Most startup ideas don't fail because the product is bad. They fail because one assumption nobody tested turned out to be wrong.
So when founders ask me how to validate a startup idea, I don't start with surveys or landing pages. I start with a question - what's the one thing that, if it's false, kills this? Then you test that thing, cheaply, before you spend a rupee on anything else. Here's how that works, with four ideas from the directory as examples.
Step 1: find the killer assumption
Every idea rests on a handful of beliefs. Usually one of them carries most of the weight, and that's where you start.
Take Dimmerhall. It fits controllers to the flexible loads in big housing societies, like water pumps, sewage plants, shared EV chargers and opt-in ACs, and moves that load into cheaper solar hours. The society pays from its savings.
What's the killer assumption? That the savings are big enough to care about. The page names thin savings as the main risk, and it's right. If a society saves a small amount a month then no committee will sit through a vendor pitch for it, however clever the tech is.
Step 2: test it without building anything
You don't need controllers to test Dimmerhall. You need a society's electricity bills, its tariff and an honest guess at which loads can shift.
Sit with one society treasurer and work out the number on paper. If the saving looks real then ask whether they'd sign a letter of intent at that price. If it doesn't then you've learnt something worth far more than a prototype, and it cost you an afternoon.
That's the whole trick. A validation test should cost hours or a few thousand rupees, not lakhs.
Step 3: talk to the person who pays
Founders love talking to users. Users are lovely, but they don't always pay.
Enzymark is a lab test for one gene, DPYD, that tells an oncologist whether a patient may break down certain chemotherapy drugs too slowly and suffer severe toxicity. Patients benefit. Oncologists order it. Patients pay a modest fee.
So who do you validate with? The oncologist, because if they don't order it then nothing happens. The page names "no mandate" as the main risk, meaning nothing forces a doctor to test. Your interviews need to find out whether oncology day-care centres would build it into their routine without being told to, and that's a different conversation from asking patients whether it sounds useful.
Step 4: set pass and fail targets before you start
This is the step people skip, and it's the one I care about most.
Before you run any test, write down what counts as a pass. Something like "3 of 10 interviewed buyers agree to a paid pilot at ₹X". If you set the bar after you see the results then you'll always find a way to call it a win.
Eaveswell is a good one to practise on. It sells owners of older houses one yearly plan with a pre-monsoon condition survey, scheduled servicing of roof, drains, tanks and wiring, and fixed labour rates. Capital is ₹55–95 L.
A fair pass target might be a number of pre-paid survey bookings in one neighbourhood before the monsoon. The page names the seasonal crunch as the risk, so your test should also check whether you can deliver all those surveys in the same few weeks. Can you? Better to find out with ten houses than a thousand.
Step 5: sell before you build
The strongest signal is money, or at least a signature.
Moveproof gives brokers and property managers a guided walkthrough that records a rented home's condition on move-in and move-out day, so the deposit settles on evidence. The agent pays per report, and capital is ₹30–60 L.
You could test this one with a printed checklist, a phone camera and a shared folder. Do five reports by hand for one broker and charge for them. If the broker comes back for a sixth then you have something real. The page says agent adoption is the risk, so that repeat booking is exactly the signal you want.
Common mistakes I see
- Asking friends. They'll be kind, which is useless.
- Asking "would you use this?" Everyone says yes. Ask what they do today and what it costs them.
- Building the app first. If the manual version doesn't sell then the app won't either.
- Testing five things at once. You won't know which one failed.
How much should validation cost?
There's no fixed number, but I'd keep it to a small slice of your launch budget. For an idea that needs ₹30–60 L to start, spending a few thousand rupees and two or three weeks before you commit seems like a steal to me.
Every idea in the directory has a free validation kit with a 14-day sprint and pass or fail targets, which is a good template to borrow. You'll find more lower-budget ideas on the startup ideas under 1 crore hub and the main startup ideas in India list.
Questions founders ask
Find the one assumption that kills the idea if it's false, set a pass target and test it with real buyers in two or three weeks.
How many customer interviews do I need to validate an idea?
Around 10 with the person who pays is a sensible start. Stop earlier if the answers are clearly no.
Is a landing page enough to validate a startup idea?
Rarely on its own. Sign-ups are cheap. A paid pilot or a signed letter of intent tells you far more.
Talk to Yash
If you're trying to work out how to validate a startup idea and can't pin down the killer assumption, that's a short conversation I'm happy to have. Talk to Yash →
For illustration only. Ideas are hypothetical and names fictional; sourced facts may be out of date. Nothing here is legal, financial or professional advice, and Yash Khatri is not personally liable for any use of it. Read the full disclaimer.
For illustration only. Names are fictional and ideas hypothetical; sourced facts are real but may be out of date. Nothing here is legal, financial or professional advice, and Yash Khatri is not personally liable for any use of it. Read the full disclaimer.