The GEO Revenue Engine Playbook

A note on why this playbook exists

Selling regulated fintech in Europe used to be a distribution problem. You had a product, you had a market, and the job was to buy enough attention — paid search, paid social, sponsorships, retargeting — to get in front of the 3% of buyers who were ready to move.

That model is quietly dying for MiCA-regulated firms, for two reasons that arrived at almost the same time.

First, the rails you'd use to buy that attention are being pulled up. Under MiCA, and the European Securities and Markets Authority (ESMA) guidance that sits underneath it, almost every channel a growth team reaches for — internet commercials, banners, pop-ups, social ads, affiliate campaigns, retargeting, influencer content, even sponsorships and event invitations — is now a regulated communication subject to the "fair, clear, and not misleading" standard, and in many cases must be consistent with a published crypto-asset white paper (ESMA, Guidelines on reverse solicitation under MiCA). The old growth-hacking playbook doesn't just underperform in this market. Large parts of it are legally radioactive.

Second, and more fundamentally, the buyer stopped starting on Google. A compliance officer evaluating an AML platform, a CTO scoping custody infrastructure, a CFO comparing stablecoin rails — increasingly, their first move is to ask an AI. "What are the leading transaction-monitoring tools for a MiCA-licensed CASP?" "Which providers handle Travel Rule compliance for EU crypto firms?" The answer that machine gives — the specific names it says out loud — now decides who makes the shortlist before a single human at your company knows the opportunity exists.

If you are not in that answer, you are not slow. You are invisible. And in a 6-to-18-month, multi-stakeholder, board-level buying cycle, invisible at the start usually means absent at the finish.

This is the intersection nobody has written the playbook for: a market that can't advertise, selling to a buyer who starts with AI. So we wrote it.

This is not a theory document. It is an operating manual. It explains how the machine actually works, why MiCA changes the math, and — critically — how to build the revenue engine that captures and converts the sliver of high-intent demand that GEO produces. Because visibility without a capture-and-nurture system behind it is just expensive awareness.

How to use this playbook

Read for your role. Each path picks up the chapters where your part of the deal is won or lost.

Every statistic in this document is cited to its primary source. Where the data is young or contested, we say so. There are no growth hacks here, because there are none. There is a system.

Table of contents

The Invisible Fintech Problem

A story you've already lived

A compliance officer at a mid-sized European payments firm has a problem. Her institution is applying for a crypto-asset service provider (CASP) licence, and her existing transaction-monitoring stack won't pass a MiCA-standard audit. She needs a new one. She has budget. She has authority to shortlist. She is, in the language of sales, a live, qualified, in-market buyer — the 3% every vendor in your space is desperate to reach.

Here is what she does not do. She does not type a keyword into Google and click through ten blue links. She does not respond to a cold email. She does not see your LinkedIn ad, because — as we'll cover in Chapter 1 — you may not legally be allowed to run it.

Here is what she actually does. She opens ChatGPT, or Perplexity, or Google's AI Overview, and she types something like: "What are the best MiCA-compliant transaction monitoring platforms for a CASP handling stablecoin flows?"

And the machine answers. It names three or four vendors. It describes what each is good at. It sounds authoritative, because it is fluent and confident, and she treats it as a serious starting point — the same way she'd treat a recommendation from a trusted peer.

This is the invisible fintech problem, and it is happening thousands of times a day across every category MiCA touches.

This is a step-change, not a trend

It is tempting to file "AI search" under the same folder as every other marketing fad — a thing to dabble in, delegate to an intern, and revisit next year. That instinct is wrong, and for regulated fintech it is dangerous, because two structural forces are compounding.

Force one: search is decoupling from clicks. As of 2025, roughly 60% of Google searches end without a single click to any website — a figure that has climbed from around 25% five years earlier (Superprompt, zero-click search analysis). Google's AI Overviews, which barely existed at the start of 2025, expanded from appearing on 6.49% of queries in January to roughly 18–20% by mid-year (SeoProfy, Google AI Overviews statistics; Search Engine Land). The answer is increasingly the destination, not a signpost to your site.

Force two: an entire generation of buyers has moved their research into conversational AI. The point is not that Google is dead. It is that the first, framing question — the one that decides who's even in consideration — is now often asked of a machine that responds with a short list of names rather than a page of options.

Neither force is reversing. Both favour the same thing: being the brand the model already associates with your category, and being the page it pulls and quotes when it searches live. That is what Generative Engine Optimization (GEO) is. It is not SEO with "AI" bolted on. As we'll show in Chapter 4, it's a different machine with a different win condition, and the habits that made you good at SEO can quietly work against you.

Why this is uniquely brutal — and uniquely winnable — for MiCA fintech

Every B2B company faces the AI-search shift. But most of them can compensate. If ChatGPT doesn't recommend a project-management tool, that vendor can still buy Google Ads, run LinkedIn campaigns, retarget visitors, and sponsor a podcast. They have a full deck of channels.

You don't. MiCA and ESMA guidance have turned most of those channels into regulated communications — constrained, disclosure-heavy, and in some configurations simply not permitted (Chapter 1). The paid escape hatch that every other category uses to buy its way back into visibility is, for you, welded shut.

That sounds like bad news. Reframe it. The constraint is the moat. In a market where nobody can out-spend anybody on ads, the firm that wins is the one that earns the machine's trust through presence, consistency, and genuinely useful, quotable content — exactly the things MiCA rewards rather than restricts. Being clear, accurate, and substantive is both the winning GEO strategy and the compliant one. Your competitors who are addicted to paid acquisition are about to discover their favourite tools are off-limits. You can be building the one asset that compounds and can't be switched off.

The 3% problem, applied to you

Business author Chet Holmes' framing — that at any moment only about 3% of your market is actively buying, another 7% is open to it, and 90% is not thinking about it at all — is the quiet foundation of the i3strategies FinCrime playbook, and it's true here too. But GEO changes where the 3% show up first, and MiCA changes how you're allowed to reach the other 97%.

Every chapter that follows serves one goal: make you the named brand at the start, and build the engine that converts that naming into revenue across a long, cautious, multi-stakeholder sale.

What you'll be able to do by the end

• Explain, precisely, why your buyers can't find you and what to do about it.
• Map the AI-first journey for each stakeholder in a regulated deal.
• Build GEO visibility that survives how models actually store and retrieve knowledge.
• Stand up the website and capture infrastructure that turns a rare, high-intent GEO visitor into a fast-followed lead.
• Run the multi-month nurture sequences that are, in this market, the difference between a pipeline and a graveyard.
• Do all of it inside MiCA's guardrails — using them as an advantage, not fighting them.

