How to Assess AI Feasibility: A Product Manager's Guide
Includes Two Free Templates to Assess Your Own AI Product Feasibility
You're likely inundated with ideas for AI-powered features and products. But which ones are actually feasible? How do you cut through the hype to identify the AI applications that are both technically achievable and likely to deliver real business value?
In this post, I will walk through a practical framework for assessing AI feasibility. By systematically evaluating AI ideas against a set of key criteria, you can prioritise the initiatives that have the best chance of success and make more informed decisions about where to invest your AI resources.
This is a great companion to my post on Productising AI, check it out if you haven’t already. I would recommend doing this AI feasibility analysis as part of any AI product development process, usually in the discovery phase.
The AI Feasibility Matrix
Central to this assessment process is what I call the AI Feasibility Matrix. This is a tool for scoring AI ideas on two critical dimensions:
Technical Feasibility: How achievable is this AI application given the current state of technology? Do we have the right data, algorithms, and infrastructure in place to bring it to life?
Business Impact: How much value will this AI application deliver to the business? Will it drive efficiency, reduce costs, improve the customer experience, or open up new revenue streams?
By plotting AI ideas on this 2x2 matrix, we can quickly visualise which ones have the highest potential and deserve our attention.
Here's what each quadrant represents:
❌ Low Technical Feasibility, Low Business Impact: AI applications that fall into this category are low-priority. They're difficult to achieve with current technology and unlikely to deliver significant business value. Avoid allocating resources here.
⚠️ High Technical Feasibility, Low Business Impact: These AI applications are technically achievable but won't move the needle for the business. They're tempting to pursue because they're within reach, but they're ultimately a distraction from higher-value initiatives. Only consider these if they are incredibly easy and could be used to prove a concept as part of a wider push towards AI projects.
🔎 Low Technical Feasibility, High Business Impact: AI applications in this quadrant could be game-changers for the business, but the technology isn't ready yet. Keep these on your radar and reassess as AI capabilities evolve or consider gathering more information to create a compelling business case for investment in the project.
✅ High Technical Feasibility, High Business Impact: This is the sweet spot. AI applications that land here are both technically achievable and likely to deliver significant business value. These are the initiatives to prioritise and rally resources around.
I’ve created a free template that you can use to map out your own ideas, by yourself, or as part of a collaborative workshop.
AI Feasibility Assessment
To take this idea a step further, consider performing an AI Feasibility Assessment as part of your product discovery/AI PoC process. An AI Feasibility Assessment goes beyond the 2x2 grid and provides you with concrete scoring to compare and contrast a range of ideas. Scores are assigned on a 1-5 scale against the following factors.
Technical Feasibility:
Data availability and quality.
Technical complexity.
AI capabilities.
Internal AI expertise.
Business Impact:
Strategic alignment.
Potential ROI.
User experience impact.
Competitive advantage.
Risk and compliance.
For each factor, assign a score from 1 (low) to 5 (high). Then average the scores and rank your ideas from High Feasibility to Low Feasibility.
📃Free Template
To help automate the scoring in the assessment I have created a free Google Sheets template. Simply list out your AI ideas across the top, and then systematically rate each one against the Technical Feasibility and Business Impact criteria.
The built-in formulas will automatically calculate the average scores and overall feasibility ranking, and a “guidance” tab goes into more detail on each factor and how to rate your own ideas.
Assessing AI Ideas in Practice
Let's walk through an example to see how this assessment might play out in practice. Imagine your team is considering an AI-powered chatbot for customer support.
On the technical side, you have a robust customer interaction dataset to train the chatbot, and there are proven NLP models for handling common queries. For an initial PoC, you could even consider an “off-the-shelf” solution via the OpenAI API, Amazon Bedrock, or another AI platform. You have experienced engineers on staff who can adapt these models to your domain. The technical pieces are in place. You give it a 4 on technical feasibility.
On the business side, a chatbot could significantly reduce support costs while improving response times. It aligns with the company's "customer-first" values and would put you ahead of competitors. There are some concerns about edge cases and liability, but on balance, the business impact looks strong. You give it a 5 on business impact.
The assessment sheet generates a “High Feasibility” score, and plotting this on the matrix, the chatbot idea lands squarely in the high feasibility, high impact quadrant. This signals that it's a strong candidate to prioritise and pursue.
Going through this exercise for each AI idea under consideration provides an objective basis for comparing them. It separates the truly promising opportunities from those that need more research or aren’t possible.
The Path Forward
Of course, even high-feasibility, high-impact AI applications aren't guaranteed successes. There are still significant hurdles to bringing them to life and realising their full potential.
But, by using the AI Feasibility Matrix to systematically evaluate ideas upfront, product teams can focus their efforts on the AI initiatives that are most likely to succeed. They can more confidently make the case for AI investments and rally stakeholders teams around a shared vision.
As you embark on your own AI feasibility assessments, remember that the goal isn't perfect precision. It's to provide a structured way to think through the key considerations and make more informed decisions. Use the matrix as a starting point, and adapt the criteria to fit your specific context and needs.