Running a market analysis: which data actually supports a decision

Which four figures does a decision require?

A market analysis becomes useful as soon as it answers these four questions. Everything else is decoration.

• How large is the addressable market? Not the industry, but the part you can actually reach with your offering, your sales channel and your region. This number is regularly a fraction of the number in the study.
• Which direction is it heading? Over three years, not one. A single year shows the business cycle, three years show a trend.
• What share can you win? Limited by capacity, sales reach and customers' willingness to switch. A number without these limits is a wish.
• Who is making money today? If nobody earns a margin in a growing market, growth is no argument for entering it.

The fourth question is the one most often skipped, and it overturns most entry decisions. A market can grow and still be unattractive for everyone involved, for example when customers negotiate hard or an upstream input dictates the price.

Before the four questions comes a decision that determines their answers: defining the market. "The market for industrial valves" is not a definition. "Shut-off valves from DN 200 up for waterworks in Germany and Austria" is. The definition needs three elements: product type, customer type and region. Without them, every source delivers numbers for a different field. The analysis then ends with values that differ by a factor of ten.

Why is total market volume the least important number?

Because no company serves an entire market. The number impresses in presentations and answers none of the questions on the table.

Here is the math in an example: according to association figures, the European market for a component group is four billion euros. You manufacture in a tolerance class that covers roughly one fifth of that volume, you deliver to three countries and you serve buyers above a certain lot size. That leaves roughly 180 million addressable. Your capacity covers 25 million. The number a decision hinges on is the last one, not the first.

The big number is only useful as a cross-check. If your bottom-up projection and your share of the association figure are far apart, one of the two contains a flaw in reasoning. That is where the real value of a study lies: in the comparison with your own calculation, not in its number.

Where does the data come from in mid-sized companies?

From five sources. The most expensive one is rarely the best.

The order in the table is also a recommendation for how to proceed. If you start with your own quote statistics and fifteen customer conversations, you will know more about your market after two weeks than you would from reading a study.

Two sources are deliberately missing from this list. We do not talk to competitors' employees, because nothing comes out of it that you could cite as a basis in a leadership meeting. And figures from consulting proposals are sales material, not findings. Both are questionable and, on top of that, practically useless.

How many customer conversations are enough?

Twelve to twenty, provided they fall within the same segment and follow the same structure. After that the answers start repeating, and that is exactly your signal to stop.

The selection matters more than the number. Three groups belong in it: customers who buy from you, customers who buy elsewhere, and companies that solve the problem themselves. The second and third groups deliver the insights. The third is the hardest to reach and the most valuable.

The questions decide whether it's worth anything. "Would you buy this?" gets you politeness. "What did you buy last time, how was the decision made, and who was involved?" gets you behavior. Past behavior is the only reliable information a conversation can give you.

How do you build the potential from the bottom up?

Through the number of possible customers, their annual demand, and a share you can justify. This calculation is fallible and still better than any top-down estimate, because every assumption in it is named and therefore testable.

A scenario: according to a company database, your target region has around 900 companies with a matching profile. Based on customer conversations, you estimate that about a third have the need on a regular basis. Average annual revenue per customer is €40,000. That puts roughly €12 million in the addressable market. With one field sales rep and twelve to eighteen months of lead time, 5% of that is a plausible target for year two.

Write down every assumption next to where it came from. If the number doesn't materialize later, you can say which assumption was wrong. Without that link, a missed number leaves nothing but the impression that market analysis is reading tea leaves.

When is the analysis done?

When the decision it was made for can be made with it. That's the only useful standard. It makes a lot of analyses much shorter.

In practice: two to four weeks for a market entry decision, at half a day of work per week. It only takes longer when nobody has written down the decision at stake, so people collect whatever they can find.

What's left afterward is a three to four page document: the four data points above, the assumptions with their sources, the implication, and the question of what development would make you reassess the decision. That last point turns an analysis into a tool that still has value a year later.

Contradictory numbers are normal and no reason to keep digging. If the industry association's statistics show a growing field and your own quote statistics show falling prices, both are usually right: the market is growing in a segment where you don't compete. Contradictions like these belong in the document instead of being resolved. They're often the most useful statement in it, because they show a shift before it shows up in your numbers.