Consider the ethics of data collection, conference hears

In a conference panel session, Jake Steadman, vice-president research and data at Canva, said that ethics needs to be considered when carrying out any project involving potentially revealing personal data.

He cited the example of a Canva project at one of its conferences that involved tracking attendees’ movements through the use of a chip in name badges in order to get insight on which parts of the event were most popular. In this case, attendees were given the opportunity to opt out and were asked permission.

“Just because we can, doesn’t mean we should. And if we do, we need to have clear permissions and need to be clear about it,” Steadman said.

“Just being clear and not doing something that’s creepy. Be a good human and don’t do something that is going to feel weird and that you wouldn’t want doing to yourself.”

He said that when done appropriately, combining event and social data could be very beneficial for understanding the effectiveness of events.

Steadman added: “My job is to help Canva do better marketing – that’s ultimately why I am here. The art of what is possible is exciting and interesting, but it always has to be rooted in the ‘so what?’ – what can I do, and what decisions can I help the company make.”

Also speaking on the panel, Richard Preedy, executive director at Verve, said that artificial intelligence (AI) could help provide deeper analysis of social media data in the future.

“Given the scale, volume and size of data you are able to collect now from social platforms, AI becomes an invaluable tool,” Preedy said. “It enables us to do things that just wouldn’t have been possible before at a scale that humans simply couldn’t do.

“AI helps us to make sense of big, messy, complex, unstructured data sets. What it does very well is to distil down that data into more manageable, digestible themes. It does not give you perfect answers – it is not a silver bullet – but what it can do is get you 80% of the way there.”

Preedy said that human input was crucial to successfully embedding AI in research more broadly.

“We need that expertise and knowledge of our category and of the data in order to get the best from it,” he explained.

“With things like AI, we can see hallucinations and things appearing, but if you don’t know your subject matter, you might not spot those things. It is important that humans are involved still.”

Synthetic data also offered a lot of possibilities, Preedy added, if it is derived from quality datasets, providing confidence in the model and data produced.

“I think then jury is still out on where exactly it is going to go, but I still think synthetic data can play a big role in the industry. It just needs to be managed in the right way.”

Deanna Tserkezie, research director at Pulsar Platform, said that social media data should be combined with other sources, such as print media, broadcast and podcasts, to “build out that digital ecosystem and where people are getting their information from”.

She added that it could also complement survey-based research: “I think social can often be used as an ‘early indicator’ of what survey results might reveal. Survey results might give you that added depth from the core audiences you are interested in, but social gives you that unprompted view to shape what you might ask in a survey.”

However, it is important to tailor the choice of social media platform to the target segment, which involves having a more detailed knowledge of the types of people using each individual platform, Tserkezie added.

“You have to understand their behaviour might not be represented in more traditional media, but will be really active on somewhere like TikTok,” she said. “It depends on understanding where your audience might be talking or where might be influenced.”