Competitive analysis: how to map a market without flattering yourself

What it is for

Competitive analysis is the structured identification of everyone solving the same customer problem — including substitutes and the status quo — followed by an honest assessment of where each is strong, in order to locate a position that is defensible rather than merely different.

Two words carry the weight. Problem, because organising the analysis by product category systematically misses the competitors that take your deals. And strong, because an analysis that catalogues weaknesses is a marketing document, and the reader knows it.

The purpose is not to prove you win. It is to find out where you can win, which requires being able to discover that you cannot.

The four types

Most analyses cover the first and stop. The other three are where deals are actually lost.

Direct competitors

Same solution shape, same buyer. The ones you already know, because you meet them in sales calls and they appear when you search your own category.

Easy to find, and the least informative part of the exercise for exactly that reason. You already knew.

Indirect competitors

Same problem, different solution shape. A consultancy doing manually what your software automates. A general-purpose tool — a spreadsheet, a project tracker, a database product — bent into the role. An in-house build by a customer's own engineering team.

These compete for the same budget and rarely appear in a category search, because they are not in your category. They are also the hardest to displace, because they are already embedded in a workflow.

Substitutes

What the customer uses instead of buying anything in your category at all. Email and a shared folder. A part-time contractor. A process that runs on somebody's institutional memory.

Substitutes look weak on a feature grid and win constantly, because they cost nothing incremental and require no decision.

The status quo

Doing nothing.

The most common competitor in most markets, and the one that is almost never listed. Every purchase requires someone to decide the current pain exceeds the cost of change — money, migration, training, risk, and the political capital of having recommended it.

If your analysis has no row for "they keep doing what they do now", it is missing the option that wins most often.

Finding the ones you would not think to search for

Four methods, in rough order of yield.

Search the problem in the customer's words. Customers describe symptoms, not categories: "spending all Friday on invoices", not "AP automation". Whatever ranks for the symptom is competing for the customer's attention at the moment of pain, whether or not it is in your category.

Read the comparison queries. Search patterns of the form "X alternative", "X vs Y" and "best X for Z" name who the market believes the incumbents are. This is frequently a different list from the founder's, and where the two disagree, the market is right — it is describing observed behaviour rather than a mental model.

Mine the negative reviews. One-star reviews say what a tool does not do. The three-star reviews are more useful: they come from people who kept using it anyway, which tells you what is load-bearing enough to tolerate the flaws for.

Read your customers' job postings. A target company's engineering and operations postings list the tools they run. That is the real stack you are displacing, stated by the customer, with no sales filter.

The grid, and why most of them are worthless

The standard artifact is a feature matrix: competitors as columns, capabilities as rows, checkmarks in the cells.

The standard failure is that the author's column wins every row. This is not persuasive; it is self-refuting. A reader cannot distinguish a genuinely strong position from a self-servingly chosen row set, so they discount both — which means a company with a real advantage loses the ability to communicate it.

Three changes make a grid informative:

Choose rows before you score them. Pick the capabilities that matter to the buyer, then fill in the cells, including the ones you lose. Choosing rows after scoring is how every-row-wins happens.

Write each competitor's strength first. One sentence on what they are genuinely good at, before anything about weaknesses. If you cannot write that sentence, you have not researched them enough to be making claims.

Include price and switching cost as rows. They decide more deals than features, and they are the two most often omitted.

Differentiation versus defensibility

Different is not defensible. The question is not whether you are distinct today but whether a well-resourced competitor could erase the distinction next quarter.

Most early-stage differentiation claims sit in the top half of that table. That is not fatal at seed stage, when speed matters more than moat. It becomes fatal when the pitch presents a top-half advantage as though it were a bottom-half one, because the reader can tell.

The honest version: "our current advantage is execution speed and a better workflow for this specific job; the durable advantage we are building is the data our customers generate by using it." That is a position with a timeline, and it survives questioning.

Reading a crowded market correctly

An empty landscape is usually a research failure or a demand failure — either you searched the wrong vocabulary, or nobody cares enough about this problem to have tried.

A crowded landscape is more ambiguous than it looks. Crowding proves demand, which is genuinely good news. Whether it is survivable depends on questions the crowd count does not answer: is anyone actually winning, or is it fragmented? Are the incumbents growing or coasting? Are customers happy, or just locked in? Fragmented markets with unhappy customers are attackable. Consolidated markets with satisfied customers are not, regardless of how good your product is.

This is also where competitive analysis interacts with everything else in an audit. A market judged saturated can override a strong score elsewhere — in INITE Studio's methodology, an unresolved contradiction between lenses blocks a GO verdict outright, precisely so that "huge growing market" and "dominated by three incumbents with 90% share" cannot both be waved through in the same document.

What a finished analysis contains

Five things, and it is short:

1. The four categories populated — direct, indirect, substitute, status quo. Five to nine entries in total.
2. One sentence of genuine strength per competitor, written before the weaknesses.
3. A grid whose rows were chosen by buyer relevance, including rows you lose.
4. A named gap — a specific buyer, with a specific job, currently served badly by everyone listed.
5. A defensibility statement with a timeline: what your advantage is now, what it becomes, and what would have to be true for it to compound.

The fourth is the output. Everything before it is the evidence, and an analysis that stops at the grid has gathered evidence for a conclusion nobody wrote down.