Your company has more AI projects than it can manage. Which of them deserve the money?
Most large companies now run dozens of AI programs. Different teams started them, different budgets pay for them, and few of them can show a measured result. The AI Radar puts every program on one page and measures the portfolio against the strategy you have already set.
One Radar on your own business. Up to three documents. Work email required.
| Dimension | Ring 1 | Ring 2 | Ring 3 | Composite | Tier |
|---|---|---|---|---|---|
| Marketing | 4 | 2 | 2 | 2.52 | Drop |
| Sales | 4 | 3 | 3 | 3.30 | Wedge |
| Pricing | 3 | 3 | 5 | 3.56 | Wedge |
| Discovery | 3 | 4 | 5 | 3.91 | Selective |
| Development | 4 | 4 | 4 | 4.00 | Selective |
| Embedded Intelligence | 4 | 4 | 5 | 4.31 | Concentrate |
| Supply Chain | 3 | 5 | 4 | 3.91 | Selective |
| Operations | 4 | 5 | 4 | 4.31 | Concentrate |
| Customer Service | 4 | 4 | 5 | 4.31 | Concentrate |
| People | 4 | 3 | 3 | 3.30 | Wedge |
| Finance | 4 | 2 | 1 | 2.00 | Veto |
| Risk | 3 | 3 | 3 | 3.00 | Wedge |
The four questions
The AI Radar answers four questions. Each answer is one number between 0 and 100, with a plain sentence that explains it.
| Question | Index | What it tells you |
|---|---|---|
| Are we investing enough? | AI Investment Index (AII) | Your actual AI spend as a share of what your stated ambition requires. |
| Is the investment in the right places? | AI Alignment Index (AAI) | The share of your AI spend that matches the priorities in your strategy. |
| Are the programs mature? | AI Maturity Index (AMI) | The share of your AI spend in programs that are in production with a measured result. |
| Is each capability built at the right level? | AI Layer Index (ALI) | For companies with several businesses: the share of spend that is built once where it should be shared, and in the business where it should be local. |
How the Radar works
- It starts from your strategy. The Radar reads your business strategy before it looks at any AI program. This order keeps the reading independent of what you already spend.
- It scores twelve parts of your business. The twelve dimensions cover demand (Marketing, Sales, Pricing), the offering (Discovery, Development, Embedded Intelligence), delivery (Supply Chain, Operations, Customer Service), and the enabling functions (People, Finance, Risk).
- It applies three tests to each one. Can AI do this job today? Does your industry reward it? Does your strategy value it? A dimension must pass all three to become a priority. The result is your recommended allocation, drawn as a polygon.
- It reads your programs as one ledger. Each program gets a home, a stage from 1 to 5, and a work mode: AI that assists a person, AI that runs a workflow, or AI inside a new offering.
- It compares the two. The four indices measure the distance between where your money is and where your strategy says it should be.
The twelve dimensions and the job each one does
| Anchor | Dimension | Job AI is hired to do | Outcome metric | Lag (months) |
|---|---|---|---|---|
| Demand | Marketing | Create qualified demand | Pipeline value | 3 to 6 |
| Demand | Sales | Convert demand into revenue and expand accounts | Win rate, deal velocity | 3 to 6 |
| Demand | Pricing | Capture more value per unit | Realized price | 3 to 9 |
| Offering | Discovery | Decide what to build | Hit rate of launches | 12 to 36 |
| Offering | Development | Build and ship the offering and the software the firm runs on | Time to release | 6 to 18 |
| Offering | Embedded Intelligence | Put AI inside what the customer buys | Willingness to pay, attach rate | 12 to 36 |
| Delivery | Supply Chain | Source and plan | Forecast error, fill rate | 6 to 18 |
| Delivery | Operations | Make the product or perform the service, and assure it | Throughput, yield, defect rate | 6 to 18 |
| Delivery | Customer Service | Support customers after the sale | Gross retention, cost to serve | 6 to 12 |
| Enabling | People | Hire, develop, and support employees | Time to hire, employee support resolution time | 3 to 12 |
| Enabling | Finance | Run the books, plan, allocate capital | Close cycle time, forecast accuracy | 6 to 12 |
| Enabling | Risk | Stay legal, safe, secure, compliant | Loss rate, incidents | 6 to 18 |
Stages and work modes
Each program is placed on one of five stages, and the stages are absolute: Stage 3 means the same thing in a bank as in a retailer. A program is placed at the highest stage whose evidence it can show. The line between Stage 2 and Stage 3 matters most, because a control group and a measurement separate a pilot that can be judged from a pilot that can only be described.
- Assist.A person does the task and the AI helps.
- Run.The AI executes the workflow and a person supervises by exception.
- Invent.The AI is part of a new offering that the customer buys.
What you get
- The polygon.Your recommended allocation across the twelve dimensions, with the dimensions to refuse stated as clearly as the dimensions to fund.
