Using Twitter polls for user research @ONS

With a 235,000-strong following, the ONS Twitter channel is a great avenue for gauging public opinion about our statistics from users of our statistics and people who are engaged with us as an organisation.

Alison Davies, one of our user researchers at ONS, explains how she approached the Social Media team when looking to undertake some research on our website:

“In the process of looking at content and prioritising products through a ‘top task’ survey, I naturally got round to thinking about the homepage. I’d spoken about this recently at some user testing in Birmingham – what people would like to see, what they’d expect to see, etc., and while the topic was still fresh wanted to keep the momentum on this. While chatting with our social media gurus around promoting our top task survey, they brought up the subject of a Twitter poll – and this seemed like an ideal opportunity to get some quick opinion on what we should be offering in a homepage. Using some of the feedback and comments we’d already had, I came up with these suggestions:

  • links to latest releases
  • table of key figures
  • most searched for content
  • topical content / in the news

Our users have had time to get used to the new site, so we’re possibly in a better position to make a long-term decision on homepage preference or priority, with their needs in mind.”

We quite often approach our followers on social media for help with consultations, surveys and to provide feedback on new products and formats, but this means clicking a link off onto a separate site, submitting a form or sending an email. The Twitter poll functionality is a great way to get instant feedback with minimal effort as users stay on the app and aren’t redirected to any other pages. The limitations to the poll feature are you’re only able to have 4 options, and only 1 option can be selected, but the benefit of it being hosted on Twitter is that the open platform gives people the opportunity to respond to give further details and context alongside responses, and this is all captured and fed into the user research. We can also follow the poll up thanking those that contributed and link to a full survey where people can select more options from a longer list.

https://twitter.com/NorthWesterly/status/747793206067212288

The Twitter polls feature was at first only available on verified accounts and with only 2 answer options and a set 24 hour limit, before it was rolled out to all users with up to 4 options and the ability to set a custom time for the poll for up to 7 days. It’s a new way for us to engage with our audience and understand exactly what people think, and it’s a simple and easy way for our users to get their voices heard with a click of a button/tap of a phone. As with many new features, we’ve been eager to learn and the best way was to get stuck in, test it for ourselves and see the results.

We trialled using the poll feature around Christmas but as a quiz feature. It wasn’t perfect as, while the polls are great for testing people’s perceptions of what they think the correct answer is, they don’t necessarily as a quiz feature as there’s no clear way of revealing the correct answer. Although really popular, it was a bit clunky logistically.

It was exciting to see real-time engagement coming in and we had our first couple of hundred votes within the first 2 hours, with a total of 542 votes after 24 hours. Our most popular poll from our #quizmas poll series was a baby names question which got 376 votes – really high engagement for a 24 hour period. We’re keen to use the polls again where suitable, but only where we feel we can get real value, as we don’t want to overuse it and be pestering and spamming people on issues they’re not interested in. But considering we have a large follower base willing and able to help, it’s a no-brainer to tap into this when we can. People like to have their say and to know they’re valued and listened to on topics they care about.

Twitter polls are an effective way to run opinion polls, hosted natively for neat, bitesize pieces of product feedback that can be broken down to 2, 3 or 4 answer questions. Although votes are anonymous, it’s a good starting point to get the ball rolling to gauge opinion and give us some sample data to get started, fuelling bigger decisions and user testing in the future.

Polls can be used to build stronger connections, and shows us to be open and responsive as an organisation, asking our users what they’d like to see more or less of, allowing them to play a part in deciding content and format. It was interesting to see the shift in the live results as 3 different options were in the lead at one point. One limitation to the polls feature is that images can’t be included, it would be good to do some simple A/B testing to gauge public opinion over some of our content and preferences to different designs.

Alison concludes:

“So, what are we doing with this? The results of this poll have definitely given a good indication of what our users want, and the poll has backed up feedback we’ve had from other sources…no doubt we’ll have further work to do, but we’ve got great direction from our users and will be looking at the homepage in our next sprint.”

Lauren Bradford

Social Media Manager