The Delta 4 Framework
Score old vs new solution out of 10; if the efficiency gap isn't 4+, the product won't stick.
- Difficulty
- Easy
- Time to result
- ~days to results
- Steps
- 5
- Confidence
- 97%
Delta 4 replaces the unmeasurable "10x better" heuristic with a scoring exercise anyone can run. Rate the incumbent way of doing a job on efficiency out of 10, rate your product out of 10, and take the difference. Only when the delta is 4 or greater do three effects appear: adoption becomes irreversible, users tolerate failures, and users brag about it — which collapses CAC toward zero.
Origin
Kunal Shah's own framework, developed as a philosophy major reverse-engineering why his first startup FreeCharge (exited ~$450M in 2015) succeeded when academically stronger peers did not. He credits its intellectual roots to studies of entropy and evolutionary biology. He notes US VCs including Sequoia now use it in analyst training.
Core principles
- 01"10x better" is not measurable; a 0-10 efficiency score is.
- 02Efficiency is judged end-to-end by the user, not by how much technology is involved.
- 03A delta of 4 or more makes switching irreversible.
- 04Delta 4 products earn a Unique Brag-worthy Proposition (UBP), so distribution is word-of-mouth, not paid.
- 05Below Delta 4, users have zero tolerance for failure and will revert to the old way.
- 06Applies at three altitudes: feature, product, and business.
How to run it
- 1
Name the incumbent behaviour precisely
Define exactly how the user solves the job today without you — hailing a street cab, buying a suit in a store, wrangling Excel formulas. The comparison must be the real alternative, not a strawman.
Pro tip Ask actual users to describe the current way in their own words before you score anything.
Watch out Founders routinely compare against a worse incumbent than the one their users actually use.
- 2
Score the incumbent on efficiency out of 10
Have users (not you) rate the existing solution's overall efficiency from 1 to 10. Street cab: roughly a 3. Buying a suit offline: roughly a 5.
Pro tip Score the whole experience including the end result, not just the moment of transaction.
- 3
Score your product on the same scale
Have the same users rate your solution 1-10. Uber: roughly a 9. Buying a suit online: still roughly a 5 because the suit doesn't fit.
Pro tip If you cannot get anyone but yourself to score it, you are being delusional — the exact failure mode the framework exists to catch.
Watch out Being a 'tech product' guarantees nothing; digitising a workflow can leave efficiency flat or worse.
- 4
Compute the delta and read the verdict
Subtract. Delta of 4 or more: expect irreversibility, high failure tolerance, and organic brag-driven growth. Delta under 4: expect reversion, zero failure tolerance, no word of mouth, and CAC you will have to pay for forever.
Pro tip Test the brag proxy directly: would a user spontaneously demo this to a friend? If not, the delta isn't there.
Watch out A sub-4 delta cannot be rescued with marketing spend; it can only be rescued by changing the product.
- 5
Re-run at feature and business level
Apply the same scoring to individual features and to the business model, not just the headline product. Kill or rebuild anything that scores under 4 against the alternative it displaces.
In the wild
Shah asked Lenny to score both. Lenny scored the old cab a 3 and Uber a 9 — a delta of 6. Shah pointed out Lenny didn't discover Uber through an ad or performance marketing; someone showed him a demo and he said 'my god, this is crazy.'
→ Delta 4 cleared: irreversible adoption, tolerance for the occasional bad ride, and near-zero CAC via word of mouth.
A fully digital experience with all the features, yet Lenny scored online and offline both around 5 — the suit he bought online didn't fit at all. Being a tech product broke the assumption that tech implies efficiency.
→ Delta of zero: reversible, no bragging, zero tolerance for error. Shah's verdict: 'you will never going to buy a suit online.'
Shah reads the efficiency gap of doing knowledge work before vs after LLMs as a Delta 4 event — 'feels like magic' — which is why it spread without advertising.
→ Organic, brag-driven adoption at unprecedented speed, consistent with the framework's prediction.
Common mistakes
Scoring your own product yourself
The whole point is escaping delusion. If the founder supplies both numbers, the delta is fiction. The scores must come from users.
Assuming digital equals more efficient
Putting a workflow online adds features but can leave the end-to-end efficiency identical or worse (the suit). Tech is not a delta.
Trying to buy growth for a sub-4 product
Low-delta products have no brag-worthy proposition, so CAC stays permanently high. Paid growth masks the product problem until the money stops.
Is it for you?
Best for
Founders and PMs deciding whether a product or feature is actually worth shipping — and investors sanity-checking why a 'better' product isn't taking off.
Not ideal for
Products whose value is status, taste, or identity rather than efficiency, where a user would not score 'efficiency' as the operative axis.
From the transcript
“every time you see that the the product efficiency delta is greater than equal to four, three things happen.”
“So, every time humans unlock a delta four product or service, they cannot stop talking or sharing about it. And therefore, all Delta 4 products…”
“when the Delta is lesser than four, uh what you will discover is that it is reversible.”
From the episode
Kunal Shah on winning in India, second-order thinking, the philosophy of startups, and more