Innovation Strategy
Structuring innovation with horizons, discovery practices, time-boxed bets, kill criteria and paths to production, including how to approach GenAI innovation without pilot purgatory.
Executive explanation
Innovation strategy defines where to look for new value, how much to invest, how ideas are tested cheaply, and how successful experiments reach production. Without it, companies either do nothing new or accumulate pilots that never scale ("pilot purgatory").
Business scenario
A bank has 40 GenAI proof-of-concepts across departments; none are in production after 12 months, costs are rising and risk teams are nervous.
Decision framework
- Focus areas tied to strategy (e.g. customer service cost, advisor productivity, fraud).
- Funnel: many cheap discovery experiments → fewer pilots with real users → few scaled products.
- Explicit hypotheses and kill criteria for each bet (metric, threshold, date).
- Shared platform (gateway, guardrails, evals, data access) so pilots are production-ready by design.
- Path to production: owners, risk review tiers, funding for scaling.
flowchart LR IDEAS[40 ideas] --> DISC[Discovery: 2-week tests] --> PIL[8 pilots with users<br/>success metrics] --> SCALE[3 scaled products] PIL -->|kill criteria met| STOP[Stopped, lessons captured]
For the scenario: consolidate to 6 use cases with the highest value and feasibility (contact-center summarization, advisor knowledge assistant, document extraction), build them on the shared AI platform, and stop the rest.
CTO interview questions and model answers
How do you avoid pilot purgatory?
Choose use cases with measurable business value and an owner willing to change their process, design pilots on production-grade foundations, define success and kill criteria up front, and fund scaling for winners in the same planning cycle.