How to do market research

Market research is not about producing a document. It exists to answer one question: does anyone want to buy this, and at what price? Anything that does not answer that question can be cut.

Here is the method in six steps, in the order they are actually done. Each one produces a number, and each number can stop you before you go further: that is the point.

1. State the question as a product, not an idea

"I want to get into sports" cannot be measured. "750 ml insulated water bottle" can. The first task is to come down from the idea to a product a customer would actually type into a search box.

The test is simple: if you cannot picture someone searching for those exact words, it is still too broad. Go down one more level.

2. Measure real demand

Demand is measured in monthly search volume: how many people look for this product, each month, in the country you target. It is the single most important figure in the whole study, and the one most often skipped.

There is no universal threshold: it depends on your unit margin and on what a customer costs you to acquire. What matters is setting one and sticking to it, otherwise every idea eventually looks acceptable. It is the first of the five criteria we cover in finding a profitable niche.

For a sense of scale, here are the ones MarketRaccoon applies in its own verdict: below 100 monthly searches a market is flagged red whatever the other data says; below 300 it can never go beyond amber. These are not laws of commerce, they are our guardrails, but they keep you from starting with no reference point at all.

Look at the twelve-month trend too, not only today's volume. A large market that is contracting and a small one that is growing are not played the same way.

3. Collect the prices actually charged

Price is not decided, it is observed. Record what the existing players actually sell for, and note three figures: lowest, median, highest.

Use the median, not the average. This is not statistical pedantry: in a market of thirty products around €20, a single €400 item is enough to move the average by several euros, while the median does not budge. The average describes a product that does not exist.

More interesting still are the gaps in the range. If products cluster at the low end and a middle band is empty, that band is either an opening or a graveyard. To find out which, look for someone who occupied it and left.

4. Look at who is already there

Counting competitors is useless. What matters is the structure of the market: the share held by the largest seller in the visible offer.

The higher that share, the more one player sets the reference price and the more you become their follower. The lower it is, the more open the market, but then ask the opposite question: if entering is that easy, what will stop the next entrant after you?

As a reference point, MarketRaccoon calls a market "concentrated" when the top seller exceeds 30% of the collected offer, and "fragmented" below that. Again: our display convention, not an economic truth.

One point of method matters more than the threshold: do not count as a competitor a seller with a single listing. They appear and vanish between two collections. We only track one once at least three of their products are found: below that the measurement is too unstable to say anything.

This step deserves more than a few lines. We gave it a full guide, analysing the competition, covering the three questions to ask and the traps of price collection.

5. Read the reviews of existing products

This is the most profitable step and the most neglected. Customer reviews of a competing product tell you, for free, what the market holds against the current offer.

Do not read the 5-star reviews, they teach you nothing. Do not stop at the 1-star ones either: they are often about delivery problems that have nothing to do with the product. The 2 and 3-star reviews are the useful ones: customers who bought, kept the item, and explain precisely what was missing.

Read until the complaints start repeating. That happens sooner than you would expect, and every recurring complaint is a selling point the market wrote for you.

6. Decide, and write the decision down

By now you have five things: volume, trend, median price, market structure and the list of recurring complaints. The decision is made here, not after three more weeks of thinking.

Write it in one sentence, with the figure that justifies it: "I am launching at €32 because the €28-45 band is empty and the top complaint is durability." A decision without its figure is instinct in disguise, and it will be defended just as fiercely six months later, when sales do not follow.

The mistakes that come up most

• Mistaking your enthusiasm for demand. You are not your customer, and the people close to you are not a sample: they are responding to your excitement, not to your product.
• Looking at average prices. Always the median, and always the full distribution.
• Stopping at volume. High volume in a contracting market costs more than a small market that is growing.
• Studying only one idea. After days spent on a single idea, you always find good reasons for it: the effort already spent weighs on the judgement. Comparing five ideas makes the decision obvious where studying one made it arguable.

How long it takes

Done by hand, chasing each data point separately, this is several days of work for a single product. That is the main reason the step gets skipped: not that people think it useless, but that it costs more than the time they have.

All six steps above rely on public data: search volumes, listed prices, published reviews. They can therefore be collected automatically, which changes the quality of a study less than it changes the number of ideas you can afford to examine before picking one.