Loading...
Tracking competitors' AI mentions means systematically monitoring how often, where, and how favourably they're cited across ChatGPT, Claude, Gemini, Copilot, and Perplexity when users ask product-recommendation questions in your category. The winning approach combines frequency tracking, position-weighted citation analysis, sentiment scoring, and automated alerts, so you can spot gaps in competitor coverage and move faster than they can react. I'll walk through the exact framework we use at MentionOwl to help SaaS teams turn this into a repeatable competitive intelligence process.
I've spent enough time analysing AI visibility data across dozens of SaaS categories to say this with confidence: the brands winning right now aren't necessarily the ones with the best product. They're the ones that understood, earlier than their competitors, that generative AI engines have become a discovery layer with their own rules, citation logic, and blind spots. If you're not tracking what these engines say about the companies you compete against, you're making positioning decisions with half the picture.
The shift underway is bigger than most marketing teams have registered. Gartner predicted in 2024 that traditional search engine volume would drop by 25% by 2026 as users increasingly turn to AI chatbots and virtual agents for the kind of research that used to start with a Google query. That's not a marginal channel shift. It's a restructuring of how buyers move from "I have a problem" to "here are my three options."
The scale involved is already substantial. OpenAI reported in February 2025 that ChatGPT had surpassed 400 million weekly active users worldwide, and McKinsey's 2024 global survey found that 65% of organisations were regularly using generative AI in at least one business function, nearly double the share from the prior year. Meanwhile, 23% of US adults said they'd used ChatGPT as of February 2024, rising to 43% among 18–29-year-olds, precisely the demographic increasingly responsible for software procurement decisions at growing companies. Adobe's data adds a useful, if imperfect, analogue: generative-AI referral traffic to US retail sites grew by more than 1,200% between July 2023 and February 2024, and by a further 1,300% during the 2024 holiday season versus the year before. Retail isn't SaaS, but the direction of travel is unmistakable.
Here's what makes this different from traditional SEO competition, and why I think most marketing teams are underestimating the urgency. When a user searches Google for "best project management software," they see ten blue links, then twenty more on page two, then infinite scroll beyond that. Ranking eighth isn't great, but it's not invisible. You're still in the room. AI-driven recommendations don't work that way. When someone asks ChatGPT or Claude the same question, they typically get two to four named options in a single synthesised answer. There is no page two. You're either in the shortlist or you don't exist in that conversation at all.
This compression is what makes AI mentions binary rather than gradual, and it's exactly why competitor tracking here carries different stakes than competitor tracking in conventional search. A company can maintain excellent traditional SEO rankings while receiving few or no AI mentions for commercially important prompts, because these engines synthesise answers from product pages, review sites, forums, and documentation in ways that don't map cleanly onto PageRank logic. I've seen this play out repeatedly in the data: a brand ranking on page one of Google for a category term, yet completely absent when the same query is put to Gemini or Perplexity.
The strategic opportunity here is real, and it's time-limited. Most SaaS marketing teams have mature processes for tracking competitor SEO rankings, competitor ad spend, and competitor pricing changes. Almost none of them have built an equivalent process for competitor AI citations. That gap won't last. But right now, in the UK and elsewhere, teams that build this competitive intelligence capability early are operating with information their rivals simply don't have.
Once you accept that competitor AI mentions deserve monitoring, the next question is what, specifically, to measure. Raw mention counts alone are misleading, and I want to be direct about that up front, because it's the single most common mistake I see teams make when they first start looking at this data. A brand mentioned frequently but always last in a list of five, or mentioned with heavy caveats about pricing, is weaker competitively than a brand mentioned less often but always positioned first with unambiguous enthusiasm.
Here's the full set of dimensions worth tracking:
At MentionOwl, we roll these dimensions into a single proprietary AI visibility score, scaled 0–100, so that teams can track directional movement without needing to manually reconcile five separate metrics every week. But even if you're building this analysis manually, the underlying principle holds: measure the full picture, not just presence or absence.
