Quick answer
When a competitor gets named and you don't, the cause is usually one of four distinct problems: a sources problem (they're cited, you're not), a clarity problem (your page doesn't clearly state the fact being asked about), a framing problem (you're named but described poorly), or a genuine gap (they serve the use case better). Diagnose which one before deciding what to fix.
The instinctive reaction to seeing a competitor recommended instead of you is to assume you need more content, more SEO, more reviews, something. That instinct is usually premature. Before deciding what to fix, it's worth spending fifteen minutes diagnosing why a specific AI answer, for a specific prompt, named them and not you, because the four most common underlying causes call for completely different responses, and guessing wrong wastes real effort on the wrong fix.
The four-way diagnostic
Almost every case of "they got recommended and we didn't" for a prompt you'd reasonably expect to be named on falls into one of four buckets. The order below is roughly the order to check them in, cheapest and fastest checks first.
1. A sources problem
The AI answer names a source, a comparison article, a review site, documentation, that covers your competitor and doesn't cover you, or covers you only in passing. This is the easiest to diagnose and, often, the easiest to act on: if the platform shows citations, look at exactly what was cited. If it doesn't show citations directly, ask a follow-up question in the same conversation, most assistants will tell you what informed an answer if you ask directly, even without a formal citation feature.
2. A clarity problem
You have a page that could plausibly answer the question, but it doesn't state the specific fact the prompt is asking about clearly enough to be extracted and cited with confidence. This is common with pricing, feature-availability and compatibility questions in particular: the information might technically be present on your site, but buried in a PDF, phrased vaguely, or split across two pages in a way that makes it hard for anything, human or AI, to extract a confident, specific answer from it.
3. A framing problem
You are named, but described in a way that doesn't help, or actively hurts, your case for the specific question asked. This is subtler than the first two and needs a careful read of the actual answer text, not just a check of whether your name appears in it. Being named third with the word "basic" attached is a framing problem even though, technically, you weren't omitted.
4. A genuine gap
The honest, less comfortable possibility: the competitor actually does serve this specific use case better, and the AI answer is reflecting a real difference, not a visibility or content problem at all. This is worth taking seriously as a real option, not a worst case to be avoided at all costs, because treating every omission as a content problem when the real issue is a product gap wastes effort and delays the conversation that would actually help.
Running the diagnostic, step by step
Start with the specific prompt and the specific answer that surfaced the problem, not a vague sense that "we're not showing up enough." A diagnosis needs a concrete thing to diagnose.
- 1Re-run the exact same prompt, several times, to confirm this is a pattern and not a one-off answer. If it only happened once, it may not be worth diagnosing further yet.
- 2Check what sources the answer cites or references, directly if shown, or by asking a follow-up question if not. This tells you whether you're dealing primarily with a sources problem.
- 3If sources look reasonably balanced between you and the competitor, check whether your own site has a page that directly, specifically answers the exact question asked, not a page that's adjacent to the topic. This tests for a clarity problem.
- 4If a page exists and is clear, read exactly how you're described when you are named, word by word, not just whether you appear. This tests for a framing problem.
- 5If none of the above explains it, consider honestly whether the competitor has a real capability, price point, or fit advantage for this specific use case that the answer is correctly reflecting.
A worked, illustrative example
Say a mid-sized project management tool, call it, purely as an illustration, "Illustrative Co.", notices it's rarely recommended for the prompt "best project management tool for a construction company." Running the diagnostic: the prompt is re-run five times across two platforms, and the pattern holds, this isn't a one-off. Checking sources shows the answers consistently cite two industry-specific comparison articles, neither of which mentions Illustrative Co. at all, only the two competitors most associated with the construction niche. Illustrative Co.'s own site does have a features page, but nothing specifically positioned for construction, no case study, no "for construction teams" landing page, nothing a comparison article writer would have found if they'd gone looking. That's a sources-and-clarity combination: Illustrative Co. isn't in the sources being drawn from, and even if a writer went looking, there's nothing clearly construction-specific to point to.
The fix that falls out of that diagnosis is specific: build a clear, construction-specific page (addressing the clarity gap), and pursue inclusion in the industry comparison content that's actually being cited (addressing the sources gap), rather than, say, generically "improving SEO" or assuming the product itself is unsuited to the use case, which the diagnostic gave no evidence for either way.
This example is illustrative
What not to do: jumping straight to "we need more content"
The most common mistake is skipping the diagnosis and defaulting straight to "publish more content about this topic," which is sometimes right, for a clarity problem, but is the wrong response to a sources problem, where the content might already exist but isn't being found or trusted by whatever's feeding the answer, and it's actively unhelpful for a genuine gap, where more content can't substitute for a real product or service difference, and chasing it just delays recognising the actual issue.
