Competitive Analysis Has Changed. AI Search Data Is Why

Competitive analysis using generative AI search data means using the outputs of AI-powered search tools, including ChatGPT, Perplexity, Google’s AI Overviews, and similar platforms, to understand how competitors are being positioned, recommended, and described in the responses users actually see. It is a different discipline from traditional SEO competitive analysis, and it requires a different approach.

The shift matters because AI search does not rank ten blue links and let users decide. It synthesises a response, names specific brands, and frames the competitive landscape before the user has clicked a single result. If your competitors are being cited and you are not, you are losing ground at the point of consideration, not just the point of click.

I have spent the better part of two decades doing competitive analysis across thirty-odd industries, from travel and retail to financial services and B2B SaaS. The mechanics change, but the underlying question never does: where are we losing ground, and why? AI search has added a new layer to that question, and it is one that most marketing teams are not yet asking systematically.

Why AI Search Creates a New Competitive Blind Spot

Traditional competitive analysis in search has always been about visibility: who ranks for what, at what position, with what share of estimated traffic. Tools like Semrush and Ahrefs made this tractable. You could pull a competitor’s keyword profile, identify gaps, and build a content plan around them. That model still has value.

But AI search does not work like that. When someone asks ChatGPT which project management tools are worth considering for a mid-sized agency, they get a synthesised recommendation, not a ranked list of URLs. The brands that appear in that response have not necessarily ranked number one for “project management software.” They have been encoded into the model’s training data and retrieval patterns as credible answers to that question. That is a fundamentally different competitive dynamic.

The Moz team has done useful work on using LLMs for competitive research and gap analysis, and it surfaces something important: the brands that dominate AI responses tend to have strong editorial footprints, consistent brand descriptions across sources, and clear positioning in their category. They are not winning because of technical SEO. They are winning because they have become the default answer to a category question.

If you are not tracking where your competitors appear in AI-generated responses, you have a blind spot in your competitive intelligence. And in categories where AI search adoption is growing, that blind spot gets more expensive every quarter.

How to Build a Prompt Testing Framework

The starting point for AI competitive analysis is prompt testing, and it is more structured than it sounds. The goal is to simulate how real buyers use AI search tools and document which brands appear, in what context, and with what framing.

Start by mapping your category from the buyer’s perspective. What questions does someone ask when they are in-market for what you sell? These are not keyword queries. They are conversational prompts: “What are the best options for X?”, “How does Y compare to Z?”, “What should I look for when choosing a provider of X?” Build a prompt library of twenty to forty questions across the consideration and evaluation stages.

Then run those prompts across at least three platforms: ChatGPT, Perplexity, and Google AI Overviews. The responses will differ, sometimes significantly, and that variance is itself useful data. A competitor that appears consistently across all three has a stronger AI presence than one that appears only in ChatGPT. A brand that is mentioned but immediately qualified with caveats is in a different position from one that is cited as the default recommendation.

Document everything in a structured format. For each prompt, record which brands are mentioned, in what order, with what descriptors, and whether any sources are cited. Do this monthly. The changes over time are often more revealing than any single snapshot.

This is not a five-minute exercise. When I ran a version of this for a B2B client last year, the prompt library alone took half a day to build properly. But the output was a clear picture of which two competitors had effectively colonised the AI-generated narrative in their category, and why. That informed a content and PR strategy that would not have emerged from a standard keyword gap analysis.

If you want to go deeper on the technical side of how AI tools are being used for SEO-adjacent work, the Ahrefs AI tools webinar series covers practical applications that complement this kind of prompt testing approach.

What to Look for in Competitor AI Presence

Once you have your prompt testing data, the analysis falls into four areas.

Citation frequency. How often does each competitor appear across your prompt set? A brand that appears in thirty of your forty prompts has a materially different AI presence from one that appears in eight. Frequency is the baseline metric.

Positioning language. How is each competitor described when they appear? Are they the “industry standard,” the “affordable option,” the “enterprise choice,” or the “best for small teams”? This language is not arbitrary. It reflects how the model has synthesised the available information about that brand. If a competitor is consistently described as the reliable, established choice and you are described as an alternative, that is a positioning problem, not a content volume problem.

Source attribution. Perplexity in particular cites its sources. When a competitor is mentioned, what sources are being cited? If they are consistently backed by coverage in respected trade publications, analyst reports, or major editorial outlets, you are looking at a PR and earned media gap as much as a content gap. This is where the Moz perspective on AI content and E-E-A-T signals becomes relevant: the authority signals that influence AI responses are largely the same signals Google has been building toward for years.

Category ownership. Some brands appear whenever a category is mentioned. Others only appear for specific sub-queries. Mapping this tells you which competitors have achieved category-level authority and which are competing at the feature or use-case level. Category-level authority is harder to displace, but it is also more visible and therefore more actionable to target.

Connecting AI Presence to Underlying Content Signals

The question that follows the prompt testing analysis is always: why are they appearing and we are not? The answer is rarely a single factor. It is usually a combination of content depth, editorial coverage, and entity clarity.

