AI & Automation · AI Strategy · Medium · 25-35 min

Company Wants an 'AI Strategy' — Where to Start

Company Wants an 'AI Strategy' — Where to Start is a medium AI & Automation ai strategy case interview that runs 25-35 min. A retailer's CEO has told the board the company will be 'AI-first,' but there's no data foundation, unclear ownership, and a dozen disconnected pilots that never scaled. A strong answer works through 5 phases: Anchor on value pools; Assess readiness; Explain why pilots don't scale; Prioritize a portfolio; Recommend a sequenced roadmap with the operating model to sustain it.

Last updated 2026-09-05

The brief

A retailer's CEO has told the board the company will be 'AI-first,' but there's no data foundation, unclear ownership, and a dozen disconnected pilots that never scaled. You're asked to turn the ambition into a strategy: where AI creates real value, what has to be true to deliver it, and how to avoid another graveyard of pilots.

How to approach it

  1. Anchor on value pools — where AI moves the P&L (demand forecasting, pricing, personalization, supply chain) rather than tech for its own sake
  2. Assess readiness — data availability/quality, platform, talent, and governance
  3. Explain why pilots don't scale — no data foundation, no ownership, no path to production — and what fixes it
  4. Prioritize a portfolio — a few high-value use cases plus the enabling data/platform investment
  5. Recommend a sequenced roadmap with the operating model to sustain it

What a strong answer does

  • Starts from business value, not from models or tools
  • Identifies data foundation and MLOps/path-to-production as the reason pilots die
  • Balances quick wins against the unglamorous platform/data investment
  • Assigns ownership and governance, not just a use-case list
  • Sequences realistically instead of promising 'AI everywhere' at once

Red flags interviewers score down

  • Produces a list of AI use cases with no link to value or feasibility
  • Ignores data quality and the path from pilot to production
  • Treats AI as a technology purchase rather than an operating-model change
  • No prioritization — tries to do everything simultaneously

Cases are written in each firm's style, written and reviewed by working consultants; they are not the firms' own published cases.

More AI & Automation cases