Let's start with the market, and the squeeze.

The MiCA Market and the Regulatory Squeeze

What MiCA is, in one paragraph

The Markets in Crypto-Assets Regulation (Regulation (EU) 2023/1114, "MiCA") is the EU's unified rulebook for crypto-assets that aren't already covered by existing financial-services law. It governs the issuance, public offering, and trading of crypto-assets and the provision of crypto-asset services, with the stated aim of protecting consumers, ensuring market integrity, and supporting financial stability (ESMA, Markets in Crypto-Assets Regulation overview). It applies in two phases: the rules for asset-referenced tokens (ARTs) and e-money tokens (EMTs) — the stablecoin regime — have applied since 30 June 2024, and the rules for crypto-asset service providers (CASPs), including authorisation and conduct, have applied since 30 December 2024 (Sumsub, MiCA in the European Union).

If you issue tokens, run an exchange or custody service, offer crypto payments, or sell software and infrastructure into firms that do, MiCA either governs you directly or governs your customers — which means it governs how you're allowed to talk to the market.

The market you're actually selling into

MiCA created something the crypto industry never had: a single, licensed, supervised European market. From 30 December 2024, any firm providing crypto-asset services to EU clients must hold a CASP authorisation from a national competent authority (NCA) — a full financial-services licence with capital requirements, governance obligations, conduct rules, and ongoing supervision. A prior national VASP registration no longer satisfies the requirement (Citium Tech, MiCA for crypto-asset service providers).

The practical effect on your go-to-market:

• Your buyers are now licensed institutions, not startups in a legal grey zone. They behave like banks: cautious, committee-driven, audit-minded, slow. The "move fast and break things" energy is gone. Moving slowly is safe; moving slowly avoids regulatory risk; avoiding risk is in the buyer's DNA. (If that sentence sounds familiar to FinCrime veterans, it should — MiCA has made crypto buyers behave exactly like traditional compliance buyers.)
• The buying committee expanded. A licensed CASP has an MLRO, a compliance function, a risk committee, and a board that signs off on vendor risk. Your deal now touches all of them (Chapter 3).
• The category is crowded and confusing. Just as the FinCrime software market ballooned to 400+ vendors, the MiCA-adjacent tooling market — monitoring, screening, Travel Rule, custody, KYC/KYB, wallet analytics, reporting — is filling with lookalike vendors using near-identical buzzword language. Buyers can't tell you apart. The single biggest, cheapest advantage available to you is clarity: being unmistakably specific about what you do and who you're for. That clarity is also, not coincidentally, what makes an AI model recommend you (Chapter 4).

The squeeze: what MiCA actually restricts about marketing

Here is the part most growth teams have not fully internalised. Under MiCA, marketing communications are a regulated activity. Article 7 requires that any marketing communication relating to a public offer or admission to trading of a crypto-asset must be:

• clearly identifiable as marketing — you cannot dress an ad up as news, research, or neutral commentary;
• fair, clear, and not misleading — honest, understandable, and free of material omissions;
• consistent with the crypto-asset white paper where one is required — your marketing claims cannot outrun your disclosures;
• accompanied by a clear statement that a white paper has been published, with the relevant website address (Article 7 MiCA, marketing communications).

For CASPs, Article 66 imposes a parallel "fair, clear, and not misleading" conduct standard on communications with clients and prospects — substantively close to the MiFID II Article 24 standard applied to traditional investment marketing, with crypto-specific overlays (Sedric, CySEC MiCA Compliance Guide 2026).

And the definition of what counts as solicitation — and therefore falls inside the perimeter — is deliberately broad. ESMA's guidance lists, without limitation: internet commercials, brochures, telephone calls, emails, banners, pop-ups and similar tools on websites and social media, face-to-face meetings, press releases, websites, mobile apps, roadshows and trade fairs, event invitations, affiliation campaigns, retargeting of advertising, response forms, training courses, messaging platforms, and sponsorship deals (ESMA, Guidelines on reverse solicitation under MiCA).

Read that list again with a growth marketer's eyes. It is, almost item for item, the modern demand-generation stack.

The channels that closed (or got dangerous)

This is not a claim that a MiCA-regulated firm can never advertise. It is a claim that every one of these channels now carries compliance cost, disclosure burden, approval overhead, and legal risk that did not exist before — and that some formats (disguised endorsements, return-baiting, advertorial) are effectively off the table. The cost-benefit of paid acquisition has collapsed. Speed and creativity, the two things paid channels used to buy you, are exactly what the regulation strips out.

The channels that opened

The regulation doesn't leave you with nothing. It leaves you with the channels that reward substance over spend:

Market sizing: how to think about it without kidding yourself

Founders and investors love a Total Addressable Market number. For MiCA tooling and infrastructure, resist the temptation to quote a headline TAM as if it were precise. As the i3strategies playbook rightly argues for FinCrime software, TAM depends entirely on your geography, product, and delivery model — so the honest, useful number is your Serviceable Addressable Market (SAM): the specific set of licensed CASPs, token issuers, and regulated payments firms in your target jurisdictions that can actually buy your specific product this year.

Two things about that market are not in doubt:

1. It is growing and consolidating simultaneously. MiCA is pulling firms out of the grey market and into licensed status, and each licensed firm acquires a compliance-tooling budget it didn't have before. The number of qualified buyers is rising.
2. It is a slow, high-consideration market. Licensed institutions buy like banks. Expect 6–18 month cycles, multiple stakeholders, and procurement scrutiny (Chapters 2, 3, and 8).

The AI-First Buyer Journey

The journey didn't get shorter. The starting line moved.

There's a comfortable myth that AI just "speeds things up." For regulated fintech, that's backwards. The sales cycle is as long as ever — 6 to 18 months, sometimes more, because a licensed institution cannot and will not rush a vendor onto its risk register. What changed is where the journey starts and how buyers gather information along the way.

The critical shift is that consideration is now set at the top of the funnel, by a machine, before you have any signal that a deal exists. By the time a lead form gets filled in, the buyer has often already asked an AI about you a dozen times — and about your competitors.

Why regulated-fintech buyers start with AI specifically

This behaviour is more pronounced in regulated fintech than in almost any other B2B category, for reasons rooted in the buyer's situation:

• The domain is complex and fast-moving. MiCA interpretations, Travel Rule thresholds, ESMA guidance, jurisdictional nuances — buyers use AI to get oriented quickly in a landscape that shifts monthly.
• The buyer is often not a technologist. A compliance officer scoping monitoring software is an expert in regulation, not in software architecture. AI translates between those worlds, which is exactly why they lean on it.
• They can't trust vendor marketing. In a market of 400+ lookalike vendors all using the same buzzwords, buyers distrust vendor claims by default. An AI answer feels like neutral synthesis — a trusted-peer recommendation rather than a sales pitch — even though its "opinion" was shaped by what the web says about you.
• The stakes are personal and career-defining. Choosing the wrong compliance vendor isn't a productivity annoyance; it's potential regulatory failure with the buyer's name on it. That fear drives exhaustive, repeated research — and AI is the fastest way to conduct it.

