I recently moderated our special webinar with four brand leaders who shared top tips for starting and scaling a social intelligence program, as well as their perspectives on the role of AI – and what it’s really good for…or not!

The conversation was candid, lively, and full of quotable takeaways. I encourage you to view the full replay here (short registration required). But to whet your appetite, here are some key takeaways and highlights. If you are just starting or revisiting your social intelligence program, they make for a great checklist.

And if you are curious how AI-enabled consumer intelligence is being adopted in the field, the last section on the role of AI gives some great clues about the massive impact technologies like NLP and machine learning are/will have on insights and marketing projects across brands of all types and sizes.

Tip #1: Know which use case is most important to solve (first)

We hear this all the time. Know what problem you are trying to solve. Social listening and AICI use cases are super diverse! Just in this panel, we heard leaders share a wide range of uses for social intelligence, including: audience and market understanding, brand monitoring/brand health, and crisis communications. Plus content creation/performance, topic analysis, and understanding unknowns. Overall, the group agreed that social data is especially valuable to “answer the why,” and often comes to the fore in a time of market shifts or social changes.

“When people have a burning platform, that’s when they start to find the need for it (social intelligence).”

Rosa Halford, Asahi

Tip #2: Position social as enabling, not replacing consumer intelligence

Social data provides unique insights and context. Yet it’s just one part of your data universe! The panel shared a wide range of tips for positioning social data, and building support for a new social intelligence program, that included:

  • Take time to gather requirements – do interviews with future users
  • When choosing new tools, do a tech audit first to map features and capabilities of existing tools
  • Pilot what you can see with social intelligence, driven by top use cases
  • Be honest about the limitations of social data, and how it complements other sources (like surveys or search data)
  • Be realistic about your brand conversations, and what volume of mentions you should expect
  • Don’t assume users know analytics!

“Who are your actual end users? Are you talking about analysts? Are you talking about brand managers? Are you talking about someone else? Each of those people are going to be using this in different ways.”

Luke Elliott, Brown-Forman

Tip #3: Drive adoption and scale through reuse, and repetition

Like any tech-driven initiative, moving from early adopters to more casual users is critical to generating ROI for your social intelligence program. The panel offered some top tips for reaching this tipping point, where social becomes part of the consumer intelligence toolbox:

  • Showcase your social insights in context, with comparisons to other sources
  • Allow experts to build the first queries and filters, then capture/promote (in a “query library”) – so other users can use them as a starting point
  • Take a topic-centric view of your analysis (say, around consumer perceptions of EVs) vs a channel- or data-centric view
  • Train your users to understand what data is available, and how/when/why it should be used (for which questions)

“Bring your Synthesio data to everyday meetings, so that people understand what… social listening can provide.”

Jaakko Ylisipola, Fortum

Tip #4: Invest in AI, but don’t overestimate what machines (alone) can do

Synthesio has been promoting the value of “human-machine teams” ever since we became part of the Ipsos family. Yes, as the panel discussed, AI is instrumental in helping you figure out the questions you didn’t know to ask – and great for surfacing unknowns, and cleaning and processing large data sets.

“Are you clear about the questions that you are trying to answer from any data?”

John Atkins, Shell

But there can be concerns about how an AI tool or bot came up with an answer (Explainable AI addresses this), and with bigger and bigger volumes of data, whether AI can scale. The panel agreed that Topic Modeling is pretty awesome (we agree!), but that to put things in context, or understand small nuances, or interpret the various roles that consumers play (are you a consumer or influencer?), we still need humans in the loop.

To learn more, you can register to watch the full replay of our panel conversation by clicking this link.

WATCH THE REPLAY