Ideas · Framework
The WINS framework: where generative AI changes your economics
The WINS framework tells you how exposed your business is to generative AI. It measures two things: the share of your cost base spent creating or improving Words, Images, Numbers and Sounds, and how digitized that work already is. Where both are high, reinvention is mandatory. Where both are low, you have time to learn.
What is WINS work?
WINS work is any task that creates or improves words, images, numbers or sounds. It is where generative AI bites first, because the models read and write all four.
| Letter | What it covers |
|---|---|
| Words | Documents, contracts, email, code |
| Images | Designs, slides, photos, video |
| Numbers | Analysis, models, reporting |
| Sounds | Calls, meetings, voice |
WINS is more precise than “knowledge work.” A software programmer, an accountant and a marketing manager do WINS work. A heart surgeon and a chef are knowledge workers, but the value they deliver is physical. Generative AI is power tools for WINS work. Asking people to do it without AI is like asking a carpenter to build a house with a hand saw.
What are the four positions in the WINS framework?
Score your business, or any function in it, on two questions. How much of your cost base is WINS work? How digitized are the WINS inputs today? The answers place you in one of four positions.
| Position | Profile | What it means | Examples |
|---|---|---|---|
| In the Crucible | High WINS share, highly digitized | Reinvention is mandatory. Fast movers win on cost, speed and quality. | Software, entertainment, professional services, financial services, education |
| Next in Line | High WINS share, low digitization | Digitize the work first, then reinvent it, or leapfrog with a new workflow. | Home décor retailers building a “digital front door” where customers see products in their own rooms |
| Holding a Lever | Low WINS share, highly digitized | Pull leverage from the WINS-heavy pockets: bids, service, supply chain, SG&A. | RFP responses where speed wins the sale. Moderna required GenAI training for every employee. |
| In the Balcony | Low on both | Monitor and learn as the tools mature. Urgency is lower, not zero. | Businesses built on physical products, physical services or low-skill labor |
Companies In the Crucible face the shift photography brought to portrait painting, when a craft for the wealthy became a mass market almost overnight. Companies Next in Line are one digitization project away from the Crucible. Companies Holding a Lever can still gain on cost, time and quality in the functions that run on documents and data. Companies In the Balcony should keep learning, because the tools will reach them later. When they do, the gains can add to EBITDA in the year you adopt them, which few capital projects can claim.
How fast do you need to move?
Faster than past technology cycles suggest. Uber launched in 2009. New York City taxi medallions peaked at about $1 million in 2014 and traded at $250,000 or less by 2017. Disruption of that kind usually takes five to seven years to reshape an industry. In 2023 we estimated that firms with heavy WINS exposure had 36 to 60 months to act. That clock has been running since.
How do you run a WINS analysis?
Score every function, geography or process on four measures. The output is a ranked Reinvention Map that shows where to act first and how much money is at stake.
| Measure | How to score it | What it tells you |
|---|---|---|
| WINS intensity (0 to 5) | Share of time spent creating or improving words, images, numbers and sounds | How exposed the work is |
| Digitization (0 to 5) | Inputs are digital, accessible, searchable and held in systems of record | How ready the work is |
| Reinvention priority | WINS intensity × digitization | Where to act first |
| Share of cost base | Salary, vendor spend and overhead tied to WINS work | The size of the prize |
A claims department that scores 5 on intensity and 4 on digitization has a priority of 20 out of 25. A warehouse that scores 1 and 2 has a priority of 2. Start where the priority score and the cost base are both high.
What should you do first?
- Learn the tools yourself. You cannot learn to swim by reading about it. Leaders who use AI every week make better calls about it.
- Appoint a cross-functional team. Have the CEO or board charter experiments at the task, subprocess and process level.
- Find your position. Review your cost drivers and digitization to learn whether you are In the Crucible, Next in Line, Holding a Lever or In the Balcony.
- Match the plan to the position. In the Crucible or Holding a Lever, run a test-and-learn program of 6 to 24 months. Next in Line, digitize your WINS work now. In the Balcony, keep learning.
- Keep a human on high-risk decisions. Models still make things up. Use AI to augment risky work before you automate it, and set clear rules on what data may enter public models.
Questions about WINS
- What does WINS stand for?
- WINS stands for Words, Images, Numbers and Sounds. WINS work is any work that creates or improves one of those four things, from contracts and code to slide decks, financial models and recorded calls.
- Who created the WINS framework?
- Paul Baier, Jimmy Hexter and I introduced WINS in “Where Should Your Company Start with GenAI?” in Harvard Business Review in September 2023. We drew on a community of more than 3,000 generative AI practitioners at GAI Insights.
- Is WINS work the same as knowledge work?
- No. WINS work is narrower and easier to act on. A programmer, an accountant and a marketer do WINS work. A heart surgeon and a chef are knowledge workers, but most of what they deliver is physical, so generative AI reaches them later.
- Can a company with little WINS work ignore AI?
- No. A low WINS score lowers the urgency. It does not remove it. Most companies hold at least one lever, such as bids, customer service or planning, and AI savings there can add to EBITDA in the year you adopt them.
- How is WINS different from RISE?
- WINS tells you where AI can change your economics. RISE tells you how mature your organization is at using AI and what to do at your current stage. Start with WINS to find the prize. Use RISE to plan how you will capture it.
Sources
- Paul Baier, Jimmy Hexter and John J. Sviokla, “Where Should Your Company Start with GenAI?” Harvard Business Review, September 2023.
- GAI Insights, The Integrated AI Playbook: WINS, RISE, PRISM, 2026.
- Related: “The 6 Disciplines Companies Need to Get the Most Out of Gen AI,” Harvard Business Review, July 2024.