The zero-click reality

The macro data confirms what the anecdotes suggest. Around 60% of Google searches now end without a click to any website (Superprompt, zero-click analysis). Google's AI Overviews scaled from 6.49% of queries in January 2025 to roughly 18–20% by mid-year, and where they appear, organic click-through rates fall sharply (SeoProfy, AI Overviews statistics; Dataslayer, zero-click SEO data). Meanwhile conversational assistants — ChatGPT, Perplexity, Gemini, Copilot — have pulled a large share of research-style queries out of traditional search entirely. Research from G2 found that roughly half of B2B software buyers now begin their research with AI chatbots (G2, via PR Newswire), and a Gartner survey found 69% of B2B buyers turn to sales reps specifically to validate AI-generated insights — confirming that the machine now frames the deal before a human is involved (Gartner, via Business Wire).

Translation for your funnel: a growing majority of the "searches" that used to send a visitor to your website now resolve inside an answer the buyer never leaves. If you measure success by organic sessions, you are watching the wrong dial go down while the real game moves somewhere you're not looking. The right dials — share of voice in AI answers, citation frequency, brand mentions in model responses — are covered in Chapter 5.

The journey, stage by stage

Here's how a regulated-fintech deal actually unfolds in an AI-first world, and where GEO and the revenue engine intervene at each stage.

The two truths this journey forces on you

Truth one: presence at the start is necessary but not sufficient. Being the named brand in Stage 1 gets you into the consideration set. But a 6–18 month journey with five stakeholders will grind down any vendor who shows up once and then goes quiet. Visibility without a nurture engine is a leaky bucket.

Truth two: the buyer is researching you when you have no idea they exist. For most of Stages 1–3, you have zero signal. No form fill, no email, no call. The buyer is forming opinions about you inside a machine you can't see. Your only lever during that invisible period is what the web — and therefore the model — already says about you. That is GEO. It is the only marketing that works when you don't yet know the buyer exists.

The Multi-Stakeholder Buying Committee

One deal, many machines

In Chapter 2 we said each stakeholder runs their own AI research from their own angle. This is the operational heart of the modern regulated sale, and it's worth stating plainly: there is no single "buyer query." There is a portfolio of queries, one per persona, and your visibility is decided separately in each.

The CTO asks the machine about architecture, latency, and security posture. The MLRO asks about regulatory coverage and false-positive rates. The CFO asks about total cost and ROI. The CEO asks whether you're a credible, durable partner. The board asks about vendor risk. These are different questions that surface different vendors — and you can be the top recommendation for one persona and absent for another. Winning the deal means winning enough of these separate AI verdicts to build consensus.

This is why a single, generic "we're the leading platform for X" message fails. It's optimised for no one's actual question.

The personas, and what each one asks the machine

Below is the core buying committee for a MiCA-regulated deal. For each: what they care about, the fear that drives them, the kind of question they ask AI, and what your GEO and nurture must supply.

How the context shifts by ICP — a worked example

The same product surfaces differently depending on who's asking and what firm they're at. Take a single product — a transaction-monitoring platform — across three ICPs:

• A stablecoin (EMT) issuer asks about reserve-flow monitoring and redemption surveillance. The machine should associate you with EMT-specific obligations.
• A crypto exchange (CASP) asks about real-time trade surveillance and market-abuse detection. Different framing, different query, different desired association.
• A crypto-friendly payments firm asks about mixed fiat-crypto flows and Travel Rule interoperability. Different again.

If your content and web presence only ever describe you in generic terms ("transaction monitoring for crypto"), the model has nothing to latch onto for any of these specific queries, and a more specific competitor wins each one. Specificity per ICP is not a nice-to-have. It is how you get named in narrow, high-intent queries — which are exactly the queries a serious buyer asks.

The consensus sale: why one champion is never enough

The single most important structural fact about regulated-fintech selling: deals don't die because someone says no. They die because consensus never forms. A champion who loves you but can't get the MLRO, the CTO, and the board aligned will watch the deal stall for months and then quietly evaporate.

This has two direct implications for the engine we're building:

1. GEO must win multiple personas, not one. Being the CTO's favourite while being invisible to the MLRO loses the deal. Your content strategy (Chapter 5) must deliberately create quotable, retrievable material for each persona's query set.
2. Nurture must arm the champion to sell internally. The most valuable thing you can send a champion is not another product pitch — it's the material that lets them win the argument with their own risk committee. Chapter 7 is built around this.

Mapping it: the persona query matrix

For every ICP you serve, build a simple matrix (template in the appendix):

This matrix is the bridge between "who buys" and "what we publish." Every asset in your GEO program should map to a cell in it.

How GEO Actually Works

Why you need the mechanism, not just tactics

Most "GEO tips" articles hand you a checklist without explaining the machine, so you can't tell which tactics matter or adapt when the tools change. This chapter explains how the machine actually finds and names information — because once you understand the mechanism, the tactics in Chapter 5 become obvious rather than arbitrary. (This section builds directly on the model our colleague laid out in "GEO is not SEO: how LLMs really find information.")

The win condition flips

Start with the single most important difference. With Google, you fight to be a link on a page of options. The buyer still chooses. With an AI, you fight to be the answer itself — the specific name the model says out loud. There is no page two. You're either in the sentence or you're invisible.

That flip is the first sign that GEO is a different job, not SEO with a few AI tweaks. In SEO you optimise for position. In GEO you optimise for probability — the likelihood that when the model reaches "the best tool for a MiCA-licensed CASP is ___", your name is the word it fills in.

Mechanism one: trained-in memory (the model "just knows" you)

Before a buyer ever types a word, the model was trained. It read an enormous slice of the internet — websites, articles, forums, documentation, reviews — and adjusted billions of internal weights to capture the patterns in all that text.

Crucially, it does not keep a copy of those pages. The right analogy is how you know your favourite film: you can't recite the script, but you deeply know the characters, the plot, the feel. That's compression — the details are gone, the patterns remain. An LLM's knowledge of your brand works the same way. It doesn't store your homepage. It stores an impression of what the web collectively says about you.