Content depth means having substantive, well-structured content that directly addresses the questions buyers ask. Not thin blog posts optimised for a keyword, but content that actually answers the question with enough specificity to be useful. AI models pull from content that demonstrates genuine expertise, and they tend to surface brands whose content treats the reader as an intelligent adult. The Semrush overview of AI SEO considerations touches on how content structure and topical authority feed into AI visibility, and it is worth reading alongside your prompt testing data.

Editorial coverage means third-party mentions across credible sources. When I was running iProspect, we grew the team from around twenty people to over a hundred over a few years. One of the things I noticed during that period was how differently brands with strong PR footprints performed in any kind of third-party evaluation, whether that was analyst reports, award submissions, or later, AI-generated recommendations. The brands that invested in consistent earned media had a compounding advantage that was hard to replicate quickly.

Entity clarity means that the model has a consistent, unambiguous understanding of what your brand is, what it does, and who it serves. Brands with inconsistent positioning across their own channels, their PR coverage, and their partner ecosystem create confusion at the entity level. That confusion tends to result in lower citation rates or more qualified, hedged mentions when they do appear.

If you want a broader view of how AI is reshaping marketing practice across channels, the AI Marketing hub at The Marketing Juice covers the strategic and tactical dimensions in more depth.

Turning the Analysis Into a Competitive Brief

Competitive analysis is only useful if it produces something actionable. The output of an AI search competitive analysis should be a brief that answers three questions: where are we losing ground in AI-generated responses, why, and what do we do about it?

The “where” comes from your prompt testing data. Identify the specific query types where competitors consistently appear and you do not. These are your priority gaps.

The “why” comes from your content and entity analysis. For each gap, is the issue a lack of relevant content, a lack of editorial coverage, or a positioning inconsistency? These require different responses.

The “what” is where most teams go wrong. They treat AI visibility as a content production problem and commission a batch of new articles. Sometimes that is the right answer. But if the gap is editorial coverage, you need a PR brief, not a content calendar. If the gap is entity clarity, you need a brand and messaging audit. If the gap is content depth, you need a topical authority plan, not more thin posts.

Early in my career, I taught myself to code because the MD said no to a website budget. The lesson I took from that was not “be resourceful” in some generic sense. It was: understand the actual problem before you reach for a solution. The same principle applies here. AI visibility gaps look similar on the surface but have very different root causes, and the fix depends entirely on which one you are dealing with.

The brief should also include a monitoring plan. AI search responses are not static. Models are updated, retrieval patterns shift, and the competitive landscape changes. A monthly prompt testing cadence is the minimum. Quarterly deep-dives that include source analysis and entity audits are more thorough.

The Limits of This Approach

There are things this approach cannot tell you, and it is worth being honest about them.

First, AI search responses are not fully deterministic. The same prompt can produce different outputs on different days, across different accounts, or in different geographic contexts. Your prompt testing data is a sample, not a census. Treat it as directional intelligence, not a precise measurement.

Second, there is no direct attribution line between AI citation and commercial outcome. I have spent enough time managing P&Ls and ad spend to be deeply sceptical of any metric that cannot be connected, even loosely, to revenue. AI citation frequency is a leading indicator, not a lagging one. It tells you about brand positioning and consideration-stage presence. It does not tell you how many deals it influenced. Build the business case for this work carefully, and do not overstate what the data shows.

Third, this is a relatively new discipline and the tools are still catching up. Platforms like Semrush and others are beginning to build AI visibility tracking into their products, but the methodology is still maturing. The manual prompt testing approach I have described is more labour-intensive than a dashboard pull, but it is also more reliable right now because you control the inputs and can interrogate the outputs.

The Ahrefs AI SEO webinar with Patrick Stox is worth watching for a grounded perspective on what AI search means for organic visibility more broadly. It does not overclaim, which is refreshing in a space where overclaiming is the norm.

Where This Fits in a Broader Competitive Intelligence Programme

AI search competitive analysis is not a replacement for traditional competitive intelligence. It is an addition to it. You still need to track keyword rankings, share of voice in paid search, competitor content strategies, and pricing and positioning changes. AI search adds a new layer: how are competitors being framed in the synthesised responses that are increasingly the first thing buyers see?

The brands that will have an advantage in AI search over the next few years are not necessarily the ones with the biggest content budgets. They are the ones with the clearest positioning, the strongest editorial footprints, and the most consistent entity presence across the web. Those are things that take time to build, which is exactly why starting the analysis now matters.

I judged the Effie Awards for a period, and one thing that process reinforced was how rarely marketing effectiveness comes from a single clever tactic. It comes from clarity of strategy, consistency of execution, and the discipline to measure what actually matters. AI competitive analysis is no different. The prompt testing framework is the tactic. The strategic question underneath it is: what does it mean to be the credible answer to our buyer’s most important questions, and are we building toward that or away from it?

The AI Marketing section of The Marketing Juice covers how these shifts in search behaviour connect to broader questions of brand strategy, content investment, and marketing measurement. If this article raised more questions than it answered, that is probably the right place to continue.

About the Author

Keith Lacy is a marketing strategist and former agency CEO with 20+ years of experience across agency leadership, performance marketing, and commercial strategy. He writes The Marketing Juice to cut through the noise and share what works.

Frequently Asked Questions