I’ve been a big fan of Chris Field’s “Sunday Soundbites” for some time now. Chris owns Fieldworks, an award-winning retail specialist PR and digital content marketing agency.
Chris wrote a piece back in November 2017 entitled “Six things I just learned about Artificial Intelligence” which I’ve been meaning to get round to thinking about, as it is focused on the use of AI by retailers and his concerns about the path that some retailers are taking.
In it, he makes, as usual, some sensible observations about the hype around AI and specifically compares the potential for AI in retail as touted by various technology providers with the reality as experienced by retailers.
Retail isn’t really my area, but in my role as Analytics Lead for Beacon, I like to spend time making sure that I & my team have a good understanding of how traffic behaves differently and the same across disparate sectors.
Chris makes the very good point that “AI can’t fix bad processes”; this applies to any new technological innovation and I couldn’t agree more. He also says “AI will not save retailers that have too many broken, missing or disintegrated processes”, but that is where I feel there is a more nuanced conversation to have as I feel it’s possible that AI – or specifically AI data evaluation – may help such organisations.
Software has a phrase that highlights the problem of bad data clearly: “garbage in; garbage out”. Actually, it is more complex than that – you can have great data, but a poor process, resulting in bad insights against that data; equally, you can have awful input data and great processes for manipulating it, but your learnings will still be wrong. The phrase “Garbage data and/or garbage processes in; garbage out” is somewhat less pithy however!

Where AI comes into this is that in order for AI to work properly, you have to have a great set of data to firstly train the AI, and then for the AI to act upon.

So, by adopting AI, you enforce your organisation to also look at your own data; to identify where maybe it isn’t so great and to make it better, either by making it cleaner at source, or making sure it gets clean before it goes into the AI tumble dryer, or by adding additional data from other sources (internal or external) to enhance the data you do have. This is one of the main thrusts behind Process Marketing Automation – bringing together many sources of data into a whole that gives better visibility of the value in the data.
The takeaway here is that if your organisation starts implementing AI (which is a long and possibly never-ending job – at least until the Terminators arrive), this will give your poor data and poor processes nowhere to hide – you’ll have no choice but to improve both. If you’re lucky to already have good data and good processes, well done! There’s always room for improvement, however, and AI implementation is a good way to prompt it.
What does that mean for retail specifically? I’ve no idea, really, but hopefully Chris can have a think and write a blog about it – I, for one, am looking forward to it.
Campaign intelligence free demo