Excerpt: Most ideas fail not because they were bad, but because no one checked they were needed. Here is a simple way to test an idea before spending time or money on it, and how one Chennai company began.

"Make something people want."

Paul Graham, motto of Y Combinator

In our last post, we described what the E-Cell SCSVMV team is building: a pipeline that carries an idea from first curiosity to public showcase. This post covers the first thing any student can do to move an idea along. It does not need a budget, a team or a prototype. It needs evidence.

Start with the problem, not the product

Write your idea as one sentence: "[A specific person] struggles with [a specific problem] when [a specific situation]." If the sentence will not come together, the idea is not ready for testing yet.

"First-year students struggle to find reliable notes in the week before exams" can be tested. "An app for students" cannot.

How Freshworks began

In 2010, Girish Mathrubootham was working at Zoho in Chennai. A poor customer-service experience, after a television was damaged while he was relocating, left him frustrated. Soon after, a discussion on the tech forum Hacker News about rising helpdesk software prices showed him that many others felt the same pain.

He and his Zoho colleague Shan Krishnasamy left their jobs in October 2010 and launched Freshdesk in Chennai. The company later became Freshworks, reached unicorn status in 2018, and listed on Nasdaq in September 2021.

Two lessons stand out. The idea came from a problem he had personally lived through. And he found a public signal that others shared it, rather than trusting his own frustration alone. It also began in Chennai, not Silicon Valley, a reminder that a global company can start close to home.

Ten conversations before any code

The most useful test costs nothing: talk to people who actually have the problem. Aim for ten.

  1. Choose the right people. Talk to those who match your sentence, not friends who will be polite.

  2. Ask about the past, not the future. "Tell me about the last time this happened. What did you do?" beats "Would you use my app?"

  3. Listen for workarounds. If people already spend time or money patching the problem, it is real.

  4. Write down their exact words. Your memory will quietly edit them.

  5. Look for patterns. If seven of ten describe the same pain without prompting, you have a signal.

Mistakes to avoid

  • Treating compliments as evidence. "That's a cool idea" costs the speaker nothing. A sign-up, a pilot, or someone offering their time is stronger.

  • Leading the witness. If your question contains the answer, you have learned nothing.

  • Building first. Months of work on something nobody asked for is the most common and most expensive early mistake.

Write your guesses down

The Business Model Canvas, the tool at the heart of Illuminate, is a one-page sheet that splits a venture into nine blocks, including customer segments, value proposition, channels, revenue streams and costs. Treat each block as a guess to be tested, not a fact.

Try this week

Write your one-sentence problem statement. List ten people who fit it. Speak to three of them before the week ends.

In our next post, we look at what comes after testing: how to present an idea clearly to an expert panel, drawing on what we saw at Yukti '26.