One nuance worth flagging, because it trips people up: the major AI engines don't behave identically. ChatGPT search relies heavily on live web citations. Gemini draws on Google's broader search and product ecosystem. Claude is used disproportionately for analytical and enterprise workflows. This means testing the same query set across all of them matters. Treating any single model's output as representative of "AI search" generally will give you an incomplete, sometimes misleading, view of your actual competitive position.
The real value of this data isn't descriptive. It's diagnostic. Once you have frequency, position, and sentiment data for both your brand and your competitors across the same query set, a few specific analyses tend to surface the most actionable opportunities.
Start by reading the delta between your AI visibility score and a competitor's across identical queries. Categories where the gap is largest and consistently in their favour tell you exactly where to focus content and technical remediation efforts. But don't stop there. Also look for questions where no brand is confidently recommended at all. In our experience running this analysis across dozens of SaaS categories, these "no confident answer" queries are some of the highest-value opportunities available, because becoming the default cited answer for an underserved question is considerably easier than displacing an entrenched incumbent.
Sentiment data deserves particular attention here. If a competitor is repeatedly described with the same criticism, whether "expensive," "steep learning curve," or "limited integrations," that's not noise. It's a recurring signal the AI has picked up from review sites, forums, and comparison content it has ingested. You can directly counter-position against that criticism in your own comparison pages, pricing pages, and content briefs.
Finally, and this is a point I don't think gets enough attention in most AI SEO discussions: AI legibility gaps often explain citation lag better than content quality does. We run 16 technical checks as part of our AI legibility audits, and a recurring pattern we see is a competitor with genuinely strong traditional SEO rankings who nonetheless underperforms in AI citations, simply because their site structure makes content difficult for AI crawlers to parse and extract cleanly. If a competitor is winning on Google but losing on ChatGPT, their technical AI legibility is often the reason, and it's a gap you can close faster than you might expect.
Manual, occasional checking doesn't scale, and worse, it gives you a false sense of stability. AI answers are probabilistic and shift based on model updates, content changes, and even query phrasing. Here's the process I'd recommend building:
Data without action is just a dashboard nobody looks at. Here's how the insights above should translate into actual work:
Manual prompting doesn't scale because AI answers vary by phrasing, time of day, and model updates. A monitoring platform like MentionOwl automates this by running a consistent, auto-generated set of customer questions against ChatGPT, Claude, Gemini, Copilot, and Perplexity daily, recording competitor mentions, position, and sentiment automatically so you're comparing consistent data over time rather than one-off snapshots.
Frequency alone is misleading. The metrics that actually predict competitive strength are position-weighted citations (where in the answer they appear), share of voice (their mentions relative to the total category conversation), and sentiment (how favourably they're described). A competitor mentioned often but last in a list, or mentioned with caveats about pricing, is weaker than raw frequency suggests.
In our experience monitoring daily across multiple LLMs, meaningful shifts can occur within 3–7 days, particularly after a competitor publishes new content, updates their site structure, or a model receives an update. This is faster than traditional SEO ranking changes, which is exactly why daily tracking beats monthly or quarterly audits for this channel.
Yes. This is the core value of automated LLM monitoring. Rather than checking dashboards manually, you can set threshold-based alerts that trigger when a competitor enters a query they previously had no presence in, when their sentiment shifts, or when share of voice moves beyond a defined percentage, so your team reacts to real signal rather than combing through data daily.
Managing AI visibility monitoring for dozens of clients? I break down the API workflows, standardized metrics, and dashboard tactics UK agencies need
Learn how to define, calculate and report AI share of voice in the UK, using AI visibility scores and brand monitoring to track performance over time.
Solo founders don't have time to query ChatGPT, Claude, and Gemini every day. Here's why a weekly AI visibility digest gives you the same insight in 1