were only available to order. In other cases, a newer version or a similar product could do the job.

The standard product page left the customer to work that out for themselves.

The problem was not simply that stock information was missing. It was that the page could bring someone to a dead end when a useful alternative existed.

Start with the customer’s task – I led the discovery work and proposed showing relevant alternatives in a pop-up on the product page.

The recommendations favoured products from the same brand or product line. Where that was not possible, they used similar specifications. We also included key product details so customers could make a more informed comparison before clicking through.

The recommendation engine already existed. The work was to adapt its rules to this particular problem: helping someone who had arrived at an unavailable product find a suitable next option.

The developers built the pop-up and the recommendation rules. I worked with an independent analyst on the results, so the person calculating the uplift and revenue was not directly involved in creating the solution.

Test the approach, then learn across markets
The control group saw the existing product page without the pop-up. The test group saw the page with recommended alternatives.

We began in North America, then expanded the testing to Europe and Asia before rolling the approach out globally.

The European result was around a 6–7% uplift. Asia was a little lower. Europe contributed the largest share of the value because it was a larger market than North America. At peak, the estimated incremental revenue was around £11,000 a day.

Those figures describe this case. They are not a forecast for every recommendation tool or out-of-stock journey.

A recommendation needs to help someone decide
It is easy to judge a recommendation by whether someone clicks it. That tells you the recommendation attracted attention. It does not, on its own, tell you whether it helped the customer complete their task or created additional revenue.

The question we wanted to answer was whether offering a relevant alternative could turn an otherwise unproductive visit into a useful buying journey.

That meant testing the recommendation in context, comparing it with the existing experience and measuring its commercial effect. The independent analysis also helped keep the evaluation separate from the people who built the solution.

The useful question is not “Can we recommend something?”

A recommendation engine can show alternatives. The harder question is whether those alternatives are relevant enough to help someone move forward.

For an out-of-stock product, that might mean finding a newer model, a close equivalent or another item with the specifications the customer needs. The right answer depends on the task they came to complete.

This case started with a clear customer problem: too many visits ended at products that could not be bought. The solution made the next step easier to find, then we measured whether that change helped.

When a product is unavailable, the page should not assume the customer is finished. It should help them decide what to do next.