- The four indices.Each one comes with a range and one plain sentence.
- A verdict on each program.Grow, hold, fix or stop, with the next milestone and its date.
- A full report.Written for the executive team and the board, with every score traced to dated evidence.
- A baseline you can refresh.Each refresh shows what moved since the last reading.
For companies with several businesses
A business unit radar answers the questions for one business. The parent company has two more decisions: how to divide AI investment among its businesses, and which capabilities to build once for all of them. The corporate radar rolls the business radars up into one view. It shows where business units build the same capability twice, which proven programs other units can reuse, and which shared programs belong in one business. With one business, the corporate radar and the business radar are the same.
| Dimension | Smart Home Solutions (weight 0.80) | Building Technology (weight 0.09) | Robotics and Automation (weight 0.11) | Corporate | Corporate tier |
|---|---|---|---|---|---|
| Marketing | 4.00 | 2.52 | 2.52 | 3.70 | Wedge |
| Sales | 4.00 | 4.00 | 3.30 | 3.92 | Selective |
| Pricing | 2.88 | 3.00 | 2.62 | 2.87 | Drop |
| Discovery | 3.63 | 3.00 | 3.63 | 3.57 | Wedge |
| Development | 4.00 | 4.00 | 4.64 | 4.07 | Selective |
| Embedded Intelligence | 4.31 | 4.64 | 4.64 | 4.38 | Concentrate |
| Supply Chain | 4.22 | 3.30 | 3.00 | 4.00 | Selective |
| Operations | 4.64 | 4.00 | 3.63 | 4.47 | Concentrate |
| Customer Service | 4.00 | 4.64 | 4.00 | 4.06 | Selective |
| People | inherits 3.63 | inherits 3.63 | inherits 3.63 | 3.63 | Wedge |
| Finance | inherits 3.30 | inherits 3.30 | inherits 3.30 | 3.30 | Wedge |
| Risk | inherits 3.00 | inherits 3.00 | inherits 3.00 | 3.00 | Wedge |
What we ask from you
We ask for three things, and each one feeds an index.
Send the ones that already exist, and write nothing new.
Each program takes one row: its name, one sentence on what it does, who builds it, who uses it, the people and spend on it, and six yes-or-no questions.
Revenue by business, total AI spend if you track it, firm-wide AI licenses, and data platform spend.
The six questions
- Does it have a committed budget?
- Do people use it in live work?
- Does it have a baseline or a control group?
- Has it reported a measured result?
- Does another program build on it?
- Does it act without a person who approves each action?
You describe your programs in plain words. The Radar does the classification.
Sample Radars
See the Radar applied to public companies.
A business radar for an industrial company whose AI is in the machine and on the assembly line.
A corporate radar for a company with several businesses, with programs tagged by the level that builds them.
A fictional industrial group, with all four indices computed from a complete program ledger.
Sample Radars for real companies use public sources only, so they show the recommended allocation and the stage of each program. They do not show the indices that need spend data. A Radar built on your own documents and program list is more precise.
How the AI Radar relates to other methods
Leading consulting firms have shown what good AI investment looks like: concentrate on a few priorities, redesign the work, and move past the pilot. The AI Radar measures how far your own portfolio is from that standard, in your own money, against your own strategy.
The AI Radar is the proprietary framework of Professor Mohanbir Sawhney. It is not related to the BCG AI Radar survey or to the Gartner AI Opportunity Radar.
Who it is for, and the framework behind it
Who it is for. Chief AI Officers, CIOs, CTOs, CFOs, strategy leaders, and the consultants who advise them.
About the framework. The AI Radar is based on the research of Mohanbir Sawhney, McCormick Foundation Professor of Technology at the Kellogg School of Management, Northwestern University. It descends from the Innovation Radar, first published in MIT Sloan Management Review in 2006. The method is set out in the working paper "The AI Radar: A Portfolio Framework for Diagnosing Enterprise AI Investment, from the Business Unit to the Corporation" (version 5.0, November 2026).
What is new in version 5.0
- Four indices in your own currency.Investment, alignment, maturity and layer replace the two earlier indices.
- A corporate radar.Companies with several businesses now get one view across all of them.
- Twelve revised dimensions.The names now work in every industry, and each program has one clear home.
- A lighter data request.One program list and six yes-or-no questions replace the earlier portfolio template.
- Three work modes.Assist, run and invent show how much of the work AI does.
Pricing
14 days, one Radar, work email required
Billed yearly in advance
Billed yearly in advance
Corporate radar: twice the price of a business Radar, for up to five business units. Full pricing →
One Radar on your own business. Up to three documents. Work email required.
The AI Radar is the proprietary framework of Professor Mohanbir Sawhney. It is not related to the BCG AI Radar survey or to the Gartner AI Opportunity Radar.