If you asked five people on your team who your competitors are, you would probably get five different lists. One person names the company the CEO mentioned in a board meeting. Another names whoever ranks first when they Google the product category. A third just lists the tools they personally tried before joining. None of those lists is wrong exactly, but none of them is complete either, and that is the problem this guide solves.
Learning how to identify competitors the right way means separating three things that get lumped together by habit: the businesses your buyers actually choose between, the sites that show up when your buyers search, and the brands that get named when your buyers ask an AI tool a question. Answering "who are my real competitors" well means keeping those three lists apart instead of merging them the first time a name shows up twice. By the end, you will have a labeled, evidence-backed list of your true competitors instead of a guess based on whoever you happen to remember.
Define the buyer and market you're testing
Before you can find who your real competitors are, you need to say exactly whose decision you're studying. A one-sentence scope does this: name the product or job to be done, the ideal buyer, the stage they're at, and the market and language you're looking at. If your product serves more than one kind of buyer with genuinely different alternatives, treat each one as its own exercise rather than blending them into a single list.
A useful test here is to ask what the buyer would do if your product didn't exist at all. That question surfaces the status quo, which is very often the thing people forget to count as a competitor. It also stops the exercise from turning into a popularity contest, since the goal isn't to rank every company in your category, it's to find the ones your specific buyer actually weighs against you.
You'll know this step is done when someone else on your team could read your scope sentence and immediately tell which searches and deal records belong in the exercise and which don't. An example: "English-language searches by founders evaluating a way to produce accurate SaaS content without adding a full-time content team." That's a scope, not a claim about how many people are searching for it.
Where people usually go wrong here is starting with a list of companies the founder admires, or mixing buyers from very different segments, countries, or use cases into one supposed competitor set. If your enterprise buyer and your self-serve buyer choose between different tools, you need two lists, not one.
Write down the alternatives your buyers actually consider
Now go looking for evidence instead of opinion. Pull whatever discovery notes, demo call notes, and lost-deal reasons you already have, and read them for what buyers actually said. If you have live conversations available, ask plain questions: "What were you using before?" "What else did you look at?" "What would you have done if you hadn't chosen us?" Write down their exact words, which segment they're in, and whether the alternative came up in a real evaluation or just as a passing mention.
Don't skip the non-vendor answers. Spreadsheets, an agency, a freelancer, or simply doing nothing are all real alternatives, and for an early-stage product they're often the most common one. If your notes are thin, it's worth a short round of calls with recent prospects rather than declaring the list finished on guesswork.
Common mistake: treating a competitor directory or a shared feature list as proof that buyers compare you to that company. A tool can look similar to yours on paper and never once come up in an actual sales conversation. Marketing consultant and positioning expert April Dunford calls these unencountered rivals phantom competitors, and they're worth a watchlist column of their own rather than a spot on your confirmed list.
If your team already keeps a working list of assumed competitors somewhere, that's a reasonable place to start collecting names, but it isn't proof on its own. Deep IQ, DeepSmith's brand context layer, keeps an editable competitor list alongside your product and persona details, which is a fine place to hold your team's working assumptions while you go verify them against the evidence above.
Give this step real time. It's tempting to skim a handful of deal notes and call the business-competitor list finished, but a short list built on thin evidence will quietly shape every content and positioning decision that comes after it. Ten minutes spent rereading actual buyer language, in their own words, catches more than a long meeting spent guessing.
Find the searches buyers make before naming search competitors
Your business competitors and your finding SEO competitors are two different lists, and this is where the second one starts. Business competitors come from what buyers choose. Search competitors come from what shows up when buyers look for answers, whether or not the site selling those answers is a business rival at all.
Start by collecting the non-branded questions your buyer would actually type: problem questions, solution questions, category questions, comparison questions, and the ones people ask right before they pick something. Keep the exact wording, since small phrasing differences can return very different results.
If your own site already gets some traffic, open Google Search Console's Performance report and look at the Queries dimension. Use "Add filter, Queries, Queries containing" to pull in the terms that match your scope, or "Queries not containing" to strip out your own brand name. A custom regex filter can group several related patterns into one view if you need it.
You're done with this step when every query in your working list has a clear, defensible connection to the buyer and job you defined in step one, and you've separated branded traffic from the rest. There's no fixed number of queries you need. Use enough to cover the different stages your content addresses, then add more if a bigger list changes the results meaningfully.
Where people go wrong: assuming Search Console shows every search anyone makes. Google hides a meaningful share of queries for privacy reasons and caps how many rows it will show you, so a thin query list from a new or low-traffic site doesn't mean there's no demand out there. It usually just means you don't have the data to see it yet from your own domain.
This step matters more than it looks like it should, because everything downstream depends on it. Finding SEO competitors without a solid query list just means you're guessing at a different point in the process instead of skipping the guesswork entirely.
Look at what actually shows up for those searches
With your query list in hand, run the searches yourself and record what actually comes back: the date, the query, the URLs, what kind of page each one is, and whether it genuinely answers what the buyer was after. Look at the page, not just the domain. A product page, a how-to article, a directory listing, and a forum thread can all show up for the same query while serving completely different parts of the buyer's journey.
Keep paid results out of this list entirely. Google labels its ads clearly, and an advertiser bidding on your category is not the same thing as an organic search competitor. Treat any single search as a sample too, since results can shift based on who's searching and from where.