A second common mistake is treating a framing problem as if it were an omission. If you're named but described poorly, adding more content elsewhere often does nothing, because the issue isn't that there's too little information about you, it's that the specific information being surfaced casts you in an unhelpful light. That calls for improving or clarifying the specific pages and third-party content that are actually shaping the framing, not adding unrelated new material.
When the honest answer is a genuine gap
This is the hardest outcome to sit with, but it's the most valuable one to recognise correctly, because acting on a false diagnosis wastes real effort. If a careful check of sources, clarity and framing all come back clean, and a competitor is still being consistently, correctly recommended over you for a specific, well-defined use case, the most useful next step usually isn't content work at all. It's taking the finding to whoever owns product or strategy, framed exactly as what it is: evidence that, for this specific segment or use case, buyers and the sources that inform AI answers currently see a real difference. What they do with that is a strategic decision, not a visibility one, and pretending otherwise just delays a conversation that would actually help.
How a product team should receive this kind of finding
A diagnosis that reaches product or strategy leadership framed as "the AI thinks competitor X is better than us" tends to get dismissed quickly, and understandably so, since stated that way it sounds like an impressionistic complaint rather than an evidenced finding. The version worth bringing to that conversation looks different: the specific prompt, run a specific number of times, on specific platforms, with the specific diagnosis reached (sources, clarity, framing, or genuine gap) and the specific evidence behind that diagnosis, what sources were cited, what the competitor's page says that yours doesn't, or what capability difference was actually confirmed. Framed this way, it reads as exactly what it is: a small, bounded, evidenced observation about one use case, not a sweeping claim about overall competitive standing, and it's correspondingly easier for a product leader to engage with seriously rather than wave off.
It also helps to bring a specific, bounded ask rather than an open-ended one. "Here's what we found; what would it take to close this for this specific segment, and is it worth doing" invites a real decision. "We need to catch up to competitor X" invites exactly the kind of vague, hard-to-act-on response that leaves the finding sitting in a deck with no follow-through.
Turning this into a repeatable practice
A one-time diagnostic is useful for a single prompt that caught your attention. The real value comes from applying the same four-way check consistently, across your most important category prompts, on a schedule, so you can tell whether a gap is improving, stable, or getting worse after you act on it, rather than re-diagnosing from scratch every time someone on your team notices a bad answer and raises it.
The role of timing in this diagnostic
It's worth checking when a competitor's advantage in a specific answer first appeared, if you have any historical record to check against, since that timing is itself diagnostic information. An advantage that's been stable for a long time points toward a durable, structural cause, a genuine capability difference, an established set of third-party sources, that's unlikely to shift quickly regardless of what you do. An advantage that appeared suddenly, within the last check or two, points toward something more recent and specific, a new comparison article getting published, a competitor's recent launch, a sudden change in how a particular prompt is being interpreted, which is often easier to trace to a specific, identifiable cause and therefore easier to respond to with a targeted fix rather than a broad, uncertain one.
A quick gut-check before escalating a finding
Before taking any diagnosis up to product or leadership, it's worth running one final gut-check: would a reasonable, informed person looking at the same evidence reach the same conclusion, or does the diagnosis depend on reading the evidence in the most unfavourable possible light. It's easy, especially after spending real effort running a diagnostic, to want the finding to justify that effort by being more dramatic or more actionable than the underlying evidence actually supports. Asking a colleague who wasn't involved in running the checks to look at the raw answers and reach their own independent read, before the framing is locked in, is a cheap, effective safeguard against that bias, and it tends to produce findings that hold up better under scrutiny once they reach a wider audience.
Documenting the diagnosis for the next person who asks
The same "why aren't we recommended for this" question has a tendency to resurface every few months, often raised by a different person each time, a new hire, a different stakeholder who just noticed it independently, someone preparing a board update, and without a written record of a prior diagnosis, each occurrence triggers a fresh investigation from scratch. Writing down not just the conclusion but the actual evidence checked, which sources were cited, what the competitor's page said, what your diagnostic log from the earlier worked example recorded, turns a one-time investigation into a reusable answer. The next time the question comes up, whoever's asking can be pointed to that record, updated with a quick re-check if enough time has passed, rather than the whole diagnostic process starting over.
