How do I discover unknown competitors?

Collect the questions buyers actually ask when researching your category, run them across search engines, AI assistants and review sites, and record every vendor that appears per question. Vendors who show up consistently across several questions - but not on your internal list - are your unknown competitors.

Who else do buyers see when they search my category?

Whoever appears on the surfaces they consult: vendors named in search results and comparison articles, vendors placed beside you by review-site categories, and vendors recommended by AI assistants for specific questions. This set is usually broader than your tracked list and shifts per question.

Why do unfamiliar vendors win my deals?

Because buyers meet them during self-directed research - in AI answers, review categories and comparison content - before your sales team ever hears the evaluation exists. Lost-deal records often hide this by logging the loss against price or timing rather than the unfamiliar vendor who won.

How is competitor discovery different from competitor tracking?

Discovery finds who is actually present in buyer-facing answers, including vendors you have never heard of; tracking follows the performance of vendors you already know. Discovery feeds tracking: consistent unknowns get promoted into the tracked set and measured over time.

How often should I re-run competitor discovery?

A few times a year as a full scan, plus after any category shift such as a funding wave or a major platform launch. Known competitors can be tracked more frequently; discovery exists so that new entrants do not stay invisible until a lost deal reveals them.

How to find competitors you didn't know you had

Guide · Market Growth · 5 min read · last verified 2026-07-27

Reviewed before publication Editorial board Independent commercial review

In shortUnknown competitors are found by running real buyer questions across search, AI answers and review sites, and recording every vendor that appears per question - then tracking the consistent ones.

You find competitors you did not know you had by reading the market from the buyer's side: take the questions buyers actually ask when researching your category, run them through the surfaces buyers actually use - search engines, AI assistants, review sites - and record every vendor that appears, question by question. Your real competitor set is whoever shows up next to you in those answers, and it is almost never identical to the list your team tracks. The gap between the two lists is where deals are quietly lost.

This is a discovery exercise, distinct from tracking known rivals. Tracking answers the question: how are the vendors we watch performing? Discovery answers a prior and more uncomfortable one: who is actually in the room when our buyers evaluate options?

Why your competitor list is incomplete

Internal competitor lists are built from the inside out. They contain the vendors you hear about in deals, the ones your investors ask about, the ones founded around the same time as you, and the ones your sales team has battlecards for. That list ages quickly, and it has structural blind spots.

It misses adjacent-category vendors who added a feature that overlaps your core. It misses regional players strong in markets you have just entered. It misses open-source projects and in-house builds, which are competitors in every sense except having a sales team. And it misses newer companies that invested early in the surfaces buyers now consult - who can be prominent in AI answers while remaining invisible in analyst reports.

There is also a reporting bias: buyers rarely tell you who else they looked at, and lost deals are often logged against price or timing rather than against the unfamiliar vendor who actually won. So the internal list systematically undercounts, and it undercounts precisely the competitors you most need to learn about.

Where unknown competitors actually surface

Buyers assemble their view of a category from a handful of surfaces, and each one can introduce vendors you have never heard of.

Search results seed the early longlist - category pages, comparison articles and listicles put specific names in front of buyers before any vendor knows the evaluation exists. Review sites shape the shortlist through their own category structures; the vendors a review platform places beside you become your comparison set whether you accept the framing or not. AI assistants now matter most, because they answer the exact question buyers used to research manually - which tools should I consider for this problem - with a short, named list. Anyone recommended alongside you in that answer is a competitor for that question, full stop.

The important shift is granularity. Competition happens per question, not per category. A vendor may never appear for your core positioning yet dominate three specific questions your buyers ask on the way to a shortlist. If you only look at category level, you will never see them.

The per-question method

The method is simple to describe and tedious to run by hand, which is why it usually goes unrun.

First, assemble the questions. Draw them from sales calls, onboarding interviews, search query data and support conversations. You are looking for the twenty to fifty questions a real buyer asks between first symptom and shortlist - covering the problem, the category, the comparisons and the objections.

Second, run every question across the surfaces: at minimum, a major search engine, one or two AI assistants, and the review sites relevant to your category. For each question, record which vendors appear, in what role - recommended, mentioned, cited as a source - and alongside whom.

Third, build the presence table: questions as rows, vendors as columns, presence per cell. Then sort the vendor columns by how often each appears. This table, not your battlecard folder, is your actual competitive landscape as buyers experience it.

This is the reading Magrios automates: it researches your buyers' questions, scans the answer surfaces per question, and surfaces every vendor encountered - including the ones you were not tracking - so discovery becomes a repeatable scan rather than an annual workshop. However you produce the table, what matters is that it is built from buyer questions rather than from your existing list.

Reading the results without overreacting

Expect three findings. First, familiar rivals will be missing from questions you assumed they owned - useful, but not the point of the exercise. Second, a few names will appear once or twice; note them, but do not mobilise over a single mention, since answer surfaces vary between runs. Third - and this is the finding that matters - one or two unfamiliar vendors will appear consistently across several questions. Those are your unknown competitors, and they deserve real analysis.

For each consistent unknown, work out which questions they win, what content earns them that presence, and whether their appearances sit early in the journey - awareness questions - or dangerously late, next to comparison and pricing questions. Late-stage presence by an unknown vendor is the strongest signal, because it means buyers meet them at the moment of shortlisting.

Resist the urge to copy whatever the unknown vendor does. The correct output of discovery is a decision: which of these vendors are strategically relevant, and on which questions will we contest their presence?

From discovery to tracking

Discovery is a scan; competition is a film. Once an unknown vendor is confirmed as relevant, move them from the discovery pile into your tracked set: include them in future scans on the same questions, watch whether their presence spreads or fades, and re-run discovery on a regular cadence so the next unknown does not stay unknown for a year.

A practical rhythm is to re-run the full discovery scan a few times a year and after any category shift - a funding wave, a platform launch, a new regulation - while tracking known competitors more frequently. The discipline to keep is modest but strict: the competitor list is an output of buyer-side measurement, never an input to it.