Interview: The Science Behind Brandwatch Search With Aykut Firat
By Phill AgnewSep 28
Combining high-quality mobile survey technology, a robust polling methodology, and expert data analysis,
our bulletins will be essential reading to get the pulse of the nation
AI has a pretty polarizing reputation.
Some hail it as the savior of the modern worker, saving us thousands of hours which are wasted on simple tasks and finally offering us the prospect of a four day week.
Others view it as an evil villain, ready and willing to take our jobs and leave us unemployed and useless.
In reality, many of us are undecided on how we truly feel about AI. I love the personalized movie choices Netflix’s AI provides me, and I’m genuinely grateful that AI helps doctors detect cancer at a higher rate than ever before. Yet I still worry about what the 297,000 taxi drivers in the UK will do once autonomous cars take over.
Regardless of what you think, change is coming. AI already is present in most aspects of work and its importance will only grow in the future, and social listening is no exception..
For us at Brandwatch, AI is now part of our product. This week we launched Iris, an AI-analyst built to help you find better insights faster.
I wanted to know what a real analyst thought of Iris, so I spoke to Ben Ellis who headed up social listening at BT, Microsoft, Groupon and most recently We Are Social, before joining Brandwatch last week.
Ben says that defining a specific job that Iris supports is difficult. “You could easily argue that it helps with every job – It spots successful campaigns, a potential brand crisis, new influencers, customer pain points, unseen consumer insights and more.”
Ben outlines three key ways it can help analysts working with Brandwatch:
Manually analyzing peaks takes time. Clicking through and analyzing a peak while comparing it to the historical data would take a user several hours.
Most users take shortcuts to get around this, Ben says. “They click on the peak and analyze a small subset of data. This could lead to inaccurate results, but it takes them less time.”
He continues: “Still, in a typical week, your brand, products, campaigns and competitors will generate 20 major peaks. I’d say an analyst spends around 3 hours and 20 minutes conducting this analysis – in other words, a whole morning. Iris would take less than two seconds to conduct the same analysis.”
“Iris is much faster than human analysis, providing arguably more reliable results.”
Often analysts don’t have the time to compare a peak with 20 days of historical data manually. They could miss significant insights.
“They’ll miss the news story trending today, the video released commenting on the issue, the Reddit thread that popped up hours earlier or the new influencer that joined the conversation,” Ben says.
“Iris highlights these anomalies for you by constantly comparing to historical data. It keeps you aware of all the important insights around your brand.”
Ben thinks that some analysts, who have spent years working with data, might turn their noses up at a virtual assistant to help identify important trends. But, he says, a second opinion can always be helpful.
“If they look at a peak and decide that an influencer caused the growth, there’s no harm in checking that against Iris. A second opinion could help solidify their insight or make them aware of something they didn’t see before.
“Iris provides expert users with a crucial second opinion, to make their insights even more reliable.”
Ben says he’s seen social listening tools cram AI in their platform, but often this isn’t done with the analyst in mind.
“When poorly implemented, these newly AI-powered tools only bring up patterns with little to no context. Some of it can be useful, and some of it has served me well in the past, but as an analyst I found myself and other analysts around me being lost in the mechanics of these tools.”
He says this could include re-programming tools to fit the user’s needs, having to re-categorise mentions for accurate reporting, or having to adjust and correct the tool where it went wrong, hoping that it’ll learn from its mistakes moving forward.
“These implementations just don’t work for analysts. I’ve used so many of these tools, enough to comfortably say that Iris is different. Insights are observations with a reason. Iris surfaces and highlights these observations, by looking at the data, structuring the information within it, identifying the patterns while taking context into account. This is what an analyst would spend hours doing. This is what Iris spends a second on. Iris is different because it works for analysts.”
“I’ve seen so many tools adapt to AI, some well and some not so well. The examples I love the most come from tools that use AI for predictive analytics, a practice that I wish analysts would adopt more,” Ben says.
He’s got some examples of his favorites:
“Tools like Leadza analyze ad campaigns and send optimization tips to stay ahead of the curve. Management tools like Metigy analyze the performance of owned channels and competitors, offering community managers a live-stream of practical suggestions they can take to get better engagement and stay on target. Then there’s the HR use case of tools that use AI to automate candidate evaluation and predict team-fit based on your social footprint, for example Frrole Deepsense.”
Ben doesn’t think AI will replace jobs for social intelligence specialists. Instead, he says, it’ll replace some of their tasks while contributing to the natural evolution of these specialists’ jobs.
He says: “Social management platforms are slowly rolling out AI to quickly prioritize and route inbound messages. This speeds up tasks for community managers while helping analysts when it comes to reporting back on the key drivers behind these inbound messages. The analyst’s job will have to accommodate the increasing availability of social data and the speed with which trends appear and change. This would normally slow down an analyst, but AI stops that from happening.”
“AI won’t replace social intelligence specialists for one solid reason: it has the capability for insight detection, but no capability for wisdom. AI can surface observations in no time, but that alone doesn’t translate to action.”
For that reason, he’s confident that AI isn’t here to replace the social intelligence specialist – AI is here to help the specialist be more intelligent while saving time in the process.
“With most internet trends, we have an explosion and an implosion. For instance, years back we had a boom in influencer marketing, with the focus on big-time influencers and big follower numbers, while now we’re seeing an implosion, with local and regional influencers taking the spotlight as people call out for bigger transparency and deeper relatability.”
Ben explains that the face of social intelligence has started its implosion phase: social platforms aren’t as generous as they used to be with who they give their data to and how much of it they make available.
“Dark social, social or digital sources which can’t be easily tracked by digital analytics tools, is now the norm, especially now that active social media use on mobile devices is a reality for 42% of the world population, a 13% increase from last year.
“Tribalism on social media is on the rise, whereby people create and flock to their own subcultures on social platforms due to higher relatability. This gives a voice to issues and topics that the mainstream on the platform just won’t cover – Black Twitter, for example, or the make up artist community who are genuine trailblazers on image and video platforms.”
“With this backdrop, it’s clear what the next five years will hold for social intelligence specialists,” Ben says. “Putting together the pieces from where information can be measured with new technology.”
Ben thinks technology can only take us so far.
“Ultimately the onus of bringing all these pieces together from multiple sources to surface insights and weave a story remains on us, the social intelligence specialists.”
Find out more about Iris here.
Combining high-quality mobile survey technology, a robust polling methodology, and expert data analysis, our bulletins will be essential reading to get the pulse of the nation.