A closing caution against over-diagnosing
There's a failure mode at the opposite end of the spectrum from skipping the diagnosis altogether: running this framework so exhaustively, on so many prompts, that the team spends more time diagnosing than actually fixing anything. The framework is a tool for deciding what to do about a gap that matters, not an exercise to run comprehensively against every possible phrasing of every possible question your category might generate. Reserve it for the prompts that genuinely matter to the business, the ones a real, qualified prospect is actually likely to ask, and resist the pull to diagnose every minor variation just because the method is available and relatively cheap to apply.
A reasonable discipline here is to cap how many prompts get a full diagnostic treatment in any given review cycle, prioritizing by how directly each one connects to an actual buying decision, and accepting that some lower-priority gaps will simply go unexamined for now rather than trying to diagnose everything at once. That's a better use of a team's limited time than spreading the same effort thin across a long list of prompts, most of which matter far less to the business than the handful at the top.
| Diagnosis | What you'll see | Typical fix |
|---|---|---|
| Sources problem | Cited sources don't mention you at all | Earn inclusion in the specific sources being cited |
| Clarity problem | No page clearly, specifically answers the exact question | Create or sharpen a page that directly answers it |
| Framing problem | You're named, but described unhelpfully | Improve the specific content shaping that framing |
| Genuine gap | Sources, clarity and framing all check out fine | Raise it as a product or strategy question, not a content one |
Why confirming the pattern matters more than it seems
It's worth dwelling a little longer on the first step of this diagnostic, confirming the pattern across repeated runs, because skipping it is the single most common way a team ends up diagnosing, and then fixing, a problem that was never really there. AI assistant answers are sampled, not looked up, which means the same prompt can genuinely return a different set of named brands from one run to the next even with no change in the underlying sources or facts. A single bad answer, seen once, screenshotted and forwarded around a Slack channel, can trigger a flurry of concern and effort that a second or third run would have shown was simply an unusual draw, not a real, consistent gap.
The discipline here costs very little, running the same prompt three to five times takes a few minutes, and it buys real protection against the much larger cost of a team spending weeks on a content or outreach effort aimed at fixing a pattern that, on closer inspection, wasn't actually consistent. It's worth treating this confirmation step as non-negotiable before any of the later diagnostic steps, not as an optional nicety to skip when time is short, because skipping it is exactly when the diagnosis is most likely to go wrong.
A second worked example: a services business
Product comparisons aren't the only place this diagnostic matters. Consider, again purely as an illustration, a regional accounting firm, "Illustrative Partners," that notices it's never recommended for the prompt "accounting firm for a small business switching from a bookkeeper." Running the four-way check: the pattern holds across repeated runs on two platforms. Checking sources shows the answers tend to cite general listicle-style articles about "how to choose an accountant," which don't name specific local firms at all, they stay generic. That rules out a sources problem in the usual sense, there's no competitor-specific source beating them out, because the sources being drawn from aren't naming specific firms either way.
Digging further, it turns out Illustrative Partners has no page at all that addresses this specific transition scenario, moving from a bookkeeper to a full accounting relationship, even though it's a common client journey they handle constantly in practice. That's a clarity problem wearing a sources-problem disguise: the generic sources aren't naming anyone because no one, including Illustrative Partners, has published anything specific enough about this exact scenario for a generic source to point to. The fix here isn't earning inclusion in existing comparison content, there isn't any specific enough to get into, it's being the first to publish a clear, specific page addressing exactly that transition, which both serves real prospective clients searching for it and gives something concrete for a future AI answer, or a future comparison article, to eventually cite.
This example is illustrative
Building a diagnostic log your team can actually reuse
A diagnosis that lives only in one person's head, or in a one-off Slack message, doesn't compound. The teams that get real value from this framework over time keep a simple running log: the exact prompt, the date and platforms checked, the number of runs, which of the four categories it was diagnosed as, what action was taken as a result, and, critically, a follow-up date to re-check whether that action actually changed the answer. This doesn't need to be elaborate, a shared spreadsheet with one row per diagnosed prompt is enough to turn a series of individually useful diagnoses into an institutional record of what's been tried, what worked, and what's still an open gap.
That log earns its keep most clearly when someone new joins the team, or when the same "why aren't we recommended for X" question comes up again six months later. Without a log, that question gets re-diagnosed from scratch, possibly reaching a different, inconsistent conclusion the second time. With one, the answer, and the reasoning behind it, is already there.
What this has to do with sales objection handling
There's a useful, underused connection between this diagnostic and the objection-handling material your sales team probably already has. If an AI assistant is consistently framing a competitor more favourably for a specific use case, that framing is very likely to turn up, in a prospect's own words, as an objection during a real sales conversation: "I asked around and it sounds like competitor X is better for construction companies specifically." A sales team equipped with the actual diagnosis, is this a sources gap, a clarity gap, a framing issue, or a genuine difference, can respond with something far more credible than a generic rebuttal, because they understand precisely why that perception exists and, where it's fixable, that work is already underway.
