When someone asks AI for a solution, does your brand appear? If so, in what role?
I check which AI answers mention you, which recommend you and which only link to your page – and what stands in the way. The analysis and recommendations stay with me.
A focused first project starts at €3,000, excluding applicable taxes. English-language hourly consultations and team training are not part of this offer.
I reply within two business days: e•••••@f•••••.pl
Ewelina Podrez-SiamaHost of the I ❤️ PRO: GEO & LLM Optimization conference
The engagement at a glance
Since 2009
Working in SEO. Today also on brand visibility in AI search.
One primary market
A defined question set and selected AI surfaces for the first phase.
Led by me
I interpret the evidence and make the recommendations myself.
Agency support
A team of specialists helps coordinate work in English and deliver agreed changes.
Two starting points
For brands crossing a market boundary
The first phase is deliberately narrow: one primary market and the questions that matter commercially there, checked on selected AI search surfaces. I add a second market or language only when a decision depends on the comparison.
01 / Entering Poland
Your brand is known elsewhere, but Polish customers ask different questions.
I collect the questions Polish customers ask and the competitors they see. Then I check whether Polish-language search and AI answers still recognize your brand once it crosses the border – and what got lost on the way. The report is in English.
Which Polish questions should you be present for?
Do AI systems connect your local site with the parent brand?
What needs local evidence rather than a translated page?
02 / Several markets
You have multiple sites and no common way to decide what to fix first.
I set up one way of measuring for the main market and, where useful, a second one, so the results can be compared. Then I work out what kind of problem you have. A brand problem needs different work from a content gap or a blocked bot, and sometimes nothing has changed except the way a tool counts. Better to know which before every team starts fixing its own version of the problem.
Do the domains present one consistent entity?
Which changes can be shared, and which must be local?
Choose the question first
Is the main problem how systems identify the person behind a brand?
If Google or AI answers confuse a founder or expert with someone else, or describe them with outdated facts, the work starts with their identity. See Personal brand SEO. When both the person and the buyer questions matter, I assess them together and propose one scope.
A mention, a recommendation and a linked source are different outcomes
A brand can be recommended without a link. A page can be used as a source without its publisher being recommended. I record those events separately, because they point to different decisions and different chances of a visit.
Mention
Your name appears in an answer, without an endorsement or a link. It tells us the system recognizes the brand in this context.
Recommendation
The system presents you as an option for a particular need. I check the reason, the audience and the alternatives named alongside you.
Linked source
An answer points to a page on your site. That creates a possible route to your content, though a citation alone is not a recommendation.
In Google Search: AI Overviews and AI Mode are the features where an answer appears. A supporting link to your page within one of them is an example of a linked source. I check link impressions in Search Console’s generative AI report and use your analytics to assess visits.
I use “AI search visibility” for this work. “Ranking in ChatGPT” is a useful shorthand for some buyers, but a generated answer is not a fixed list of positions that anyone can promise to secure.
A bounded project
What I investigate
I choose the checks that answer your question, not every check that exists. The proposal spells out what is in scope and what is not. There is no generic GEO checklist here – the evidence decides what comes next.
1
Buyer questions and competitors
I build a set of questions from your products and from the problems customers try to solve before they buy. I record which brands appear alongside yours and why their pages may be easier to use as sources.
2
Baseline across selected surfaces
I separate mentions, recommendations and linked sources. Repeated observations use documented conditions; I do not present one answer as your share of every private AI conversation.
3
Access and retrieval
I check what the site returns with and without JavaScript, what robots.txt and CDN or WAF rules let through, what server logs show where available, and whether the site is in Common Crawl. AI bots do different jobs. OAI-SearchBot, Claude-SearchBot and PerplexityBot fetch pages for AI search; GPTBot and ClaudeBot collect content that may be used for training; ChatGPT-User, Claude-User and Perplexity-User fetch a page when a user asks for it, and OpenAI and Perplexity say robots.txt rules may not apply to those fetches. Blocking GPTBot does not remove you from ChatGPT search. Opting out of OAI-SearchBot in robots.txt excludes the site from ChatGPT search answers, although it may still appear as a navigational link. WAF blocks require a separate access check. In Google, I also check that the site is not excluded from generative AI features in Search Console.
4
Content and site architecture
On the pages that matter most, I check whether the answer to a buyer’s question is stated clearly and still makes sense as a passage read out of context. I also look at the internal links leading there and at questions customers ask that the site never answers. Google says content does not need to be split into chunks for AI, so for me the passage test is about readability, not a ranking rule.
5
Entity and external evidence
I look at how the company, people, products and regional sites are connected through visible claims, structured data and credible references outside your own domains. If the main issue is an individual expert’s identity, see Personal brand SEO.
6
Decisions and ownership
I put the work in order and show the evidence behind that order. Then we split it: what I can change myself, what your team owns and where Fox Strategy specialists can help.
What you receive
An English report your team can act on
The deliverable is a decision document, not only a visibility score.
The agreed question set, recorded observations and method notes.
A comparison of the brand, competitors and linked sources.
A technical and content diagnosis tied to specific pages or barriers.
A roadmap that puts the work in order, with owners, dependencies and the next measurements.