So the model's built-in knowledge is a statistical echo of its training data. Three consequences follow, and they are the foundation of GEO strategy:

1. If the web talks about you often, clearly, and consistently — always tying your name to the right category — that echo is strong and accurate. The model "just knows" you belong when someone asks about your space.
2. If you're barely mentioned, or the web is confused about what you do, the echo is faint or wrong — and the model will happily "remember" you incorrectly, or not at all.
3. Consistency of association is what survives compression. Scattered, contradictory, or thin coverage gets averaged into nothing. The same clear description of who you are and what you're best at, repeated across many credible places, is what makes it through.

You can't edit the model's memory directly. But you shaped it, and you keep shaping the next version. Your goal is to make the web's story about you abundant, consistent, and unambiguous.

Why this is the mechanism that matters most for MiCA fintech: it's the one that works when the buyer is researching you invisibly (Chapter 2), and it's the one that can't be switched off by an ad-platform policy change. It's also the one that rewards exactly the behaviour MiCA demands — clear, consistent, non-misleading description of what you do.

How the model builds a sentence — and where you compete

The model doesn't write an answer whole. It breaks language into tokens (word-chunks) and predicts the single most likely next token, adds it, predicts the next, and so on at speed. Everything it "knows" shows up as which word gets the highest probability.

Most tokens are boring and obvious. But at the moment the model has to name a brand, several companies are literally competing to be the high-probability word. Getting cited is, mechanically, about becoming the most probable next token when the model reaches a slot like "the best Travel Rule solution for EU CASPs is ___". That probability was set in training by how often, and how confidently, the web pairs your brand with that exact context.

This is why being named inside relevant, well-written sentences beats a thousand keyword-stuffed pages. "For MiCA-licensed exchanges, [Brand] is the go-to transaction-monitoring platform" — repeated, in credible places, in natural language — is what tilts the probability. You're not optimising a rank. You're tilting a distribution.

The frozen-memory problem (and why presence compounds slowly)

Because trained-in knowledge is baked in during training, it stops at a date. Picture an expert who walked into a cabin with no internet on a certain day: everything before it they may know cold; everything after — last week's launch, your new product, yesterday's review — simply isn't in their head. Ask about something recent and the model will either admit it doesn't know or, worse, confidently invent something plausible (a "hallucination").

Two implications for your strategy:

• Presence compounds with a lag. Content you publish today mostly influences the next training cycle, not the model already shipped. GEO is a compounding asset, not a growth hack — which is exactly why starting now, before competitors do, is the advantage.
• The frozen-memory problem is why assistants bolted on live search — the second, faster mechanism.

Mechanism two: live retrieval (the model looks you up mid-answer)

This is where SEO people assume GEO collapses back into their world. It doesn't. Modern assistants — ChatGPT with search, Perplexity, Google's AI Overviews, Gemini, Copilot — can call a search tool mid-answer. But this is not the ten blue links. The model runs the search for itself, casts a wide net, reads the best passages, and rewrites them into its answer.

When a buyer's question looks recent, specific, or beyond what the model confidently remembers, it reaches for search instead of answering from memory alone. Then — and this is the key point — the model isn't ranking you, it's selecting a passage to quote and deciding which brand to name, guided by the same trained instincts from mechanism one.

The evidence here is young and studies disagree — some still show high overlap between AI-retrieved results and traditional top-10 rankings (e.g. ChatGPT matching Bing's top results a large share of the time), and we flag that honestly (Search Engine Land, AI Overviews data). But the mechanism and the trend point the same way: passage relevance and brand presence matter more than raw position. One line of research even found brand-search volume predicts AI citations better than backlinks.

So classic SEO hygiene — be crawlable, be indexed — is the price of entry for the retrieval lane, not the strategy. Even here the model is selecting a quotable passage and a brand to name, so the retrieval lane rewards the clearest, most quotable answer plus real brand presence. SEO gets you into the room; GEO decides whether you get quoted.

The context window: winning the model's "desk"

There's a third dynamic worth understanding. Everything in a single conversation — the buyer's prompt, the pages just retrieved, the earlier back-and-forth — sits in the model's context window: its working desk. And the model weighs what's on the desk very heavily, often more than its hazy long-term memory.

So when your page is one of the few things on the desk — because you won retrieval — your framing can override the model's baked-in impression. The retrieved text is fresh, specific, and right in front of it, so it tends to trust and echo it. If you win retrieval, you effectively get to write part of the model's context for that answer. Structure pages so the facts a model wants — who it's for, what it does best, proof — are easy to find and lift. The page that lands on the desk in clean, quotable form tends to become the answer.

The two paths in — neither is "rank #1"

There are exactly two ways your brand ends up in what the model says:

Serious GEO works both at once. And notice: neither is "be number one on Google."

GEO vs. SEO: what carries over, what hurts you

The trap is treating your GEO plan as your SEO plan wearing a new hat. If it is, you're optimising for the wrong machine.

The compliance dividend

Here's the part that should make a MiCA-regulated firm sit up. Everything that makes you quotable and trusted by a model is also what makes you compliant:

• Clarity and accuracy — the model rewards them; MiCA requires them ("fair, clear, not misleading").
• Consistency of association — the model needs it; a clean, non-contradictory brand story is also what keeps your risk committee and NCA comfortable.
• Substance over hype — the model quotes substance and ignores fluff; MiCA prohibits the return-baiting hype anyway.

For most industries, GEO and compliance are unrelated. For yours, the GEO-winning content and the MiCA-safe content are the same content. That is a structural gift. Chapter 5 turns it into a plan.

Building GEO Visibility

This is the practical playbook. It maps directly onto the two mechanisms from Chapter 4: build trained-in association and win live retrieval. We'll go pillar by pillar, then cover measurement, then the MiCA guardrails that wrap all of it.

Pillar 1 — Entity and brand-association strategy

The goal: make the model "just know" that your brand belongs in your category. That means teaching the web — repeatedly, consistently, everywhere credible — the exact association you want.

• Define your one-sentence category claim, per ICP. Not "we do compliance software" but "we are the transaction-monitoring platform for MiCA-licensed stablecoin issuers." Write it once, then use exactly that language, consistently, everywhere. Consistency is what survives compression (Chapter 4).
• Fix your entity footprint. Ensure your company is described identically across your site, LinkedIn, Crunchbase, industry directories, review sites, and any knowledge-graph sources. Contradictory descriptions ("we're a KYC tool" here, "we're a monitoring platform" there) average out to a faint, confused echo.
• Own your category language. Decide the precise terms you want to be associated with (e.g. "MiCA Travel Rule compliance," "EMT reserve monitoring") and use them naturally and repeatedly in substantive content. You're training the probability distribution.
• Get named alongside the category, by others. The strongest signal isn't you describing yourself — it's third parties saying "for X, [Brand] is the go-to" (Pillar 3).

