The "local maximum" trap

Sometimes, the most dangerous thing for a growth team is when something starts working. 

The curse of the local maximum

Small wins give you just enough validation to keep doing the wrong thing – you’re climbing a hill when there’s a mountain nearby. (e.g. Ads that only pay for themselves. Outreach campaigns that bring just enough leads.) It's almost better if nothing works; at least then you’re allowed to think.

How do you know if your small wins are real progress, or just blinding you to bigger opportunities?

The clearest tell

The metric moves and the business doesn’t.

Gains in one area are offset by losses lower in the funnel – you improve click rates or free trial signups, but activation, paywall conversion, or retention gets worse.

If you’re working on your global maximum, you’ll see conversion improvements amplify, rather than diminish as you move down the funnel.

How to find your global maximum

Funnel data tells you where people drop off, not why. Drop-offs cluster at decision points – a paywall or a demo booking. But the decision didn't create the doubt, it just exposed it. Whatever made them hesitate happened earlier.

So don't start with drop-off points. Start with what needs to be true for someone to buy:

• You're talking to the right person

• Your product helps them with a top-3 priority right now

• They understand how you’ll help them achieve that goal

• They believe it will work…

• And won't embarrass them in front of colleagues, family or friends

• They have no unanswered questions or worries

• Their first experience of the product gets them to the “aha moment”

If any one is missing, they can drop off, and the fix may be anywhere upstream of the drop-off point.

Study people who did purchase. For them, these things lined up: right customer, right message, right timing, right experience. That's your winning formula.

Now study the drop-offs and find out which piece is missing.  That’s where your global maximum is hiding.

Helpful? Read all my posts and subscribe here.