A method you can challenge
How I keep the measurement honest
Language models are nondeterministic. The same question can produce a different answer later. I compare repeated runs of an agreed question set, record the conditions and read the trend; I do not treat a single screenshot as proof of a market-wide change.
01 / Public interfaces
Observe the answer as it is presented
Where possible, I use tools that automatically collect responses from public interfaces available without signing in. This is closer to what an anonymous user may see than an API-only test, but it does not reproduce logged-in history or personalization. I record the available model label, language, location, date and search mode.
02 / API and first-party data
Control what can be controlled
API runs allow defined parameters and repetitions. I keep them separate from interface observations. I also use available Search Console, Bing Webmaster Tools, ChatGPT Business Hub stats (for businesses with a verified profile), analytics and server logs to check links, visits and bot access. No one dataset reveals every private conversation with an assistant.
If a tool changes its model, interface, search function or counting rules, I flag that before comparing periods. Otherwise a neat trend line may describe a change in measurement rather than a change in your brand’s visibility. My server-log analysis shows why bot activity alone is not a result.
Personal direction, supported delivery
You know who makes the decisions
My role
I lead the thinking
I frame the research question with you and make the diagnosis from the evidence. The recommendations in the English report are mine, and so is the responsibility for them. When access and scope allow it, I also make selected WordPress or WAF changes myself. Sometimes fixing one blocking rule is faster than describing it in a report.
Fox Strategy’s role
The team helps deliver
Agency specialists support English-language project coordination and work that needs several disciplines or parallel implementation. The proposal names the people and responsibilities before the project begins; your brief does not disappear into an anonymous delivery queue.
01Send a brief
Tell me the site, target market, language and decision you need to make. I review fit before suggesting a scope.
02Agree the boundaries
The proposal fixes the scope and the fee, and names who is responsible for what.
03Research and roadmap
You get the documented baseline and diagnosis, plus a plan of work in English.
04Act and measure again
We scope implementation or ongoing monitoring separately, then return to the same questions and measurement conditions.
Format boundary: This is a project engagement. I do not currently sell one-hour consultations or team training in English. I lead the work and the conversations; specialists support coordination and delivery where needed.
The next useful step
Find the constraint before adding more content.
It may be a missing answer, a fragmented brand identity or a bot that never reaches the page. I will tell you what the evidence supports and what remains uncertain.
Starting fee€3,000for a defined first phase, excluding applicable taxes
The starting scope covers one primary market and one main website, a focused question set, selected AI surfaces and priority pages. You receive a documented baseline, findings on the most relevant barriers, an English report and a prioritized roadmap. I confirm the exact sample and checks in the proposal.
Deeper technical work across multiple sites, additional markets or languages, implementation, publication budgets and recurring monitoring are priced separately. I will propose a narrower diagnosis when it can answer the business question.
This fee buys research and direction for a specific decision. If foundational indexing or site access is the real blocker, the proposal will say so.
See how I reason
My work is open to inspection
Start with the conversation, then explore the English-language analyses. They show the questions I ask before recommending a fix and the limits of the evidence.
Watch · Marketer+ podcast
What does it mean for a brand to be visible in AI?
I talk with Sławomir Kruczek about mentions, citations and recommendations, and about the evidence behind those answers. It is a quick way to hear how I approach the work before reading a project proposal.
Language: The conversation is in Polish. To follow it in English, turn on captions (CC) on YouTube, then select Settings → Subtitles/CC → Auto-translate → English, if available.
Yes. I can research Polish-language questions and competitors while delivering the report in English. A Polish market project still needs local validation of what customers mean, not a word-for-word translation of an existing content plan.
Can you cover every market and AI platform at once?
No. The first engagement has one primary market and selected surfaces. We add a comparison market or language only when the decision requires it, with a separate scope and fee. A wider rollout can follow the first findings.
Does the starting fee include implementation?
It covers the agreed research, English report and roadmap. Implementation and recurring monitoring are scoped separately. I can handle selected technical changes myself; Fox Strategy specialists or your team can deliver larger work under an agreed division of responsibilities.
Will I speak with you personally?
Yes. I lead the conversations and the analysis, and I write the recommendations myself. Specialists join project meetings and implementation where agreed.
Can you guarantee that ChatGPT or Google will recommend us?
No. Generated answers vary, and no consultant controls which brand a system recommends. I can document the baseline, remove identified barriers, improve the evidence available to the systems and measure what changes over a comparable series of observations.
Start with context
Send me the decision you need to make
Send a short project brief by email or through the form on my contact page. I’ll review it and propose a defined scope – or tell you that your first step should be elsewhere. If a call helps, we’ll arrange one and confirm who joins.
Useful details for the first message
Your company and website.
The primary market and language, plus any market you want to compare.
The products, services or buyer questions that matter most.
What prompted the inquiry and what decision the work should support.
Ewelina Podrez-Siama
I lead the analysis and recommendations; a supporting team joins English-language meetings or implementation where agreed.
Cookies with no tricks at all. I only turn on analytics cookies (Google Analytics) if you agree – they help me see which content is genuinely useful. You can change your mind any time by clicking the cookie in the corner.