Competitive analysis: five steps to findings you can actually base a decision on

What is a competitive analysis for?

A competitive analysis exists to make a decision better. That sounds obvious, but it's the distinguishing feature: if the analysis produces the same decision you would have made anyway, it was busywork.

That's why the question comes first, not the data collection. "Should we enter this segment?" leads to a different analysis than "Can we raise our price by seven percent?" or "Why do we lose bids to this one specific provider?" Each of these questions needs only a little data, and each needs different data.

The typical competitive analysis in mid-sized companies hasn't recognized this problem. It takes shape as a slide deck with one page per competitor, neatly researched, and gets flipped through in the leadership team. Afterward someone says "interesting," and the discussion picks up right where it stood before the analysis. The reason is a missing question, not poor work.

Which five steps lead to solid conclusions?

The sequence carries more weight than the method. Each step reduces the material instead of expanding it. That's the real difference from a collection.

Two to four weeks is realistic for an analysis like this if someone has half a day per week for it. What takes longer is almost always step two: the circle gets drawn too wide, and the data collection grows with every additional name.

Where does the data come from when competitors publish nothing?

Among mid-sized companies in Germany, that's the norm. Most relevant competitors are private companies with no press work, no investor presentation, and a website that hasn't changed in years. Even so, the most useful data lies closer than it looks.

• Your own lost bids. The strongest source there is, and the one least often analyzed. Sorting the last twenty lost bids by who won them and why gives you a more precise picture than any market study.
• Your own salespeople. At every meeting they hear what competitors offer and what they've promised. That knowledge exists, but it's recorded nowhere. It takes half an hour per person and a structure.
• Annual financial statements in the German Federal Gazette. Disclosure requirements mean that revenue magnitude, balance sheet total, headcount, and often the earnings situation are public. Two years side by side show the direction, and that's usually more telling than the absolute number.
• Job postings and certifications. A company hiring three people for a new technology is building something. A company obtaining an approval wants into a regulated market. Signals like these come months before the product.

What's off limits is just as clear: no posing under a false name, no fake inquiries, no conversations with competitors' employees about internal matters. That's risky and practically useless. Information from sources like these can't be named as a basis in a leadership meeting, and what can't be named won't carry a decision.

Which points can actually be compared in a meaningful way?

What's comparable is what means the same thing for everyone involved. That's less than it seems: revenue per employee says nothing about performance when vertical integration differs, and market shares among mid-sized companies are mostly estimates, a gut feeling expressed as a percentage.

Four levels hold up. What the competitor offers and what it doesn't. Which customer it's built for: almost everything else hangs on that. What price range it operates in, in ranges and without decimal places. And how it sells, because the sales channel reveals the cost structure.

One level above that, it helps to ask about the forces in the industry: how easily a new provider can enter, how much leverage customers and suppliers have in price negotiations, what fallback option the customer has. This view goes back to Michael Porter and is useful where the issue is less about a single competitor than about whether there's money to be made in this industry at all.

And one level is almost always forgotten: the fallback option. In many mid-market industries, the most common "competitor" isn't a provider at all. It's the customer's decision to keep doing it in-house or to do nothing. If you only compare providers, your biggest opponent isn't in the table.

What do you do with the result?

Three to five conclusions, each with a consequence. No leadership team can keep more than that straight, and less doesn't justify the effort.

It helps to sort each conclusion into one of three buckets. It confirms what you assumed, in which case it costs nothing but buys confidence. It contradicts an assumption underlying a decision already in motion, in which case it's the most valuable thing in the entire project. Or it's new and unrelated to the current question, in which case it belongs in the file, not in the discussion.

The conclusions that contradict are the real payoff. They're also the ones that get talked away fastest in leadership meetings, because they're uncomfortable. That's why every conclusion belongs on the same line as its source: a statement backed by "three of five lost bids, July analysis" holds up longer than one backed by nothing but an impression.

Which mistake makes any analysis worthless?

The attempt to be complete. A matrix with twenty providers and thirty attributes contains a justification for every position, and therefore justifies nothing. It also creates a false precision that no one questions, because the table looks so thorough.

The second mistake is the mirror comparison: you compare on the attributes where you're strong. The result is predictable, and it doesn't explain why customers buy elsewhere anyway. The better cut is the reverse: compare on the attributes the customer actually checks when buying, even if you come off worse there.

The third is timing: a competitive analysis that lands after the decision is a justification. If the decision comes in six weeks, a lean analysis in four weeks is worth more than a thorough one in four months.