framework · strategy
Ansoff matrix for product management
Best for: Strategy interview questions of the form "should company X enter market Y?" or "how should company X grow?" Use it to eliminate options quickly and commit to a recommendation, not to enumerate all four quadrants.
The Ansoff matrix is not a checklist. It is an elimination tool. Igor Ansoff published it in the Harvard Business Review in 1957 to map growth options against risk, and the core logic has not changed: the further you move from what you already know on either axis (product or market), the higher the risk. In a PM interview, the matrix is useful only if you use it to cut options and arrive at a recommendation. Candidates who enumerate all four quadrants without committing have just demonstrated framework recall. That does not clear the bar.
The four quadrants and the risk gradient
Market penetration: existing product, existing market. Sell more of what you have to the people who already know you. The lowest-risk quadrant because both the product and the market are understood. The strategic question is competitive: in a winner-takes-most market, penetration has a very different risk profile than in a fragmented one. If your moat is distribution or brand, penetration is often the default right answer.
Market development: existing product, new market. Take what you have and bring it to a different segment, geography, or channel. One axis moves, so risk is moderate. The real question is whether the product travels. Does it fit the regulatory environment, the buying behavior, and the workflow of the new market without significant rework?
Product development: new product, existing market. Build something new for people who already trust you. One axis moves. You have distribution and customer insight, but you are making a product bet. The risk is that the new product fails to achieve fit in a market where users have high expectations because they already know your brand.
Diversification: new product, new market. Both axes move. Highest risk. This is the only quadrant where two types matter. Related diversification means you carry existing capabilities into the new space (Apple from Mac to iPhone: new market segment, but deep hardware and software expertise). Unrelated diversification is a conglomerate move (Amazon from e-commerce to AWS to healthcare): the synergy is capital and distribution, not domain expertise. In most PM interviews, the correct answer is not diversification unless the company has an unusual asset base or the market opportunity is too large to ignore.
Worked example: Nvidia at Computex, June 2026
Nvidia is worth mapping because all four quadrants are visible in a single company’s announced strategy at one point in time.
Selling more Vera Rubin GPUs to existing hyperscaler customers (Google, Microsoft, Meta) is market penetration: the same customers buying more of the same product category. Extending the Vera Rubin platform as an integrated AI cluster stack to existing GPU buyers is product development: new product, same enterprise customers. Pursuing sovereign-AI contracts and enterprise data-center build-outs in markets where Nvidia has not historically sold direct is market development: existing hardware, new buyers. RTX Spark, the consumer AI inference chip for mainstream Windows PCs, is diversification: a new product for a consumer market that Nvidia has never owned despite being in gaming GPUs. That one example, with a real company and specific product lines, is more useful in an interview than a generic reference to Coca-Cola.
Use it, do not recite it
The most common failure mode: the candidate walks through all four quadrants as potential options for the company in question and then says “it depends.” That signals inability to use the framework, not humility.
Strong candidates do three things. First, they frame the tool upfront: “I’ll use the Ansoff matrix to map growth options against risk and then filter by what this company is actually positioned to win.” Second, they eliminate quickly. Diversification is usually off the table unless a specific asset makes it rational; say that out loud and move on. Third, they run a viability check on the remaining candidates: addressable market size, competitive intensity in that space, the company’s capability fit, and expected payback timeline. Then they commit to one quadrant and name the primary risk.
The complementary check before locking in a quadrant: TAM/SAM/SOM to validate market size, and a quick scan of competitive intensity using Porter’s five forces if the interviewer probes on the competitive angle. Do not reach for SWOT as a substitute for committing. SWOT is useful for stress-testing a recommendation after you have one, not for choosing between quadrants.
One dimension the matrix omits entirely: distribution and channel. For digital and AI products, the question is often not whether the product can reach a new market but whether the company has the channel to acquire and retain users there. When evaluating market development or diversification options, add a distribution question explicitly: “Do we have the channel, or would we be buying into a market where we start from zero on go-to-market?”
Strong and weak answers in an interview
strong
"I'll use the Ansoff matrix to map the growth options against risk, then filter by what [company] is actually positioned to win. Their current assets are: strong brand recognition with [existing users], proven monetization in [core segment], and distribution through [channel]. That rules out unrelated diversification immediately: they'd be starting from zero on both axes with no clear synergy, and nothing in their asset base earns them a right to win there. The real choice is between market development and product development. The question is whether their current product travels to the adjacent segment, or whether that segment needs something different enough to require a dedicated build. Given [specific insight about target users], I'd lean market development: the product works, the risk is distribution and regulatory fit, not product-market fit. My recommendation is market development with a scoped pilot in [new segment] to validate the regulatory and buying-behavior assumptions before full investment. If the pilot shows the product needs significant adaptation, that shifts the calculus toward product development, which changes whether we build internally or partner."
weak
"The Ansoff matrix has four quadrants. Market penetration is selling existing products to existing markets. Market development is new markets with existing products. Product development is new products for existing markets. Diversification is both new." Then the candidate lists all four as potential options. This fails because it demonstrates recall, not judgment. The interviewer already knows the quadrant definitions. Enumerating them without eliminating irrelevant options and committing to a recommendation signals the candidate can describe a tool but not use one. A strong answer picks a lane within 60 seconds and defends it against follow-up.
The 2026 reframe: feasibility is no longer the constraint
The classic framing of Ansoff risk treated the “new product” axis as primarily a build-cost and engineering-complexity problem. In 2026, that assumption does not hold. An AI-era PM can ship a working product into an adjacent market in weeks. Feasibility has compressed almost to zero as a constraint.
This reshapes which questions matter. The “new product” axis is no longer about whether you can build it. It is about whether you can build something the new user will actually choose over whatever they are already using. That is a lovability question: does it meet users where they work, does it anticipate their workflow, and does it earn adoption in a context you do not yet understand well? You can ship technically in 90 days and still fail because the product does not fit the user’s actual environment.
The viable question is unchanged and is still the first filter: is the market large enough that it can cover costs and generate profit, and will someone pay for the solution at a price that works for both sides? Size the market before committing to any quadrant.
Practically for interview answers: when you evaluate a new-product or new-market quadrant, replace “can we build it?” with “do we understand this user well enough to build something they will choose?” That is a discovery and distribution problem. Name it that way. It shows you understand what feasibility being nearly free actually changes about growth strategy, and it is a reframe none of the standard Ansoff content makes.