How #research helped @twitter conquer the world

x Sponsored content on Research Live and in Impact magazine is editorially independent.
Find out more about advertising and sponsorship.

You’ve seen the photo. Ellen DeGeneres, accompanied by Bradley Cooper, Jennifer Lawrence, Lupita Nyong’o and a crowd of other A-list stars, grinning into the camera for a group selfie at the 2014 Oscars ceremony. DeGeneres tweeted the image with the message, ‘Best photo ever. #oscars’. More than three million people have since retweeted it.

The very language that it takes to describe the incident reveals the impact Twitter has had on our world: tweets, retweets, hashtags… before 2006, none of this meant anything to anyone.

For a quirky social network whose original USP is what it can’t do – namely, send messages of more than 140 characters – Twitter’s cultural influence has been astonishing. It has helped bring down governments. It has created superstars. It’s the first place to look for major public announcements – everything from political resignations to celebrity deaths – and the current US presidential election is the third to have had Twitter at its heart.

Half a billion tweets a day are sent by 313m monthly users, helping to make the site a $2bn (£1.5bn) ad platform, with a further $224m (£168m) a year coming from data licensing. But it still needs to grow these numbers if it is to become profitable (the company’s own adjusted figures show a profit because it excludes the cost of share issues but, officially, Twitter is still a loss-maker).

To support this effort, the company has a 60-strong global research function, including teams at the company’s San Francisco headquarters that focus on user experience and product development research, rolling out new features to 1% of users and seeing what happens. Other teams look at ways to make the site safer and deal with misuse and abuse.

The international research team, which numbers about 20 – including 12 in the UK – has a simple mission: to ‘prove and improve’ the value of Twitter in each country.

Jake Steadman and Matt Taylor were among the first researchers to join Twitter outside the US. Steadman began by leading research in Europe, the Middle East and Africa, and now also oversees the Latin America and Asia Pacific regions as international research director. Taylor has just moved from European head of research to consumer insight lead. Both are prolific tweeters. A neon sign on the wall of their office says #lovewhereyouwork – and they do. “We do an intellectually stimulating, creative job for a company we’re passionate about,” says Steadman. “It doesn’t get much better.”

Twitter’s research addresses three big questions: what the platform can do, who’s using it, and whether ad campaigns are working. A big chunk of the work is thought leadership research to explain to the world what Twitter is for and why it matters – particularly important in markets where the platform is less well established.

So what is Twitter for? “Twitter is news,” Taylor says. “It’s as simple as that. Twitter’s not for the people around you in your life – there are other social media platforms for that. Twitter is for what you’re most passionate about. I’m a massive football fan, and all of the best news and the most exciting content happens first on Twitter. If you’re into food or politics it’s the same.”

The current business challenges facing Twitter’s research team are reflected in three words that keep cropping up in our conversation: mobile, video and live. 

Mobile has been Twitter’s main focus for a while: 80% of its traffic comes through mobile devices, significantly ahead of the trend for internet use generally. As a result, everything Twitter does is ‘mobile first’, Steadman says. 

“The big shift for us – and the industry in general – is the move to video,” says Taylor. “Mobile video is one of the biggest growth areas, not just for advertisers, but for people using their time. We’ve seen huge increases in people posting video.”

And now video is going live. Twitter recently acquired the live-streaming app Periscope, through which users can broadcast live to the internet from their phones. Separately, it has begun to host live streams of major TV shows and events, including the Wimbledon tennis championships, NFL American football games and the Democratic and Republican party conventions, with video streams accompanied by curated feeds of relevant tweets.

Steadman is quick to point out that Twitter doesn’t just analyse its own data. “We are Twitter’s research team, which is different from being a team that just researches Twitter data,” he says. Analysing data from Twitter and other social media probably represents about 10% of the team’s work, he says.

Twitter data offers a number of specifics – speed, raw ‘in-the-moment’ insight, huge global reach, and an ever-growing longitudinal dataset. But it’s just one tool in the research team’s toolkit, and it’s most powerful when combined with other sources.

