Skip to Content

Desk research or boots on the ground: which does your idea actually need?

Fieldwork multiplies the cost of validation several times over. Here is how to tell when it is indispensable and when it is theatre.

There is a point in every validation exercise where someone asks whether it is worth sending a person. It is a fair question: fieldwork multiplies the cost several times over. Sometimes it is indispensable and sometimes it is theatre. The distinction is more predictable than people assume.

Desk research is enough when the truth is already written down

Some markets are well documented. Regulated industries publish. Listed companies file. Trade bodies count. If your question is how large a category is, how fast it has grown, who the licensed players are, or what the compliance burden looks like, that information exists and someone has already paid to collect it.

Desk work also answers competitive questions well. Who is funded, who has shut down, what pricing is published, what the reviews say. For a software idea, or any business where the entire transaction happens online, you can get remarkably far without leaving a desk.

Choose desk research when: the category is established, the transaction is digital, your main uncertainty is size or competition, or you simply need a fast go or stop before committing to anything larger.

Interviews are the first real upgrade

Desk research tells you what a market looks like from above. It cannot tell you why a particular buyer chose what they chose, what they nearly bought instead, or what they would have to stop doing to adopt you.

A round of structured interviews is usually the highest-value step in the whole exercise, and the most commonly skipped, because it is the first one that requires talking to strangers. Done properly it is about past behaviour, not future intentions — what they bought, when, for how much, and what triggered it.

Choose interviews when: the buying decision is considered rather than impulsive, there is an incumbent you need to displace, or you cannot confidently name who signs the cheque.

Fieldwork earns its cost when the market is physical

There are questions that simply cannot be answered remotely, and they share a characteristic: the answer lives in a place.

  • Shelf reality. What is actually stocked, at what price, with what facings, moving at what rate. Distributors will describe this on the phone. It will not match what you see.
  • Channel economics. What margin the trade actually takes, as opposed to the published rate, and what it takes to get stocked at all.
  • Informal competition. The unbranded local operator who owns forty percent of the market and appears in no report anywhere.
  • Physical friction. Whether the loading bay fits the truck. Whether the clinic has room for the machine. Whether the road floods.

Choose fieldwork when: the product is physical, the channel is fragmented, the market is unfamiliar to you geographically, or you are about to sign a lease, hire locally, or commit to inventory.

A simple test

Ask yourself what would have to be true for the idea to fail, then ask where that evidence physically lives. If it lives in a filing, a report or a website, buy desk research. If it lives in a person, buy interviews. If it lives in a shop, a warehouse or a street, buy the flight.

Most people should do these in order rather than choosing between them. Desk research frequently kills an idea for ₹10,000, which makes the question of whether to spend ₹85,000 on fieldwork moot.

Testing this properly

We run this as a fixed-price service. A desk read with a written verdict starts at ₹10,000 and takes five working days; the full engagement puts one of our people in your target market for three days. See the four validation plans.

Share this post
Tags
MNB RESEARCh
BUSINESS GROwth
Archive
Sign in to leave a comment
What market validation actually costs in India in 2026
Feasibility studies are quoted in ranges so wide they are useless for budgeting. Here are the real bands, and the four things that move the number.