Written by Roger Jackson,
October 17th, 2019 | Views

We’re all about getting “so what” from data but we can’t get AI to do much to help us!

The world is full of AI conferences, AI suppliers, AI blogs, books, consultants. AI everywhere. Makes me feel we are missing out. That our business may be falling behind, maybe even facing extinction, our business is about making sense of data and getting answers to help better business decisions. At face value, we should be investing big in AI, surely?

Worried, I have been talking to lots of colleagues in the research and data business this year – I have been trying to find the case studies of how AI has changed the analytical game (not hypothetically how it might change the game). I have to confess. I’ve not come up with much so far.

What I know about AI (probably not a huge amount) is that software can now figure out or “learn” how an algorithm best connects different (repeating) and large data sets together. I saw one example at the ESOMAR conference where a company had proven that one metric from social listening could replace one measure in its brand tracker. AI found the formula to connect the dots over several years. That’s one very niche example. Nice. Superb if there are large scale, repeating patterns available to analyse.

Nearly all the CPG clients I meet are crying out for help to join up disparate data sets. Two big trends are happening around us:

  • More and more data sources
  • Less and less internal resource/time to make sense of it.

We in our company have a lot of data and spend a lot of man-hours making sense of it. And we have just two data sources ourselves (albeit big ones!) I’d love to find an AI solution that would free up our people to spend far more time talking to clients about their challenges rather than peering at data to create necessary ppt charts. We already automated a lot of our reporting, yes, but we haven’t got anywhere to augment the human ability to look at a bunch of numbers are be able to think “aha, given the business issue/client question, this data means XYZ”. Relax team.

Could AI help? Not as far as I can tell. AI needs to have thousands if not millions of repeating data to iterate and fine-tune “emerging” algorithm. Almost every presentation we create is different, even using similar data. Because the question is always changing. And in reality, we don’t do 100s let along 1000s of similar deliverables. So, in effect, there is nothing we can feed into any AI tool to build it.

And for our clients, the ability to join up sales data, market audit data, panel data, loyalty card data and perhaps our tracking behavioural data would be a phenomenal value driver. Connect the “what” to the “why” everyone cries out. To quote a prospective client recently: “no way will I give our people yet another data portal. We just need answers, not more data! Can’t you just give me a simple summary of the conclusions?” And another “we don’t even have time to log into the tools, let alone do half the potential analyses we should”. But will varied competing companies even share data to start the integration journey? And more importantly if the analyses were done are most often unique one-offs, and (in AI terms) not in large quantities, how can AI ever “learn” them?

The world where AI does the grunt work, so our people are released to be creative, forward-thinking, solution-oriented seems a long, long way off?

Or am I wrong? I’d love to be. What’s your experience so far? Am I missing something?