Taylor says: “The most important thing we do with our data is try to integrate it into other types of research. As soon as it all comes together it becomes amazing. For example, it elevates qual because you can combine it with behaviours. You can say, let’s take everybody who’s tweeted about Brand A in the past six months; we know that they’re interested in it because they’ve talked about it, so let’s take them all off to a separate room and talk about it. It just connects everything in a really cool way.”

Taylor describes the range of research techniques Twitter uses as “almost anything”. “Everything from neuroscience to implicit attitudinal testing to research communities, quant, qual, everything else.”

Observational research has helped the team understand how people use Twitter in a way that surveys couldn’t. One piece of research conducted with the TV marketing body Thinkbox showed how people carefully redrafted and honed tweets about TV shows, to try to make them funny enough to earn a retweet. When asked about this in surveys, however, users denied doing it – even though they believed that a majority of other people probably did.

When evaluating clients’ campaigns, it’s important that Twitter is able to speak the language of the rest of the media research world, says Steadman. “There’s an assumption that it’s about social data and, therefore, we’re just looking at our own dataset, but we don’t at all. When it comes to measuring the effectiveness of advertising, there has been 100 years’ worth of thought that has gone into that, and the thought that was applied to TV in the 1960s may well apply to Twitter today. We apply all that learning that our industry has.”

The Twitter team deliberately works with a wide range of agencies and makes sure it gets multiple pitches for every project. Researchers in each country select local suppliers to cover their market. In the UK, Twitter has worked with, among others, Nielsen, Millward Brown, TNS, Crowd DNA, Research Now; qual agencies Flamingo and Firefly Millward Brown; neuroscience specialists Neuro-Insight; and online community provider C Space. 

As always, new research projects are in the pipeline. A piece of semiotic research conducted with Flamingo sought to uncover the secrets of effective tweeting, and eventually boiled it down to ‘the 11 tweets that work’. 

More things along these lines are on the way, Steadman says. “The semiotics project is a really good example of pioneering research work that is being applied to Twitter data, and there will be a lot of really exciting things to come – particularly from the Twitter UK research team – in the next few months.”

For future research, Steadman wants to look into how people use Twitter in different markets – in particular, in Japan, where language differences mean Twitter is a very different beast. “One hundred and forty characters in Japanese gives you a lot more freedom, and it has created a different user
case in the way people use Twitter there.
It’s fascinating – we’re doing some work to understand how people use that, and what
that adds for them, in a way that it doesn’t in other languages.”

A key element of the research team’s culture is its creativity, Steadman says. This means experimenting with innovative methodologies, and coming up with interesting and unexpected ways to communicate the findings. Steadman encourages all the company’s researchers to ask themselves: “What if Keynote and PowerPoint didn’t exist? How would you get your story across?”

Taylor agrees that it’s vital to make the communication of research powerful and memorable. “There’s no other discipline I can think of in business that can be as impactful as research,” he says. “You have a fact here about consumers that will lead you to improve your business’s health. If people listen to that and apply it, you can change the business. But you have to be able to persuade them.”

Often, that means doing something different. “Every single marketing director in the world comes into work and they’ll have pitches from media owners, from agencies,” he says. “They see a sea of PowerPoints. The only way you can stand out in that is by attempting to be as distinct as possible.”

That’s why they have done things such as making a documentary film about music fans on Twitter, to present the results of an ethnographic study. And it’s why Steadman recently found himself sitting on a stage wearing a neuroscience helmet, with his brainwaves being displayed live on a screen, to report the results of a piece of research on people’s emotional responses to Twitter.

Steadman knows that his live brainwaves didn’t actually reveal any insight – but that’s beside the point. “It’s the theatre of it. It makes people remember it in a way that they wouldn't do if you give them a bunch of charts that technically explain it.”

The result is that when Twitter’s researchers have something to say, they always have their colleagues’ attention. 

“In some businesses you invite 30 people to the debrief of the project and eight people turn up,” says Taylor. “Here, you invite 30 people and 30 people show, take notes and listen.”