The Two-of-Three Inflection Test
Only build a new zero-to-one product when at least two of three inflections line up
- Difficulty
- Moderate
- Time to result
- ~weeks to results
- Steps
- 4
- Confidence
- 95%
A go/no-go filter for zero-to-one products. A genuinely great new-category product needs at least two of three simultaneous inflection points converging: a step-function in the technology, a shift in consumer behavior, and a shift in the business model. One alone is not enough; a better interface bolted onto no underlying shift will not carry a product.
Origin
Aparna Chennapragada (CPO, Microsoft; ex-Google Lens, ex-Robinhood) formalized this from repeatedly building new-category products the hard way.
Core principles
- 01A better interface alone will not make people switch — you need an underlying shift
- 02Two of three inflections is the bar for a really good new product
- 03Being early is the same as being wrong: the tech inflection must actually be present, not imminent
- 04'Why now?' is the investor's real question and this answers it
How to run it
- 1
Check for a technology step-function
Is there a discontinuous jump in what the technology can do? Deep learning enabled Google Lens; speech recognition enabled conversational search; LLMs and reasoning models are today's step-function.
Pro tip Name the specific capability that just crossed a threshold, not a general 'AI is better' hand-wave.
- 2
Check for a consumer behavior shift
Has user behavior changed at scale in a way the product can ride? When storage became free and phones were everywhere, people photographed everything, turning the camera into a keyboard for the real world — the behavioral basis for Lens.
Pro tip Look for order-of-magnitude changes in an everyday behavior, then ask what new value can be extracted from that volume.
- 3
Check for a business-model shift
Is there a natural inflection in how the thing gets monetized? Second-price auctions and CPCs for search; per-seat for SaaS; Robinhood's zero-fee model; and now usage- and outcome-based AI monetization.
Pro tip For AI specifically, weigh seat vs usage vs outcome-based ('have you solved the problem for me') pricing as a possible edge.
- 4
Require two of three before committing
If fewer than two inflections are genuinely present, hold. The combination is what unlocks a category; a single factor is not a strong enough tailwind.
Watch out Don't count an inflection you merely believe is coming — the intelligence has to be there now, or you are early, which equals wrong.
In the wild
Deep learning (tech), the shift to photographing everything once storage was free (behavior) converged, making the camera a viable input for the real world.
→ A defensible new-category product built on aligned inflections rather than on a novel interface alone.
Smartphones as the primary means to run a finance app (behavior) plus the zero-fee business model converged.
→ Chennapragada cites it as a case where the combination — not any single factor — unlocked the product.
Common mistakes
Betting on interface alone
Assuming a slicker, more intuitive interface will make users switch without any underlying tech, behavior, or model shift.
Mistaking imminent for present
Counting a technology inflection that hasn't actually arrived yet — 'being early is the same as being wrong.'
Is it for you?
Best for
Founders and product leaders evaluating whether now is the right time to start a new-category zero-to-one product
Not ideal for
Incremental features or optimizations inside a mature product where the inflections are irrelevant
From the transcript
“you do want to look for at least two out of these three factors uh inflection points here if you want to make a really…”
“Number one is there a shift is a step function in the tech”
“the second factor that we should look for is what is the consumer behavior shift”
“the third inflection point particularly I would say in enterprise but also in consumer is the business model shift”
“you can't just say, oh, we'll just have a much much more better intuitive interface and hope that uh you know, people will switch to…”
From the episode
Microsoft CPO: If you aren’t prototyping with AI, you’re doing it wrong
Aparna Chennapragada