Pillar 2 — Content built to be quoted

The goal: win live retrieval by being the clearest, most liftable passage the model can find. Write for the machine's "desk" (Chapter 4).

• Answer the question in the first sentence. Lead with the answer, then support it. Models lift the clean, up-front statement — not the paragraph you buried it in. Reverse the clickbait instinct.
• Structure for liftability. Clear headings that match real buyer questions. Short, self-contained, quotable statements. Definitions, comparisons, and "who it's for" stated explicitly. A model should be able to grab one paragraph and have a complete, accurate answer.
• Build the content around the persona query matrix (Chapter 3). One substantive asset per persona-question per ICP. The MLRO's "MiCA-audit-proof monitoring" question gets a regulation-mapped explainer; the CTO's "secure and integrable" question gets real architecture documentation.
• Publish the boring, high-value stuff competitors won't. Regulation-mapped guides, audit-readiness checklists, Travel Rule implementation walkthroughs, honest comparison content. This is catnip for retrieval and trust — and it's compliant by nature.
• Make technical documentation public and rich. For the CTO persona, deep public docs are among the most-retrieved, most-quoted assets you can own. Gate less; publish more.
• Include proof the model can quote. Specific, verifiable, non-return-baiting proof: "reduced false positives by X% at [named type of firm]," certifications, standards compliance. (Keep it MiCA-safe — see guardrails below.)

Pillar 3 — Third-party presence (the credibility multiplier)

The goal: get the credible corners of the web to associate your brand with your category — because the model trusts consensus more than self-description.

• Earn mentions, not just links. Being named in industry analyses, roundups, expert articles, and reputable directories builds the association that predicts citations (Chapter 4). Prioritise being mentioned in the right context over raw backlink counts.
• Contribute genuine expertise publicly. Bylined analysis on MiCA developments, standards-body participation, expert commentary in trade press. This builds the CEO/board-persona credibility signal and feeds trained-in memory.
• Cultivate review and peer-signal presence. For B2B software, presence on the review platforms buyers (and models) consult matters. Encourage satisfied customers to describe you in your category language.
• Get into the datasets models read. Structured, factual, consistent presence in the sources that feed both training and retrieval — clean company data, accurate directory listings, substantive third-party write-ups.

Pillar 4 — Technical foundation (the floor, not the strategy)

The goal: be retrievable. This is SEO hygiene repurposed — necessary, not sufficient.

• Be crawlable and indexable. If a model's search tool can't fetch your page, you can't win retrieval. Clean site structure, fast load, no accidental blocking.
• Use structured data where it helps. Schema markup that clearly states what you are and who you serve helps machines parse and trust your facts.
• Keep facts machine-readable and consistent. Who it's for, what it does, proof points — in clean, parseable form on every key page.
• Don't over-invest here. This is the floor. Once you're crawlable and parseable, further technical SEO tinkering has sharply diminishing returns compared to Pillars 1–3.

Pillar 5 — Measurement: track what actually moves

The old dial — organic sessions — is going down for reasons that have nothing to do with your performance (Chapter 2). Track the dials that reflect GEO reality:

• Share of voice in AI answers. Run a fixed set of buyer-relevant prompts (your "prompt audit," template in the appendix) across ChatGPT, Perplexity, Gemini, and Google AI Overviews on a regular cadence. Record how often you're named, how you're described, and who's named instead of you.
• Citation frequency. How often your pages are cited/linked as sources in AI answers.
• Accuracy of description. Is the model describing you correctly and in your category language? A wrong description is a GEO bug to fix at the source.
• Competitor presence. Who owns which queries. Where you're absent and they're present is your roadmap.
• Branded query volume. Rising branded search is both a trust signal and, per emerging research, a predictor of AI citations.
• Downstream: GEO-attributed pipeline. Tie it to revenue (Chapter 6), because in this market traffic vanity metrics are worse than useless — they're misleading.

Set a baseline now, before you start. The prompt audit run today is the "before" picture you'll measure everything against.

Pillar 6 — The MiCA guardrails (built in, not bolted on)

Because you're a regulated firm, every asset above must clear the compliance bar. The good news from Chapter 4: doing GEO well and doing MiCA compliance well are largely the same act. The guardrails:

• Fair, clear, and not misleading — always. Every claim, proof point, and comparison must meet the standard. This is also what makes content quotable and trusted, so it's not a tax — it's the strategy.
• Never disguise marketing as research or news. Marketing must be clearly identifiable (Article 7 MiCA). Your genuine thought leadership should be genuinely educational; your promotional material should be clearly promotional. Don't blur them.
• Keep marketing consistent with your white paper (where one applies). Claims in content can't outrun disclosures.
• No return-baiting. Frame value as risk reduction, efficiency, and compliance assurance — never as promised financial returns.
• Own your affiliates' and influencers' words. If third parties promote you, their communications are regulated too. Prefer earned, organic mention over paid affiliation (which ESMA explicitly names as solicitation).
• Route new content types past compliance once, then templatize. Build compliant templates so the program scales without a legal review bottleneck on every asset.

A 6-pillar build sequence (preview of Chapter 9)

1. Baseline — run the prompt audit; document how the machine sees you today.
2. Fix the entity footprint — consistent description everywhere (fast, high-leverage).
3. Publish the persona-mapped cornerstone content — the quotable, compliant assets that win retrieval.
4. Earn third-party presence — mentions, expertise, reviews.
5. Instrument measurement — track share of voice, citations, accuracy.
6. Feed the revenue engine — wire GEO visibility into capture and nurture (Chapters 6–7).

From GEO Traffic to Revenue Engine

The mistake that wastes everything before it

A firm can do everything in Chapters 4 and 5 correctly — become the named brand, win the citations, own the category in the model's memory — and still generate almost no revenue from it. Why? Because they treated GEO as a visibility project and never built the system that converts visibility into pipeline.

This is the most common and most expensive failure in the whole playbook. GEO is the top of the funnel. It is not the funnel. The demand it creates is unusually valuable and unusually fragile, and it behaves differently from any traffic you're used to.

