AI optimizing for the wrong strategy

This case reached me a few months ago, and it's still one of the most instructive I've ever seen.

The company had redefined its ICP, adjusted its positioning, and changed its customer qualification criteria. A solid strategic decision, carefully built by leadership.

The problem stayed invisible for months.

The AI system running qualification and prioritization had been trained on historical data. No one had built a recalibration trigger. So, for months, the system kept prioritizing the old customer profile: the one that generated volume, but low LTV.

The team thought it was following the new strategy. The system was executing the old one.

When I mapped out the problem, the cause was easy to see: the strategic change had no technical counterpart.

AI doesn't read a planning memo. It learns from what you feed it. And this system kept being fed the patterns of the March business, long after October had arrived.

What was missing was something few companies build intentionally: a formal recalibration process whenever a relevant strategic change occurs.

Whoever defines the new ICP also needs to make sure the systems will operate by the new ICP. That means documenting the change, identifying which systems are affected, and deliberately triggering recalibration.

Without that, the company makes a turn on the map. The GPS keeps guiding it to the previous destination.

Save this post if you suspect that some system in your company might be optimizing for the wrong strategy.

Tell me in the comments: how long does it usually take, in your company, between a strategic change and the actual update of your systems?