The General Mills Strategy To Defend Against AI
Why are there 13 kinds of Cheerios?
Happy Sunday and welcome to Investing in AI. Be sure to check out our AI Innovators Podcast when you get a chance. If you have someone who would make a good guest, please reach out.
Today I want to talk about a lesson from my MBA days and what it might mean for AI. It starts with a study about Cheerios, my favorite cereal. Do you know why there are 13 kinds of Cheerios?
Cheerios launched in 1941. For 38 years, it was just one brand, then in 1979, a sub-brand was launched called Honey Nut Cheerios. That strategy of launching more niche brands of Cheerios continued to this day, and there are currently 13 brands of Cheerios on the market.
Why?
The case study we had in business school on this claimed the reason is actually a defensive strategy against new market entrants. You could worry that by offering variations on Cheerios you cannibalize the main Cheerios brand. But the flip side of that, and the side that is more important, is that when you fragment the market so that each sliver of cereal commands only low single digit market share, you make it less interesting to new entrants, and more difficult for them to expand if they do get a foothold in a niche.
What does this mean for AI?
I think you will see software companies start to behave similarly in an AI world. LLMs are making it easier and faster than ever to write code. That means fragmentation at the app layer because it’s easier to build minor variations of popular software that appeals to certain user niches.
To date, application software has had to build to the most common denominator for user preferences. As AI brings down the costs to create software, smaller TAMs can still generate decent investment returns. Large software companies will have to defend against getting picked apart by small startups that can stay lean, use AI, and each into some component of their market share. How will the big companies defend against that?
I think they will try the General Mills strategy. Over the next 5 years I expect to see large software companies offer many variations and sub-brands of their core software so that they fragment the market on their own, and make market entry less attractive to new players. It’s a common playbook in the Porter’s Five Forces model when the “threat of new entrants” force becomes more of a threat. But it’s not the way software executives or investors usually think, so, maybe it doesn’t happen.
What do you think? Is app layer fragmentation a real threat? And if so, is the General Mills strategy a viable way to respond?
Thanks for reading, and as always, I appreciate your feedback.
Ha! I wasn't sure where you were going with this. But great points and very plausible.
Brilliant. A slight nuance to the General Mills case study. As with physical goods and as with brick &mortar retail environments, taking up the most shelf space was an effective strategy for preventing entrants. However, consumers became overwhelmed with choice which would backfire. It continues to be that 5 - 7 choices is an ok heuristic. Makes sense given how we know memory works.
When it comes to app layer, I agree that specialization and smaller snack size apps is a likely way to go. In my experience, many C-Level leaders keep looking for vertical, sub vertical experience from their new suppliers. So depending on how the sub vertical is defined, it can be quite a niche app.
If we stick to thr FMCG analogy, I'm curious whether firms will take the P&G route of best in breed for every sub category in their portfolio or take the Reckitt route which has been criticized in the past for not really managing a portfolio but rather than buying, exploiting and selling brands.
I'm curious if there is a nuance I'm missing that makes more sense for the app layer.