Look in places where the work is visible
Public Product Hunt launches show what makers are shipping. Show HN discussions can expose implementation constraints and user objections. Founder websites explain positioning. Public community threads may reveal a workflow the buyer cannot finish cleanly. None of these sources alone proves a market.
Start with a role and a workflow: “accounts teams reconciling invoice exceptions” beats a broad search for “AI accounting.” Save the original URL, the date, the buyer role, and the sentence that reveals the difficulty. Link readers to the original source rather than copying a long post into your own product.
Separate the observation from your hypothesis
Use two columns in a research log. The observation is what the source actually says: “the operator still exports a CSV and emails a screenshot.” Your hypothesis is what you infer: “a shared exception queue might remove a handoff.” Keeping these apart prevents a polished idea from looking like proven demand.
Attach a confidence label to money claims. A public price page shows what a product asks buyers to pay; it is not evidence that anyone paid. A founder's revenue statement is a founder report until corroborated. Our revenue-signal guide explains the distinction.
Find repetition inside one buyer group
One frustrated comment is a search starting point. Look for several independent examples of the same job, ideally across different companies with the same buyer role. Map the trigger, the current tool, the failure point, the deadline, and the person accountable. If examples differ, you may have several markets hiding under one broad label.
Also inspect the apparently dull workaround. A spreadsheet can be adequate when the job is rare. A spreadsheet that must be reconciled by three people every Friday may signal a recurring coordination problem. The buyer's cost and urgency matter more than whether the workflow looks technologically interesting.
Turn the lead into a testable brief
Write five lines: buyer, painful job, existing workaround, proposed smallest useful outcome, and a 48-hour test. Add an India angle only when a local workflow makes the product meaningfully better; do not assume that an overseas product automatically needs an Indian copy. Before building, check Indian alternatives and talk to buyers.
The Problem Finds board uses the same separation: source-linked observation, India hypothesis, and quick test. It is a lead list for your own research, not a list of validated businesses.
