Your first 10 customers will not come from ads. They come from you, a short list of names and a lot of slightly awkward conversations.
That's good news if you have no budget, because the work is free. It just isn't easy.
I've watched plenty of founders spend their first few lakh on ads before they knew who their buyer was. The ads didn't fail because of the copy. They failed because nobody had sat across from a buyer and asked what hurts, and so the ads were talking to everyone and landing with nobody. Here's how I'd find the first 10 instead, using four ideas from the directory as worked examples.
Start with one buyer, not a market
"Families" is not a buyer. "Working siblings in different cities who split the cost of an ageing parent's care" is.
Take Siblingshare, a shared ledger where brothers and sisters agree their shares once, log every bill and settle up by UPI directly with each other. The need is real. India had 149 million people over 60 in 2022, heading for 347 million by 2050 (source: UNFPA and IIPS India Ageing Report, 2023), and a lot of them depend on children who live elsewhere. But you can't sell to 149 million people. You can sell to the one sibling who lives closest to the parents, pays more than agreed and is quietly annoyed about it.
That person is your first customer. Write down 20 of them by name before you build anything.
Go where your buyer already gathers
If your buyer has a group, a meeting or a queue then that is where you sell. You don't need to build an audience when one already exists.
For Siblingshare, that means housing society groups, the waiting rooms of clinics that treat older patients and the parents' association at a school where half the families have grandparents back home. For a B2B idea it's even more direct. Here's how it looks for the others -
- Liftaudit measures how much braking energy each lift in a tall building wastes, then fits regenerative modules only where the numbers pay back. The buyer is a facility manager or a housing society committee, because lift power sits on the common-area bill. Facility managers already report energy use, and they swap notes at industry meets, so that's your room.
- Dumpcover plants and monitors cover on biomined city dumpsites for five years. Nearly 15,000 acres across 2,479 sites are due for clearance (source: PIB, 2026), and the land belongs to urban local bodies. Your first customers here may not be cities at all. They could be the biomining contractors who need to hand back a finished site.
- Studentpass sells one monthly bus, metro and shared-auto pass to college students through their campus. The college is the channel and the parent is the payer. One principal who says yes can get you a whole batch of students at admission time.
Notice the pattern? In three of the four, the person who signs is not the person who uses it.
Sell by hand, and do things that don't scale
Your first 10 sales should feel almost embarrassingly manual. That is the point.
For Siblingshare, I'd skip the app at first and run three families on a shared sheet and a weekly summary message, with you doing the maths on who owes whom. If they won't pay ₹999–1,999 a year for that (the test price on the page, and it is an assumption) then a polished app won't save you. For Liftaudit, do one cheap audit for a society you can reach through a friend and share the result with that building. The Hong Kong study measured returned energy of 17.1% to 27.0% across one government lift group (source: EMSD Hong Kong, 2026), but your buyer will trust their own building's meter far more than a foreign report.
Manual work also tells you what to build. Every time you do something by hand three times, you've found a feature.
Ask for the sale, then ask for a name
Founders are oddly shy about asking for money. Don't be.
After every conversation, ask two things. Would you pay for this, and how much? And who else do you know with the same problem? The second question is how 3 customers turn into 10, because a housing society secretary knows four other secretaries and a biomining contractor knows which city tenders are coming next. A college principal sits on committees with other principals too.
A "no" is still useful. If three facility managers tell you the lift maker won't support third-party modules, that's a risk the Liftaudit page already flags, and now you've heard it first hand.
Pick the slowest buyer last
Not every buyer moves at the same speed, and with no budget, speed matters a lot.
Dumpcover sells to government, where payment often comes months late and a small firm can run out of cash waiting. Studentpass sells to colleges, which decide slowly and often only at the start of an academic year. Siblingshare sells to families, who can say yes on a phone call. So if you're choosing between ideas with no money in the bank, the fast buyer usually wins, at least for the first 10.
That's not a rule against government or campus work. It's a rule about order. Get a fast customer to pay your bills while the slow one goes through approvals.
What I'd do this week
Here's a plain checklist -
- Write one sentence that names your buyer as a person, not a segment.
- List 20 of them by name, from your phone, LinkedIn and old colleagues.
- Find the one place they already gather and show up there.
- Offer a manual version at a real price.
- After every chat, ask for one more name.
You'll probably need somewhere around 40–60 conversations to land 10 paying customers, and that's a rough guess, not a law. It sounds like a lot until you remember it costs nothing but time.
For more ideas to test this way, browse the startup ideas for India or the ideas you can start under ₹1 crore.
Name one narrow buyer, list 20 of them by name and go where they already meet, like an association, a group chat or a campus. Sell a manual version by hand and ask each person for one more name.
Should I build the product before I find customers?
No. Run a manual version for a handful of buyers first. If they won't pay for the manual version then a finished product usually won't change their mind.
How many conversations does it take to get 10 customers?
There's no fixed number, but plan for roughly 40–60 as a working guess. Referrals from your first few buyers shorten it a lot.
Talk to Yash
If you've got an idea and a list of names but aren't sure who to call first, I can help you pick the buyer and plan those first conversations. 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.