How GEO traffic actually behaves

Understand these four properties, because they dictate the infrastructure you need:

1. Low volume, high intent. GEO won't flood you with clicks — remember, most AI interactions are zero-click (Chapter 2). The visitors who do arrive have often just been told by a machine that you're a leading option. They're not tyre-kickers; they're pre-qualified by the model. Each one is worth many ordinary visitors.
2. Mid-cycle, not top-of-cycle. Unlike an ad click from a cold prospect, a GEO visitor frequently arrives already deep in research — they've framed the problem, seen a shortlist, and are now validating you specifically (Chapter 2, Stage 3). They arrive warmer and further along than traditional traffic. Your site has to meet them there, not treat them like a stranger.
3. Multi-visit and multi-stakeholder. The same account may send several different people to your site over months — the CTO one week, the MLRO the next — each arriving from their own AI research. The engine must recognise and connect these touches, not treat each as a fresh anonymous session.
4. Invisible until it converts. You have no signal a GEO buyer exists until they act. So the moment they do act — request a demo, download the audit guide, start a conversation — is precious and rare. Wasting it is unforgivable.

The capture infrastructure GEO demands

Your website stops being a brochure and becomes the conversion layer of the revenue engine. What it must do:

• Confirm, in one screen, what the model just told them. The GEO visitor arrives with an expectation the machine set ("this is the leading MiCA monitoring platform"). Your landing experience must immediately validate and deepen it — clear category claim, who it's for, proof — not make them re-discover what you do. Meet the warm, mid-cycle intent.
• Offer the next step that matches their stage. A mid-cycle validator doesn't want a generic newsletter signup; they want the audit-readiness guide, the architecture docs, the comparison, the demo. Offer stage-appropriate, persona-appropriate conversion points (mapped to Chapter 3).
• Capture with consent, cleanly. Given MiCA's solicitation rules, build capture around clear, consented opt-in. Consented inbound relationships are your safest and most valuable channel (Chapter 1).
• Identify and stitch the account. Use analytics and enrichment that let you recognise when multiple stakeholders from one account are engaging, so sales sees the account waking up, not disconnected visits.
• Route instantly to a human. The moment a qualified lead converts, it must reach a salesperson in real time — which brings us to the single highest-leverage number in this entire playbook.

The five-minute rule: the most important number in the engine

Here is the statistic every revenue leader in regulated fintech should have tattooed on the inside of their eyelids.

In a landmark study by Dr. James Oldroyd with MIT and InsideSales.com — analysing more than 15,000 leads and 100,000+ contact attempts across multiple B2B companies, published via Harvard Business Review — researchers found that contacting a web lead within 5 minutes, versus waiting 30 minutes, makes you about 21 times more likely to qualify that lead, and roughly 100 times more likely to make contact at all (MIT/InsideSales Lead Response Management study, summarised by Rework; AInora, sourcing the study to MIT/InsideSales).

Let that land. Not 21% better. 21 times more likely to qualify. The odds of even reaching the person fall off a cliff after the first five minutes.

(A note on rigour, because this stat is constantly mangled: the 21x/100x figures come from the 2007 MIT/InsideSales study, not from Harvard Business Review's separate 2011 audit and not from any McKinsey study — those don't exist. HBR's separate research is where the "average first response takes 42 hours" and "23% of companies never respond at all" figures come from. We cite the real source so you can too.)

Why the five-minute rule is life-or-death for GEO leads specifically

For ordinary lead sources, slow follow-up is merely wasteful. For GEO leads, it's catastrophic, for three compounding reasons:

1. GEO leads are rare and expensive to earn. You fought for months to become the named brand. Squandering that lead by responding in two days is throwing away your most costly acquisition.
2. GEO leads are actively comparing you right now. A mid-cycle validator who just converted on your site is, in the same session, likely asking the machine about your competitors too. The vendor who responds in five minutes enters the conversation while you're still top of mind. The one who responds tomorrow enters after the buyer has moved on to whoever answered first — and many buyers purchase from the vendor that responds first (lead-response research compilation).
3. The regulated buyer reads response speed as an operational signal. For a compliance officer choosing a vendor they'll depend on for audit-critical work, "how fast and how professionally did they respond to my first contact" is a live data point about what working with you will be like. Speed isn't just conversion mechanics; it's a trust demonstration.

Building for speed: the operational setup

• Instant routing and alerting. New qualified lead → immediate notification to the right rep (Slack, mobile, CRM) with full context. No batch processing, no overnight queue.
• A five-minute SLA, enforced and measured. Make sub-five-minute first response a tracked, owned metric. What you don't measure, you don't do.
• Automated instant acknowledgement, human fast-follow. An immediate, compliant, human-sounding acknowledgement (email/booking link) the second they convert — buys you presence while a human mobilises. Then a real person within minutes.
• Context in the rep's hands. The rep should see which pages the lead viewed, which persona they map to, and what AI-stage signals exist — so the first human touch is relevant, not generic.
• Frictionless booking. Let a hot, mid-cycle buyer book time instantly rather than waiting for a callback. Every extra step leaks the lead.

Attribution: crediting a channel inside a black box

The hardest operational problem with GEO is that the "search" happens inside a machine you can't instrument. A buyer asks ChatGPT, gets told about you, and types your name into their browser a week later — which shows up in your analytics as "direct" or "branded search," not "GEO." This causes finance teams to under-credit and under-fund the very channel driving pipeline.

How to attribute honestly:

• Track leading indicators, not just last-click. Share of voice in AI answers and citation frequency (Chapter 5) are your leading GEO metrics. Rising AI presence that precedes rising branded/direct traffic and pipeline is the causal story — make it visible to finance.
• Ask buyers directly. Add "How did you first hear about us?" to forms and discovery calls with an explicit AI-assistant option. Self-reported attribution is imperfect but, for AI, often the truest signal you have.
• Watch the correlation. When GEO share of voice climbs and, weeks later, branded search and direct pipeline climb behind it, that lag is the fingerprint of GEO working (Chapter 4's compounding-with-a-lag effect).
• Report GEO on pipeline, not traffic. Anchor the whole program to sourced and influenced pipeline. In this market, traffic metrics mislead; pipeline doesn't.

The Nurture Engine

Why nurture is the whole game in regulated fintech

Most B2B playbooks treat nurture as a drip of newsletters to leads who "aren't ready yet." In regulated fintech, that framing is fatally wrong. Here, nurture is where the deal is actually won, because of three hard facts established earlier in this playbook:

1. The cycle is 6–18 months (Chapter 2). A lead you don't nurture is a lead a competitor nurtures. The vendor who stays present, useful, and trusted for the entire cycle wins; the one who pitches once and goes quiet loses by attrition.
2. The buyer is a committee, and consensus — not a yes — is the bottleneck (Chapter 3). Deals die in consensus-building. Nurture's primary job is to build that consensus by equipping your champion and reaching every stakeholder.
3. The buyer keeps researching you the whole time, mostly via AI (Chapter 2). Nurture keeps feeding them the material that keeps the model's answer — and their internal narrative — favourable to you.

