Inside Expedia’s London usability lab

Expedia is using facial recognition technology on holidaymakers surfing the internet for inspiration and to book their next trip. Ben Ireland sat in as the online travel agent’s scientists demonstrated their new technology. Walking through what could have been mistaken for a door into a boiler room in Expedia’s modern London offices in Angel, you … Continue reading Inside Expedia’s London usability lab

Expedia is using facial recognition technology on holidaymakers surfing the internet for inspiration and to book their next trip. Ben Ireland sat in as the online travel agent’s scientists demonstrated their new technology.


Walking through what could have been mistaken for a door into a boiler room in Expedia’s modern London offices in Angel, you wouldn’t think you were entering a hub where analysts are developing the future look of one of the world’s biggest travel website.

After brushing away a curtain, you appear in a darkened room kitted-out with multi-screen computers displaying the live results of facial recognition technology, known as electromyography (EMG).

A test-dummy – who is paid, but by an outside agency, not Expedia – arrives and takes a seat.

She’s talked through proceedings by user experience research manager Roseann Ferrara, who attaches sensors to her cheeks and eyebrows as we sit on the other side of a glass window.

Inside Expedia’s London Usability Lab

Eyebrow movements, we’re told, show when someone is either focussing or frustrated. Cheeks are a person’s “happy” or “delight muscle”.

To test if it’s working, the subject is asked whether they prefer cats or dogs and is shown pictures of their favourite animal.

When the subject begins browsing through Expedia and other sites looking for places to stay in Porto Alegre in Brazil (she’s already bought flights there), we see two bars akin to what you might see measuring a heart rate in a hospital start to twitch.

Green measures the cheek movements while red measures the eyebrows. Meanwhile, an orange dot tracks which part of the screen the user is looking at.

Inside Expedia’s London Usability Lab

Along with my fellow laymen in the room, I instantly draw my own conclusions. Ferrara tracks every movement and makes her own notes and lets on that she notices the subject is paying close attention to the price.

Yet there are some strange anomalies, like a clearly awkward laugh being measured as ‘delight’.

We learn that people with big beards, or who have had botox, can’t take part in the 45-60-minute sessions as the sensors don’t work. And, apparently, Army and police workers are less readable as “they’ve been trained to control their emotions”.

“We know it’s scientifically sound,” said Tammy Snow, user experience manager at the London lab. “We know from science that people can have very emotional responses to something before they’ve had time to process what they’ve seen.

“We want to see what is causing the emotional response and we do everything we can to eliminate bias to gather meaningful results.”

The team’s insights are used alongside insights from other research being conducted in parallel and say that after around 150 tests they will get meaningful data to use to make suggestions for what Expedia can do to make its website more user-friendly.

A similar lab was created in the US three years ago and began to influence product design after a year. Snow said the team can start to see patterns as soon as three to six months in.

“One of the things we have found is that when people have spent time to find a hotel and then find that it’s sold out, they become really frustrated. So telling them how many rooms are left when they’re shopping can be quite useful.

“We’ve also noticed that a lot of hotels have different room types and differences are not always obvious when people have to scroll up and down – especially on mobile.

“Photos can create a lot of delight,” she added. “People can imagine an environment or situation.”


So what can we conclude from seeing the research first-hand?

The researchers told me that there are a lot spikes they have to siphon out as “non-events” and it sounds like it will take a fair few tests before any meaningful results are obtained. Close attention by experts is necessary to filter through the data and compile their own reports.

A global giant like Expedia can probably afford the luxury of this sort of lab, but I doubt it’s affordable for many smaller players to use to strengthen their own websites’ usability.