The touchpoint
Your brand is described before it is seen.
Until now, a brand was conveyed where companies could shape it: on their website, in their communication, in direct contact. Today the AI-reconstructed brand increasingly takes its place - in answers, recommendations, comparisons, and more and more often in direct interaction with customers.
The consequence
Why the description decides.
An AI answer is not a ranking. It is a recommendation in language - and it works through the same mechanisms as any other recommendation.
People do not decide on completeness, they decide on what connects to their own situation. A description can be factually correct and still miss the motive behind the decision. “Technically leading” and “dependable, even when it gets tight” can describe the same brand - and mean something entirely different to different people.
That is why the analysis does not end with the question of what AI says about your brand. What matters is whether that description connects where decisions are actually made - and what has to change for it to do so.
The situational picture
A prompt is not a situational picture.
Asking an AI system about your own brand is a sensible first test. Most analyses stop right there. For reliable brand management, a single answer is not enough.
Do you appear at all?
The baseline measurement: in what share of the relevant questions is your brand named, in what position, for which types of question - and how stable is that across repetitions and across different systems.
Stability is what matters here. Language models do not answer deterministically. A single mention is an event, not a finding. Only repetition separates pattern from coincidence.
What is measured, among other things:
- mention rate by question type
- position within the answer
- coverage across topic and product fields
- variance across repetitions and systems
- visibility across different AI channels and search environments
As what, next to whom - and by which criteria?
Presence alone says little. What matters is the classification: which attributes are ascribed to you, in what tone, in what role - and in which competitive environment do you appear?
The competitive set is the underestimated part. Whoever is named alongside the wrong providers gets compared by their criteria. Defining the comparison group has always been core to brand management; now it is set from the outside.
The criteria are no longer set by the brand either. Which attributes a system raises unprompted as decision-relevant determines what you are measured against - even where you never positioned yourself that way. This is assessed across all observed providers, not only for your brand.
And visibility is not automatically an advantage. What matters is the context your brand appears in, which attributes get attached to it, and whether that presence actually pays into your goals.
What is measured, among other things:
- ascribed attributes and tonality
- competitive set by question type
- interchangeability with the alternatives named
- misattributions and false associations
- reputation and trust signals
- criteria treated as decision-relevant, by question type
- attributes no provider in the set occupies
- recurring objections and uncertainties
Which content and which sources shape that picture - and why?
This is where observation parts ways with consulting. The question is no longer what is being said, but what it comes from: which of your own content, which third-party sources, carrying what weight - and whether your content can technically be captured the way you assume it can.
Not every source counts the same. Whether a brand turns up in a trade publication, at a public authority or in a directory makes a difference to the picture that emerges. The reverse direction matters just as much: in which kinds of source is the competition named - and you are not?
The same applies, more finely, to your own site. Which of your content is actually drawn on, which gets fetched but never cited - and which page characteristics measurably go together with higher consideration? Connections like these are tested for how well they hold before they turn into a recommendation. A pattern is not yet a cause.
And the analysis does not end at the AI answer. What people find on your website afterwards, and what actual behaviour follows from it, matters just as much.
What is measured, among other things:
- sources drawn on and sources citing you, their authority and weight compared with the competition
- share of own versus third-party content in the picture that emerges
- technical accessibility and delivery differences between human and machine
- connections between page characteristics and consideration
- gaps between relevant questions and existing content
- pages that are fetched but never drawn on
This is where analysis turns into a decision.
How far is the external picture from your self-image? Which deviations are relevant? And which of them actually carry economic weight?
And it is not only about visibility. What counts is brand strength, differentiation, relevance and esteem - and whether the picture that emerges meets the decision motives of your audiences.
What is measured, among other things:
- deviation between self-presentation and external description
- fit with the motives behind decisions
- brand strength and how it develops
- differentiation from the competition
- prioritization by impact and effort
- change over time, where observation is continuous
Method
An observation is not yet a finding.
LLMs do not answer deterministically. The same question can lead to different answers, brands can be confused with one another, and apparent connections can have entirely different causes. A shift in the brand picture can equally stem from a changed market, an active competitor or a model update - telling those causes apart is part of the evaluation.
That is why I do not work with individual answers or isolated metrics. Findings are cross-checked across several systems, sources, questions and points in time. Measurements are documented and connections are tested for how well they hold.
A signal is not yet a recommendation. Every relevant finding goes through a critical counter-check: does the connection hold? Is the proposed measure actually relevant? Or is a measurable signal simply being mistaken for business impact?
What does not survive that check does not become a recommendation.
This way of working is not new. It is shaped by 30 years of digital practice, published expertise and forensic collaboration - wherever statements must not merely sound plausible, but hold up to scrutiny.
Collaboration
AI-first. Not AI-only.
AI-first brand management does not mean delegating brand management to AI. AI can capture, compare and evaluate large volumes of signals. The strategic decision stays with people.
I deploy AI where it makes analysis and delivery better and faster - and combine it with human judgment, contextual understanding and strategic responsibility.
Because visibility is not a target in itself. It is a variable - and it carries a sign.
A brand does not have to appear everywhere. It has to be relevant and correctly classified where its perception counts for decisions.
Services
Five ways in, on their own or building on one another. Which one makes sense follows from the diagnosis - not the other way round.
For organizations that know they have to act but do not yet have a framework for it.
Half a day or a full day in which we determine together where AI meets your business - in perception, in decision-making, in operations.
You see live how AI systems answer about your own brand, and how much the wording of a question decides. The result is a set of priorities you worked out yourself - not one I hand you.
Result: prioritized fields of action · shared understanding across the leadership team
The full assessment across all four levels: presence, image, origin, impact.
Several systems, several questions, repetitions for validation, comparison with the competitors actually named alongside you.
You receive a report with findings, interpretation and recommended actions - plus a version that works in a board meeting.
Result: situational picture · market picture and competitive position · root-cause analysis · brand strength · prioritized list of measures
Whether AI systems can capture your content at all in the way you assume.
Access for the relevant systems, delivery differences between human and machine, structured data, unambiguous identification of your brand.
This is the hygiene layer. It does not create a brand image - but without it, no content measure takes effect.
Result: technical findings · implementation backlog for your team or your agency
What has to be built and written so the intended picture can emerge.
Which content is missing, which exists in the wrong form, which third-party sources work against it.
Concept, prioritization and support through delivery - with AI where it brings speed and scale, and with human decisions where context and brand matter.
Result: structure and content concept · delivery plan with clear ownership
Brand images in AI systems move slowly.
Impact does not show within weeks. Shifts in the competitive environment and changed decision criteria are only noticed, without measurement, once the enquiries stop coming.
Continuous assessment with a regular review conversation: is perception shifting in the intended direction, what is moving among competitors and criteria - and where does it need correcting?
Result: ongoing situational picture · brand strength over time · shifts in the competitive and market picture · evidence of impact for the measures implemented
Let’s make visible what AI is doing with your brand.
Before the market decides it for you.