- Many marketers know they need to optimize for AI search, but most still do not know which prompts matter or what AI systems currently say about their brand.
- Adobe says 80% of companies have significant gaps in how their content appears in LLM results, showing that AI search is still as much a measurement problem as an optimization problem.
Most Brands Still Do Not Know What AI Search Says About Them
AI search optimization has become one of the loudest topics in digital marketing.
The advice is everywhere. Build authority. Earn citations. Create comparison content. Strengthen entities. Optimize for ChatGPT, Perplexity, Gemini and Google AI Overviews.
The problem is that many brands are being handed advanced tactics before they have completed the first basic step: finding out what AI systems actually say about them.
“I don’t know how to do AI SEO”
Just start at level 1.
The problem with most AI search advice is that it feels like 100 things to do all at once.
Instead, think of it like a video game:
– Complete level 1
– Unlock level 2
– RepeatHere’s my 9-level AI search game:
Level 1:… pic.twitter.com/XRhKCZy7yX
— Jake Ward (@jakezward) June 16, 2026
Digital marketing strategist Jake Ward framed the problem in a post on X, arguing that most AI SEO advice feels like “100 things to do all at once.” His solution was to treat the process like a game: complete level one, unlock level two and repeat.
That framing works because it cuts through the confusion.
For most brands, AI search optimization should not start with a complex GEO playbook. It should start with visibility: which prompts matter, where does the brand appear, where do competitors appear and which sources are shaping the answer?
AI SEO Starts With Prompt Visibility
Ward’s first level is simple: build a list of prompts where the brand should appear, group them by topic and intent and track them across major AI systems.
That sounds basic, but it is where many brands are still missing structure.
Traditional SEO starts with keywords. AI search starts with prompts, questions and recommendation scenarios.
A user may not search “best CRM software” in the old way. They may ask ChatGPT, Perplexity or Gemini which CRM is best for a specific company size, budget, use case or industry.
That changes the measurement problem.
There is no fixed rank one in ChatGPT. There is no stable SERP that every user sees.
A brand may be mentioned in one answer, ignored in another and described differently depending on the prompt, model, source mix and context.
That is why prompt visibility comes before optimization.
Before a brand can improve its AI search presence, it needs to know:
- Which prompts matter commercially.
- Whether the brand appears in those answers.
- Which competitors appear instead.
- How the brand is described.
- Which sources AI systems cite or seem to rely on.
This is also why AI search connects closely to the broader measurement problem we covered in our analysis of AI search visibility tools. Marketers want clear rankings and dashboards, but the underlying system does not behave like traditional Google Search.
Adobe Says Most Companies Have Gaps
The need for a starting point is supported by Adobe’s own research.
In a post about building a GEO practice, Adobe said its LLM Optimizer research found that 80% of companies have significant gaps in how their content surfaces in LLM results.
That is the key issue.
Many companies are not only under-optimized for AI search. They also do not have a clear view of where they are visible, where they are missing and what AI systems are saying about them.
That makes AI SEO different from a normal content checklist.
A brand may have strong website content and still be absent from AI-generated answers. It may rank well in Google and still lose visibility inside Perplexity or ChatGPT.
It may be mentioned by an AI system, but described in a vague, outdated or inaccurate way.
Those are not only content problems. They are measurement and source problems.
The First Audit Should Be Simple
The first useful AI search audit does not need to be complicated.
A brand can start by collecting 30 to 50 prompts across four groups:
- Branded prompts, such as “What is [brand] known for?”
- Category prompts, such as “best tools for [use case].”
- Comparison prompts, such as “[brand] vs [competitor].”
- Buying-intent prompts, such as “which [product/service] should I choose for [specific need]?”
Those prompts should then be tested across the AI systems that matter most for the audience, including ChatGPT, Perplexity, Gemini and Google AI Overviews where relevant.
The goal is not to pretend this creates a perfect ranking report. It does not.
The goal is to create a baseline.
Does the brand appear? Are competitors mentioned more often? Are the cited sources accurate? Are third-party pages shaping the answer?
Is the company described clearly? Are the most important products, services and use cases connected to the brand?
Without that baseline, the rest of AI SEO becomes guesswork.
Comparison Content Is Where AI Decisions Often Happen
Once a brand understands where it appears and where it is missing, the next step is usually source improvement.
AI systems often lean on pages that summarize, compare or validate options. That can include comparison lists, “best of” pages, review content, Reddit discussions, trusted industry publications and third-party explainers.
Search Engine Land’s GEO guide makes a similar point: generative engine optimization is not only about a brand’s own website. It also depends on the sources AI systems use to discover and validate information.
That makes comparison-led content more important.
Users often turn to AI systems when they are trying to narrow options. They ask which tool, agency, platform, service or product fits a specific situation.
If a brand is missing from those recommendation paths, it may never reach the buyer’s shortlist.
This is the same broader shift we covered in our article on why AI search is making generic brand content easier to ignore. AI systems need clear, useful and externally supported information. Generic brand pages are often not enough.
Entity Consistency Still Matters
The more advanced part of AI SEO is entity consistency.
AI systems need to understand what a brand is, what it does, who it serves and which topics it should be associated with.
If that information is inconsistent across the web, the model has less reason to describe the brand clearly.
Lumar’s GEO framework argues that AI search visibility depends partly on how well a brand is understood as an entity and connected to relevant topics.
For marketers, that means the basics still matter:
- Clear brand descriptions across the website and third-party profiles.
- Consistent product, service and category language.
- Strong author and expert profiles where relevant.
- Mentions on trusted industry sites.
- Useful comparison and decision-stage content.
- Structured pages that make the brand’s positioning easy to understand.
This is not separate from SEO. It is an extension of the same visibility problem across a new discovery layer.
What Marketers Should Do Now
Most brands with weak AI search visibility are not missing the advanced tactics.
They are missing the first audit.
They have not defined which prompts matter. They have not checked what ChatGPT, Perplexity, Gemini or Google AI Overviews say about them. They have not identified which competitors appear instead. They have not mapped which sources are shaping the answers.
That should come before bigger GEO projects.
Adobe’s own example shows why. According to Adobe’s published findings, applying structured GEO work to its own properties led to a 5x increase in citations for one product line and a 41% lift in LLM referral traffic within weeks.
Those results will not apply to every brand. But the lesson is useful: improvement starts with visibility, measurement and source gaps, not vague advice to “optimize for AI.”
The Query Post View
The problem with AI SEO advice is that it often starts too late.
Many brands are being told to build authority, dominate topics, earn citations and optimize for LLMs before they have answered the basic question: what do AI systems currently say about us?
That should be step one.
Pick the prompts that matter. Test them across ChatGPT, Perplexity, Gemini and Google AI Overviews. Record where the brand appears, which competitors are mentioned and which sources are being cited.
Only then does the rest of the strategy make sense.
AI search optimization does not start with a 50-point checklist. It starts with visibility: are you present, are you described correctly and are the right sources shaping the answer?
