Prereqs: Hiring machines & training them to be decision-makers.

For better or worse I approach problems like a digital marketer who studied economics and loves technology. I get bored easily, too. So naturally, I want things better, faster, and–when possible–automated. Blending the handshake between mind and machine requires concrete thinking about flexible frameworks. It goes beyond HCI (human-computer interaction) and pushes closer to bionic.

Visibility. First you need to see what’s going on.

Questions:

• Are you measuring the right things?
• Are you able to–with relative ease–create analysis and dashboards and reporting?

If you can't create reliable dashboards and reports, then you are not in a position to let machines make decisions.

Optimization. Turn visibility into insight into action.

Questions:

• Are you able to surface actionable insights?
• Can you repeat the insight --> action --> learn feedback loop process?
• Are patterns emerging? Are you codifying those insights?

If you've achieved visibility into primary areas of the business, then you have to be ability to filter through the noise and identify actionable insights. As soon as the insights become reliably repeatable, then you can prototype the decision tree through human-led optimization actions. As you build a catalog of key decisions that include the constraints and guardrails, you can slowly give up control over the space between insight and action.

Automation. Let technology make the decisions for you.

Questions:

• Are you surfacing repeated, similar insights?
• Do you have guardrails and fail-safes in place?
• Is there anomaly detection in place?
• Is the culture of the company ready?

The culture part is as hard–maybe harder–than the technical aspect. But more on all that later. Most companies aren't past Visibility and Optimization.