If GEO is how you get found, nurture is how you get chosen. In this market, the second is harder and more decisive than the first.

The mindset shift: from "drip campaign" to "consensus engine"

Stop thinking of nurture as a sequence of emails that stay in touch until the buyer is ready. Start thinking of it as a multi-threaded system that moves an entire buying committee toward internal agreement over months. Its outputs are not opens and clicks. Its output is a champion who can win the argument inside their own organisation.

That reframing changes everything about what you send, to whom, and why.

Principle 1 — Multi-stakeholder, not single-thread

A single nurture track aimed at one contact is built for a buyer that doesn't exist here. You need parallel, coordinated threads for each persona in the committee (Chapter 3):

• The MLRO gets regulatory depth: audit-readiness, MiCA interpretation, coverage proof.
• The CTO gets technical substance: architecture, security, integration, docs.
• The CEO gets strategic and credibility signals: category leadership, durability, vision.
• The CFO gets economic framing: cost of inaction, risk reduction, TCO.
• The champion gets ammunition to sell internally — the single most valuable nurture asset you can produce.

These threads must be coordinated, not contradictory. The story each stakeholder receives should reinforce the others, so that when they compare notes internally, the picture is consistent — which is also what keeps the model's story about you consistent (Chapter 4).

Principle 2 — Arm the champion to sell internally

This is the highest-leverage move in the entire nurture engine, and most vendors miss it. Your champion is trying to convince their MLRO, CTO, CFO, and board to say yes. They are selling for you in rooms you'll never enter. So give them the sales kit:

• A crisp internal business case they can forward.
• Persona-specific one-pagers that answer each colleague's objection (the risk committee's "any red flags," the CTO's "is it secure," the CFO's "what's the ROI").
• Proof and validation they can cite without exposing themselves to risk.
• Answers to the exact questions their colleagues are asking the machine.

When you make your champion look smart and safe for backing you, you convert one supporter into an internal salesforce.

Principle 3 — Content-led and MiCA-compliant by design

Nurture content lives inside the same guardrails as everything else (Chapter 1). But nurture has a structural advantage: it goes to consented contacts in a direct relationship, which is the safest footing under MiCA's solicitation rules. Even so:

• Keep every asset fair, clear, and not misleading.
• Lead with genuine education — how to pass an audit, how MiCA obligations map to controls, how to implement the Travel Rule — not product pitches. Education builds trust and stays clearly on the right side of the "don't disguise marketing as research" line by simply being genuinely useful and clearly attributed.
• Frame value as risk reduction and operational assurance, never promised returns.

The content that nurtures best in this market is the content that helps a compliance professional do their job and keep their institution safe. That content is compliant, trust-building, and — usefully — the same material that wins GEO retrieval (Chapter 5). Build once, deploy across GEO and nurture.

Principle 4 — Stay the named brand across the whole cycle

Chapter 4 showed that the model weighs what's on its "desk" and what it consistently "knows." The buyer's internal narrative works the same way. Over 6–18 months, the vendor who stays consistently present — with useful, relevant, well-timed material — becomes the default, the safe choice, the name that keeps coming up. The vendor who goes quiet fades from both the buyer's mind and, over time, the model's live-retrieved context.

Nurture is how you stay resident in both. Cadence matters less than relevance and consistency: show up when you have something genuinely useful to the stage the buyer is in, reinforce the same clear category association every time, and never go dark during the long middle of the cycle where deals are quietly lost.

Principle 5 — Re-engagement and the long game

Not every lead moves on your timeline. A trigger event — a new regulation, a failed audit, a leadership change, a licence milestone — can reactivate a dormant lead overnight. Your job is to be the vendor already present and trusted when that trigger fires.

• Keep dormant-but-consented leads on a low-frequency, high-value educational track.
• Watch for trigger events (regulatory deadlines, enforcement news, the account's own milestones) and re-engage with genuinely relevant timing.
• Treat re-engagement as relationship maintenance, not a resurrection pitch.

The sales–marketing handoff choreography

Nurture fails when marketing and sales aren't choreographed. In this engine:

• Marketing owns the multi-stakeholder content threads, the educational nurture, and the account-level presence.
• Sales owns the human relationship, the champion coaching, and the consensus-building conversations.
• The handoff is continuous, not a one-time toss. Because GEO leads arrive mid-cycle and multi-stakeholder (Chapter 6), sales and marketing must share account-level signal in real time — who's engaging, from which persona, at which stage — so the human and automated threads reinforce rather than collide.
• Speed still applies (Chapter 6): the five-minute rule governs the first human touch; nurture governs the long middle; sales governs the close.

A simple nurture architecture (template in appendix)

Selling in a Regulated, Slow-Moving Market

The buyer's operating system: risk avoidance

Everything about how a MiCA-regulated buyer behaves flows from one fact: their job is to avoid risk, and choosing a vendor is a risk. Moving slowly is safe. Moving slowly avoids blame. A wrong vendor choice can mean a failed audit, an enforcement action, and a career consequence with the buyer's name attached. "Move fast and break things" is a startup mantra; among people responsible for AML, sanctions, and MiCA compliance, it is a threat.

This is not a personality flaw to overcome. It is the rational operating system of the market. Sell against it and you lose. Sell with it — by making the safe choice obviously you — and you win.

Selling is de-risking, not persuading

The instinct of most salespeople is to build desire: bigger vision, better features, more urgency. In this market, desire is not the constraint — perceived risk is. The buyer often already wants to solve the problem. What stops them is fear of choosing wrong. So the entire selling motion should be organised around systematically removing that fear:

• Proof over promises. Every claim should be evidenced, specific, and verifiable — which is also the MiCA standard (Chapter 1) and the GEO standard (Chapter 4). "Fair, clear, not misleading" isn't just compliance; it's the most persuasive posture available with a risk-averse buyer.
• References and social proof from peers. A compliance officer trusts another compliance officer far more than any pitch. Peer validation is the strongest de-risker you have.
• Transparency about limitations. Admitting what you don't do builds more trust than claiming you do everything. Confused, over-claiming vendors read as risky.
• Make the reversible easy, the irreversible safe. Pilots, phased rollouts, clear exit terms, strong data-processing and security guarantees — anything that lowers the stakes of the first yes.