You'll know you're done when you can point to the exact query and page for every site on your emerging search-competitor list, and explain whether that page actually serves the buyer's need. Label publications, directories, forums, and actual software vendors separately by role instead of forcing everything into one bucket labeled "competitor."
Pro tip: if a teammate says "we compete with that site on Google," ask them which query and which page they mean before you accept it. That sentence describes a search result, not necessarily a sales rival, and the two get confused constantly.
Use overlap reports to expand the list, not approve it
Once you've built a query-based list by hand, a ranking-overlap tool can help you find sites you missed, but it should only ever add candidates, never confirm them on its own. If your site has established rankings, Ahrefs' Organic Competitors report and Semrush's Organic Rankings Competitors report both show sites that share top-10 rankings with your domain for the keywords sending you traffic. In Semrush, compare the Competition Level score against Common Keywords rather than trusting the default sort order, since Competition Level reflects keyword overlap, not shared customers.
If your site is new and has little or no ranking history, Semrush's Competitors Discovery tool inside Position Tracking works from a custom keyword list instead of your existing rankings, which makes it a stronger starting point than asking a report to infer competitors from a domain with almost no data.
Either way, add anything new as a candidate with the specific shared queries or ranking pages attached, then mark anything that only overlaps on irrelevant terms as out of scope. There's no fixed overlap percentage or Competition Level score that automatically makes a domain a true competitor. That judgment still has to come from you.
The mistake to avoid here is sorting by total traffic or domain authority and importing the first handful of names without checking what they're actually ranking for. A site can pull huge organic traffic on topics that have nothing to do with your buyer's decision.
Neither Ahrefs nor Semrush is required to do this work. They speed up the discovery half of finding SEO competitors, but the manual searches from the step before this one will still surface most of what matters, especially for a site too new to have much of a ranking footprint to analyze in the first place.
Check AI answers as their own category
AI answers are a separate discovery surface from organic search, and they deserve their own pass rather than getting folded into your search-competitor list by default. Take the same buyer questions you built in step three and run them through the AI platforms you plan to track. For each answer, note two different things separately: which brands get named in the text, and which pages get linked as sources. Save the exact prompt, the platform, the date, and what came back.
Repeated, relevant appearances here can earn a company a spot on your AI-answer competitor list, and that list sometimes includes sites that don't sell anything close to what you sell, just because they get cited as a source. Keep this list next to your business and search lists rather than merging them automatically. A brand named in one ChatGPT answer hasn't been proven absent everywhere else, and a company cited once isn't automatically a business rival either.
This is one of the places DeepSmith earns a real mention rather than a passing one. Its AEO module tracks the exact questions you define and keeps the full answer history, and the Competitor Citations view shows who's winning citations for those tracked prompts and on which pages, broken out by platform. That's useful evidence for this exact step: which brands and pages are actually showing up for the prompts that matter to you. It's not a verdict on whether those brands are business alternatives, and coverage depends on your plan. Pro tracks ChatGPT, Grow adds Perplexity, and Scale adds Gemini, with Enterprise and Custom covering all ten engines DeepSmith monitors.
Verify, label, and freeze the list
For every name you've collected across the last four steps, check its actual offering against the buyer you defined in step one, then weigh whatever evidence you have: a buyer mentioning it, a query and page where it ranks, or an AI prompt where it gets named or cited. Sort each one into confirmed, provisional, aspirational or watchlist, and excluded, rather than forcing a single yes-or-no call. A ledger with columns for the entity, the evidence type, the role, the decision, and the date you checked it works well here, and it's fine for "provisional" to be the honest answer when you only have one weak signal.
Once names are verified, this is also where you'd add them into DeepSmith's AEO settings so tracking reflects your actual competitor set. From there, add verified competitor sites into Content Map if your team plans to use it for topic-level comparisons later. That comparison work is the next project, not part of this one, so don't let it pull you into analyzing competitors before you've finished naming them.
You're done when someone outside this exercise could read your list and understand exactly why each name made it on, in which category, and why a company that seems obvious got left off. The same company can carry more than one label at once, and that's expected, not a sign you did something wrong.
Where people go wrong at this stage is letting a famous, well-funded incumbent stay on the confirmed list just because it's famous, or dropping a smaller company despite real evidence that buyers actually consider it. A tool-generated overlap score is a candidate, not a verdict, and a single AI answer isn't a stable ranking either. If most of your candidates turn out to be generic publishers with no buyer or search relevance, that's usually a sign to revisit your scope from step one rather than force a decision on weak evidence.
What to do next
Once your list is verified and labeled, the next project is figuring out what to actually do with each entry on it: what these competitors are ranking for that you aren't, where the content gaps sit, and which of them are worth building content against directly. That's a separate exercise from this one, and it only works well when the list you feed into it is accurate. Skipping straight to competitor analysis with an unverified list of assumed rivals just means redoing this work later, after you've already spent time analyzing the wrong companies.
The whole point of going through these seven steps instead of trusting a gut-feel list is that a founder's memory is a bad instrument for this particular question. Buyer preferences shift, search results change month to month, and AI answers move even faster than either. A verified list holds up to that change because it's tied to evidence you can go back and recheck, not to whoever happened to come up in last quarter's board meeting.
If you want a faster way to see which of these verified names are actually winning AI citations and organic visibility against you, start a free trial with DeepSmith. It lets you plug your verified competitor list straight into tracking, no separate spreadsheet required.