Why the fix usually isn't just 'be better'
There's a tempting, unhelpful shortcut available at the end of any diagnosis that lands on a genuine capability gap: concluding that the business simply needs to "be better" at whatever the gap concerns. That conclusion is true in the most trivial sense and useless in every practical one, it gives nobody a next action. The more useful version of the finding names the specific, narrow thing that's actually different, a missing feature, a slower support response time, a narrower set of integrations, for the specific use case or segment where the gap showed up, since a precisely named gap can go to product or operations as a concrete, prioritizable input, while "be better overall" cannot.
It's also worth resisting the urge to treat a genuine gap in one narrow use case as evidence of a broader, company-wide weakness. A diagnosis that an AI assistant correctly favours a competitor for, say, large enterprise deployments, does not imply the same competitor is also correctly favoured for small-team use cases, or for a different geography, or for a different pricing tier. Keeping the finding scoped exactly to what the diagnostic actually showed, rather than generalizing it further than the evidence supports, keeps the eventual response proportionate and credible rather than triggering an overreaction to what might be a narrow, specific, and entirely normal competitive difference in one corner of the market.
When more than one cause is present at once
Real situations are often messier than the clean four-way split suggests, and it's worth expecting some overlap rather than forcing every case into exactly one bucket. A brand might have a genuine, if modest, capability gap for a specific use case, and also have a clarity problem that makes the gap look larger than it actually is, because the strengths the brand does have for that use case aren't clearly documented anywhere. In that situation, closing the clarity gap won't fully fix the perception, but it will likely narrow it, and it's honest, useful work either way. The diagnostic framework isn't meant to force a single clean verdict, it's meant to stop you from reflexively reaching for the same fix (usually "more content") regardless of what's actually going on, by making you check each possibility deliberately before deciding what to do.
Key takeaways
- Diagnose before you fix: the four causes call for different responses.
- Confirm the pattern across several runs before investing effort in a fix.
- A sources problem and a clarity problem look similar from the outside but need different work.
- Being named with weak framing is a distinct problem from not being named at all.
- A genuine gap is a valid, important finding, not a diagnosis to avoid reaching.
- Keep a running log of diagnoses and outcomes so the work compounds instead of restarting each time.
Frequently asked questions
How do I know if the sources behind an answer are the real cause, versus just a coincidence?
Check across several repeated runs of the same prompt. If the same sources keep showing up and consistently exclude you, that's a real pattern worth acting on. If the sources cited vary a lot between runs, the sources problem is less clear-cut and another cause may be more likely.
What if the platform doesn't show which sources it used?
Try asking a direct follow-up question in the same conversation, such as "what did you base that on" or "what sources support this." Many assistants will explain their reasoning or name sources even without a formal citation feature, though the answer isn't always complete or precise.
Is a framing problem ever worth ignoring?
If the framing is mild and the overall answer is still reasonably favourable, it may not be worth dedicated effort. It's worth tracking over repeated runs to see whether it's a consistent pattern before deciding it's a priority.
How do I tell a genuine gap from a framing problem that just feels like a gap?
A framing problem means the underlying facts support you but the language doesn't; a genuine gap means the underlying facts, honestly assessed, favour the competitor for this specific use case. The test is whether you'd still choose the competitor yourself, given full and honest information, for that exact use case.
Should I run this diagnostic for every prompt where a competitor appears?
Prioritise prompts that matter most to your business, high-intent category questions your actual customers ask, rather than every possible phrasing. Depth on the prompts that matter beats shallow coverage of everything.
Can this diagnostic be automated instead of done by hand each time?
The diagnostic reasoning itself benefits from a human judgment call, but having the underlying data, which sources get cited, how you're described, tracked automatically across repeated runs makes each diagnosis faster and more reliable than starting from scratch by hand every time.
How much evidence is enough before escalating a genuine-gap finding to product leadership?
Enough that a skeptical colleague reviewing the same raw answers independently would reach a similar conclusion. That usually means several repeated runs, a clear read of the cited sources, and a specific, narrow description of the use case involved, rather than a single striking answer.
Cite this page
BrandWater AI Research. (2026, 2 November 2026). When AI recommends your competitor instead of you: a diagnostic framework. https://brandwaterai.in/blog/when-ai-recommends-your-competitor-instead
How we work
Figures are dated and linked to their sources. Where none exist we say so. Read our methodology and AI transparency pages.