Discovery: diagnose like a regulator, not a rep

Discovery in regulated fintech is closer to a regulatory assessment than a sales qualification. Do it in their language:

• Anchor to obligations, not features. Start from the specific MiCA/regulatory obligations they must meet and work back to how you help them meet them. This mirrors how the MLRO already thinks (Chapter 3).
• Map the full committee early. Identify every stakeholder and their distinct fear (Chapter 3) in discovery, so you can arm the champion for each one (Chapter 7).
• Surface the cost of inaction. For a risk buyer, the sharpest motivator isn't upside — it's the quantified risk of not acting: audit failure, enforcement exposure, manual overhead, reputational damage.
• Find the trigger and the timeline. Regulated deals move on external triggers (deadlines, licence milestones, audit findings). Locate the trigger; it's your only real source of urgency, and it's legitimate rather than manufactured.

Compressing time without triggering resistance

You can't rush the buyer, but you can remove the self-inflicted delays that make regulated cycles longer than they need to be:

• Answer objections before they're raised. Because you know each persona's fears, pre-empt them with ready material. Every objection resolved in advance is weeks saved in consensus-building.
• Remove friction from every stage. Fast responses (Chapter 6), instant access to documentation, pre-built security and DPA packs, easy references. Slowness is often the vendor's fault, not the buyer's caution.
• Enable parallel, not sequential, stakeholder engagement. Help the champion get the CTO, MLRO, and CFO evaluating simultaneously rather than one after another. This is where the multi-stakeholder nurture engine (Chapter 7) directly compresses cycle time.
• Never manufacture false urgency. Fake deadlines and discount pressure read as manipulation to this buyer and increase perceived risk. Legitimate urgency (their regulatory deadline) works; artificial urgency backfires.

Handling the "we'll wait" default

The regulated buyer's default answer is not "no." It's "not yet." Their safest action is always inaction. To move a stalled deal:

• Re-anchor to their trigger and the cost of delay (audit date, enforcement climate, competitor exposure).
• Shrink the first commitment until saying yes feels lower-risk than continuing to wait.
• Keep the champion equipped to fight the internal "let's revisit next quarter" drift with fresh, relevant material (Chapter 7).
• Stay present — because when their trigger finally fires, you must be the vendor already trusted and top of mind, in their head and in the machine's answer.

The 90-Day Implementation Roadmap

The sequencing logic

GEO compounds with a lag (Chapter 4), so the earlier you start, the sooner it pays off — but only if the revenue engine is ready to catch what it produces (Chapter 6). The 90-day plan therefore builds both the visibility and the capture layers in parallel, front-loading the fast, high-leverage fixes and the measurement baseline so you're never flying blind.

The KPI dashboard (what to actually track)

The failure modes that kill GEO programs

Learn these so you don't join them:

1. Treating GEO as a visibility project with no revenue engine (Chapter 6). The most common and most expensive failure. Visibility without capture and nurture is admired and forgotten.
2. Measuring traffic instead of AI share of voice and pipeline. You'll watch the wrong dial fall and conclude, wrongly, that it isn't working (Chapter 2).
3. Impatience. GEO compounds with a lag (Chapter 4). Firms that quit at week six abandon the asset right before it pays.
4. Generic messaging. One-size-fits-all content wins no persona's query (Chapter 3). Specificity is the whole game.
5. Slow follow-up. Winning the AI answer and then responding to the lead in two days wastes everything upstream (Chapter 6).
6. Fighting MiCA instead of using it. Trying to sneak paid/disguised tactics past the regulation is high-risk and misses the point: the compliant path is the winning path (Chapters 1, 4).

The one-page "so what"

The market changed underneath regulated fintech in two ways at once, and they compound.

Put together, these facts point to one strategy, and this playbook is that strategy:

1. Win GEO — become the brand the model consistently associates with your category (trained-in memory) and the page it quotes when it searches live (retrieval). Neither is "rank #1 on Google." Both reward clarity, consistency, and substance — which is also what MiCA demands. For you, the compliant content and the GEO-winning content are the same content.
2. Build the revenue engine — capture the rare, high-intent, mid-cycle GEO visitor, confirm what the machine promised, and put a human in front of them within five minutes (21x more likely to qualify, per the MIT/InsideSales study). Visibility without this is a leaky bucket.
3. Run the nurture engine — the decisive element in a long, committee-driven, cautious sale. Multi-stakeholder, consensus-building, compliant nurture keeps you the named brand and arms your champion to win internally. This is where deals are actually won.
4. Sell by de-risking — make the safe choice obviously you. Proof, peer validation, transparency, removed friction. Never manufacture urgency; use the buyer's real regulatory triggers.

This playbook draws on the mechanics of AI retrieval detailed in "GEO is not SEO: how LLMs really find information", on MiCA and ESMA primary guidance, and on the MIT/InsideSales lead-response research. Where data is young or contested, we've said so. GEO, and the revenue engine around it, is run best as a dedicated discipline — not a checkbox on an SEO retainer or a task delegated between other priorities.

Appendices

Appendix A — MiCA Marketing Do / Don't Checklist

Sources: Article 7 MiCA; ESMA reverse-solicitation guidance; Sedric, MiCA marketing standards. This checklist is a practical summary, not legal advice — confirm specifics with your compliance function and NCA.

Appendix B — GEO Prompt-Audit Template

Run this set on a fixed cadence (e.g. monthly) across ChatGPT, Perplexity, Gemini, and Google AI Overviews. Record, for each: were you named? how were you described? who was named instead?

Category / landscape prompts

• "What are the leading [category] platforms for MiCA-licensed [ICP]?"
• "Who are the top providers of [specific capability] for EU crypto firms?"
• "Best [category] tools for a [ICP] in 2025?"

Comparison prompts

• "How does [Your Brand] compare to [Competitor]?"
• "[Your Brand] vs [Competitor] for [use case]?"

Persona prompts (one set per persona from Chapter 3)

• MLRO: "Which [category] tools satisfy MiCA audit requirements for [ICP]?"
• CTO: "Most secure and integrable [category] platform for regulated crypto data?"
• CFO: "Typical cost / ROI of [category] tools for a CASP?"
• CEO/Board: "Is [Your Brand] a reputable, financially stable vendor? Any red flags?"

Accuracy prompts

• "What does [Your Brand] do?" (Is the description correct and in your category language?)
• "Who is [Your Brand] for?"

Scoring: Named (Y/N) · Described accurately (Y/N) · Competitors named · Notes. Track share of voice = (times named ÷ total prompts) over time vs. competitors.

Appendix C — Persona Query Matrix Template

Build one per ICP.

Appendix D — Nurture Sequence Template

Appendix E — Glossary

Prepared as a shareable industry resource. Statistics cited to primary sources throughout; figures on AI-search behaviour reflect 2025 data and are evolving. Not legal advice.