The Operations Bridge

Point nine in the AI Strategy Points series. Regular readers will recognize the framework in this post — it’s the same “what, so what, now what” structure from my earlier Power BI series, applied here directly to AI initiatives themselves.

The same three questions, a different subject

Back in point two of this series, I said that AI initiatives have to create value to stick, but that value depends entirely on who’s defining it. This post is where I actually answer that question: the “what, so what, now what” framework is the clearest tool I’ve found for defining value in a way that different stakeholders can agree on.

What: what did the AI tool actually produce or change? This is the easiest layer to measure and the one most AI pitches stop at — faster reports, generated drafts, automated summaries. It’s real, but it’s just visibility into activity.

So what: why does that output matter? Did it change what someone understood, noticed, or was able to see that they couldn’t see before? This is where a lot of AI initiatives quietly stall — the tool produces something, but nobody’s understanding of the operation actually improved.

Now what: what decision, action, or behavior changed because of it? This is the layer that separates a genuinely valuable AI initiative from an impressive demo.

This is where the definition of value actually comes from

I said earlier in this series that value depends on who’s defining it, and that leadership and end-user definitions often diverge. This framework is how I reconcile them. Instead of arguing over an abstract word, I ask each stakeholder to walk through their own what, so what, and now what for a given initiative.

Almost every time, this exposes exactly where the disagreement actually lives. Leadership might have a clear “now what” at the strategic level — competitive positioning, cost reduction — while the end user can’t complete their own “now what” at all, because the tool changed what they could see but never changed what they were allowed or able to do differently. That gap is the real diagnosis, and it’s far more useful than a vague sense that “adoption is low.”

Using it before you build, not just after

The framework is even more useful applied in advance. Before greenlighting an AI initiative, I now ask the sponsor to sketch out all three layers as a hypothesis: what will this produce, so what will it help us understand, and now what will we do differently as a result. If they can’t get past the first layer, the initiative isn’t ready to build yet — it’s still an idea.

A question for leaders

Take one AI tool currently in use in your organization and walk it through all three layers honestly. Where does it stall?

Next: why AI projects need to solve real problems for real team members to avoid the quiet “fail” status that so many initiatives eventually land in.