Delta 4
If the gap is big enough, people never go back. Build for that gap, not for a little better.
AFTER KUNAL SHAH
The thing you can't un-feel
Think about the last time a product changed you for good. Not 'I liked it.' I mean you genuinely cannot go back. UPI, so you haven't touched an ATM in years. Swiggy, so you haven't called a restaurant to read out your order in a decade. A smartphone, so a keypad phone now feels like a punishment. Nobody made a New Year resolution to switch. It just happened, and the old way quietly died. Kunal Shah, who built FreeCharge and later CRED, has a clean way to predict exactly which products do this. He calls it Delta 4.
The best products don't get adopted. They get un-leavable.
Rate the old way, rate the new way
Here's the whole tool in one move. Take any product. Score the OLD way of getting the job done on a 1-to-10 scale (10 = effortless and delightful, 1 = painful). Then score the NEW way the same. Subtract. That gap is your delta. Calling a restaurant to order food: maybe a 4 (you wait on hold, they mishear, no card payment). Swiggy: maybe a 9 (tap, track, done). Delta of 5. The score is deliberately subjective. It's the felt experience, not a spec sheet.
Two numbers, one subtraction. That's the entire framework.
You've built a note-taking app. It's genuinely nicer than the default Notes app: prettier, slightly faster, a couple of neat features. On the felt 1-to-10 scale, the old Notes app is a 6 for most people. Yours is an 8. A real, honest improvement.
Will people switch to your app and stick?
An 8 beats a 6, but a delta of 2 is below the line where behaviour actually changes. The improvement is real and useless at the same time, because the friction of switching eats the entire two-point gain. Your users do the subtraction subconsciously and conclude it isn't worth it. This is why 'we're better' is the most dangerous sentence a founder can believe.
A small delta is a real improvement that nobody acts on. Better than the default is not a business.
The magic number is 4
Kunal Shah's finding: the gap has to hit roughly 4 or more before the change becomes irreversible. Below 4, people abandon the new thing easily, drift back to the old way, and you have to keep paying to drag them in. At 4 and above, something flips. The old way starts to feel unbearable. Users stop choosing you; they just can't imagine going back. Smartphone over keypad, UPI over cash, delivery over cooking-or-calling. All delta 4-plus. All permanent.
Below 4 you rent your users. At 4-plus you own the habit.
Early Swiggy and early UPI were buggy. Apps crashed, payments failed and you'd refresh praying, a delivery would show up cold or late, support was a black hole. Genuinely rough experiences in the early days.
Why did people forgive all those flaws and keep using them anyway?
This is the hidden superpower of a high delta: it earns you tolerance. When the gap is 4-plus, users absorb your bugs, your bad UX, your thin support, because reverting hurts more than enduring. A low-delta product gets zero forgiveness; one bad experience and they're gone, because there was barely anything to lose. So a big delta doesn't just win users, it buys you the slack to be imperfect while you fix things.
A high delta forgives your flaws. A low delta punishes every one of them.
Big gaps talk for you
There's a second compounding gift. When a product genuinely changes how someone lives, they can't shut up about it. Nobody was paid to tell their family to install UPI or to send their parents the Swiggy app. The delta did the selling. Low-delta products have to buy every single user through ads, because no one is excited enough to recommend a marginal improvement. High-delta products grow on word of mouth, which is the only growth channel that gets cheaper as you scale instead of more expensive.
Low delta you pay for every user. High delta your users recruit the next ones.
Delta 4
Name it so you can use it. DELTA 4: score the old way and the new way on a subjective 1-to-10 experience scale, then subtract. Below 4, the change is reversible. People try it, drift back, and you pay to re-acquire them forever. At 4 or more, the change is irreversible. The old way becomes unbearable, users never voluntarily revert, and you unlock two compounding advantages: tolerance for your flaws, and organic word of mouth. The job of a builder isn't to be a bit better. It's to engineer a gap of 4 or more.
How Delta 4 plugs into value
Delta 4 isn't floating on its own; it's the felt version of value. Hormozi's value equation says perceived value goes up when you raise the dream outcome and the odds of getting there, and crucially when you crush the time and effort it takes. UPI didn't offer a bigger dream than cash; it collapsed the effort and the time to almost zero. That collapse is what produces a high delta. So when you're trying to manufacture a 4, the fastest lever is usually the denominator: make the job dramatically faster and dramatically easier, not slightly fancier.
Delta is what value feels like from the user's chair.
You can spend the next three months on ONE of two things for your product. Option A: make the experience you already have noticeably nicer, pushing your delta from 2 to 3. Option B: rethink the job entirely and chase one feature that could push your delta from 2 to 5, but it's harder and might not land.
Which is the better bet?
Delta isn't linear in payoff; it has a cliff at 4. A 2 and a 3 are basically the same business (people churn), so a guaranteed move from 2 to 3 buys you almost nothing real. The whole prize lives past the cliff. That means the rational move is to take the riskier path that has a chance of clearing 4, rather than the safe path that politely stops short. Incrementalism is the trap that feels responsible and quietly kills you.
Don't optimise toward 3. Gamble toward 4. The payoff is a cliff, not a slope.
People don't switch because you're better. They switch because going back becomes unthinkable.
Where I use this
I run Delta 4 as a filter before I build anything. Most ideas that excite me on a whiteboard are secretly delta-2 ideas: a prettier version of something that already works. Those I kill, because I've learned the market kills them for me anyway, just slower and more expensively. The ones I keep are the ones where I can honestly say the old way scores a 4 and mine scores a 9. Same discipline applies to a pitch, a landing page, a service: if the gap over what they do today isn't huge, I don't have a product, I have a feature nobody asked for.
If you can't name the old way and score the gap, you don't have an idea yet.
The honesty trap
One warning, because this is where people fool themselves. You will be tempted to rate your own product a 9 and the old way a 3, and declare a delta of 6 from your desk. That number is worthless. The score lives in the user's head, not yours. The old way is whatever they actually do today, including the workarounds they're weirdly fine with. The real test is brutal and simple: take it away from a few real users for a week. If they shrug, your delta was always small. If they riot, you've got a 4.
You don't grade your own delta. Your users do, with their feet.
- Delta 4: score the old way and the new way on a felt 1-to-10 scale, then subtract. The gap is the delta.
- Below a delta of 4 the change is reversible. People try it, drift back, and you pay to re-acquire them forever.
- At 4 or more the change is irreversible. The old way becomes unbearable and users never voluntarily revert.
- A high delta earns two compounding gifts: tolerance for your flaws, and organic word of mouth that gets cheaper as you grow.
- The payoff is a cliff at 4, not a smooth slope. A safe move from 2 to 3 buys almost nothing; gamble toward 4.
- The fastest lever to manufacture a delta is usually collapsing time and effort, not adding polish.
- You don't grade your own delta. The user does, with their feet. Take it away and see if they riot.
Your turn
Pick one product or service you're building, considering, or just obsessed with. Write down the OLD way the user solves this job today (the real one, including their lazy workaround), and score it 1-10 on felt experience. Then score your new way the same. Subtract. If the gap is under 4, do not polish it; instead write one specific change to the job itself, usually a big cut to time or effort, that could push the gap past 4. If you genuinely can't find one, that's your answer: it's a feature, not a business, and you just saved yourself six months.
Score the gap before you write a line of code. Under 4 is a no until you can change the answer.
Want the whole story, the long version? Read the deep-dive essay
This module stands on Kunal Shah’s work. The words, the examples, and the mistakes are mine.