In FMCG, the AI Project That Photographs Well Gets Funded

In FMCG the AI project that gets funded is almost always the one facing the consumer.

A campaign with a clever generative element. A branded assistant. Something that can be shown at a conference and photographs well. It gets approved because it looks like a decision, and because marketing already has a budget line that can absorb it.

The forecasting project sits unfunded. So does the work on distribution profitability by outlet. Those two are where the money actually is, and neither photographs well at all.

I understand why it happens. One is visible, the other is plumbing. But consider what is sitting in the plumbing. Stock in the wrong depot is working capital doing nothing. An out of stock on your fastest-moving line is revenue handed to a competitor with a shelf facing next to yours. Short dated product written off is margin that was never recovered anywhere.

None of that appears in a campaign review. All of it appears in the year end.

There is a harder version of this problem. Most distribution businesses in the region can tell you volume by outlet and cannot tell you profit by outlet. Until that is true, every route decision is being made on incomplete information, and no consumer-facing AI project will compensate.

Before approving the next visible thing, ask the commercial team a plain question. What would a ten per cent improvement in forecast accuracy be worth to us?

If the number is bigger than the campaign budget, you have your answer.