Starting from the right question
The wrong starting question is "how can we use AI?" The right one:
Which value drivers can AI materially improve?
That reframing forces every AI initiative back through Lesson 18's three levers and Lesson 28/29's specific growth and cost levers, rather than starting from the technology and looking for a problem to attach it to.
Examples, each tied to a lever
Concrete examples, each tied to a named lever: pricing (usage-pattern-driven tier recommendations, feeding Lesson 28's pricing lever); sales (lead scoring that improves rep productivity); churn (Lesson 26's account-health scoring, now specifically framed as a predictive-model problem); support automation (Lesson 29's automation lever, directly); operations and procurement (anomaly detection in vendor spend).
Every one of these needs the same structure to be a real initiative, not a pilot that goes nowhere:
Investment → operational change → financial impact
An AI proof-of-concept that never changes how anyone actually works, or never gets traced to a financial number, hasn't created value yet, it's stopped one step short.
| Lever | AI use case | Example |
|---|---|---|
| Pricing | Usage-pattern-driven tier recommendations | Flag accounts under-tiered for their usage |
| Sales | Lead scoring | Rank pipeline by likelihood to close |
| Churn | Account-health scoring | Predict at-risk accounts before renewal |
| Support | Automation | Deflect routine tickets before a human touches them |
| Operations / procurement | Anomaly detection | Flag unusual vendor spend for review |
Checkpoint
- Start from the value driver: (which lever, from Lessons 18/28/29), not from the technology.
- Every AI initiative: needs a traceable path: investment → operational change → financial impact, not just a working model.
If anything here still feels unclear, ask before moving to Lesson 34.