The strategy mistake I made at FunnelBud for years

For years I ran FunnelBud on a decision rule that sounds obviously sensible: always make the choice that improves the main metric right now.

The reasoning behind it was not stupid. In a complex market you cannot predict the long term consequences of a big strategic bet, but you can usually see the effect of the next small decision. So optimise locally, repeatedly, and trust that the compounding takes care of the direction.

It took me years to see what it was actually doing to the company.

What went wrong

We said yes to projects we should have said no to. Each one was defensible on its own. We reshaped our pricing and our service structure repeatedly to solve whatever problem was in front of us, and we did solve those problems. That is the part that makes this hard to catch: the method works, locally, every time.

What accumulated was a business doing a lot of different things for a lot of different customers. Each individual engagement was profitable. The whole was not.

We had lost the things that actually decide whether a company makes money: who we serve, how we serve them, and how the activities fit together. Porter's point about strategy being an interlocking set of activities rather than a list of good decisions. We had a list of good decisions and no interlock.

The uncomfortable part is that I knew the framework. I had recommended Porter to other people in a guide on understanding customers a few months before writing this. Knowing the theory turns out to be almost no protection against violating it gradually.

Why the rule fails

When you focus on the next iteration rather than the overall direction, you stop seeing the cost of misalignment, because that cost never shows up in the decision you are currently making. It shows up two years later in the shape of the company.

Every individual yes was small. The accumulated drift was not. And drift has no natural stopping point, because there is never a single decision bad enough to trigger a review.

So I am marking this method disproven in my notes today. Years later than it should have been, and only visible to me once the shape of the whole business had already changed.

What I do not yet have

A replacement rule. I know what is wrong with optimising the next decision, and I do not yet have the thing that goes in its place. Writing this down is mostly so that I stop quietly falling back into it.

Update, October 2025

A year on, here is the rule I settled on: move fast, but only along paths aligned with the direction already chosen, and prefer the ones that open more doors than they close.

The speed was never the problem, so the speed stays. What changes is the filter. A decision now has to pass two tests rather than one: does this improve things now, and does it point where I am already going. A profitable opportunity that pulls sideways is a cost, not a win.

The uncomfortable part is that this requires having a direction in the first place. The local optimisation rule was attractive precisely because it let me avoid committing to one. It felt empirical and humble. It was mostly just avoidance.

I write more about how I make decisions in how I decide.