AI search optimization (AEO + GEO) for crypto and Web3 brands

Google SEO targets the 10 blue links on a SERP. AI search optimization targets the source paragraphs that AI tools quote inside an answer. The two surfaces share a lot of infrastructure — schema, internal linking, named-expert bylines — but the optimization targets differ enough that the work splits cleanly.

AEO — Answer Engine Optimization — restructures existing pages so AI extraction layers can pull self-contained answer blocks: H1 disambiguator, Quick Facts table near the top, H2 questions with first-sentence direct answers under 30 words. That work changes the same content surface Google reads, but the tuning is for the AI extractor, not the blue-link ranker.

GEO — Generative Engine Optimization — produces new content built to be cited. Specific numbers, dated claims, named experts, primary-source links. The piece is structured around what an AI tool would lift verbatim, not around long-tail keyword volume.

Running both inside one retainer is the honest path because the operations they share — schema engineering, sameAs verification, llms.txt, AI-crawler robots, the prompt monitor — would otherwise be billed twice. Some clients combine this retainer with our Crypto SEO line for $5,200/mo instead of the additive $6,000/mo. The discovery call resolves which combination fits.

The prompt list is built during discovery — 22 commercial and consideration-stage queries selected from your ICP, current sales conversations and our crypto-specific prompt corpus. Each prompt runs weekly across ChatGPT (gpt-4o, gpt-4.5), Perplexity, Gemini, Claude and Google AI Overviews. We capture the full answer, the sources cited, the position of each source in the answer, and the brand mentions in unbranded category queries.

The 22-count was tuned across our engagements. The lower bound (15) is where category drift starts slipping through — when a prompt list is too narrow, the analyst optimizes for a moving target. The upper bound (35) is where hand-review breaks down — the analyst starts marking citations without reading the surrounding answer, which is where prohibited regulatory phrasings show up. 22 is the operating sweet spot.

Output is the monthly citation report: counts per platform, prompt-by-prompt share-of-voice against three named competitors, citation-quality scoring (favorable, neutral, comparative-against-a-competitor), and the trend deltas week-over-week. Competitors in this category tend to lead with a single number ("+5,856 AI Overview citations" is one we have seen). We lead with the monitor methodology because counts without method are gameable, and any agency that has worked in AI search for more than a quarter knows it.

Across our 8 active crypto, fintech and Web3 engagements 2024–2025, the median time to first top-3 commercial ranking is 4.7 months. That number is conservative — the fastest case (Fast Offshore Licenses on "anjouan gambling license") landed top-3 in month 4; the slowest (a MiCA-bound exchange that lost two months to a Google certification appeal) landed in month 7. The point of citing the median rather than the fastest case is the median is what a new client should plan against.

The two NDA engagements with the strongest AI-search outcomes:

A Series-A real-world-asset tokenization platform — 8 months, 44 confirmed AI citations across Perplexity, Phind and ChatGPT, inbound demo requests +108%, top-3 on 7 technical queries. The number we lead with is **44 confirmed** because confirmed is the variable that distinguishes a measured citation from a competitor headline.

An EU MiCA-licensed crypto exchange — 11 months, +340% non-brand organic, 100% compliance-approval rate on every piece of public copy shipped under the engagement. The compliance-approval rate is the metric MiCA-bound buyers actually need; citation count without a compliance gate is a liability, not a result.

Two public cases also speak to AI-search outcomes: [CryptoLawIndex](/cases/cryptolawindex/) hit #2 on "crypto law firm rankings" with 31 AI-citation appearances across 4 platforms during a 9-month cold launch. [Fast Offshore Licenses](/cases/fast-offshore-licenses/) ran 14 months with 138 donor placements and an organic share lift from 18% to 62%. The donor allowlist behind those placements currently holds 312 publications, with a 12% drop-off per quarter as publishers tighten robots.txt against AI crawlers.

Three properties carry most of the weight in YMYL verticals. sameAs — at least two independent profiles per named expert (LinkedIn plus a conference talk, regulator submission, or authored publication). knowsAbout — specific topical fields the expert is associated with. jobTitle and worksFor — establishing the institutional anchor.

The named-byline effect is one we ran internally as an A/B across a single client's content cohort: pieces shipped under a named expert with full schema.org Person markup outranked equivalent pieces under "Editorial Team" bylines, and retained that ranking through one algorithmic event during the test window. Anonymous-byline content lost more rank than named-byline content during the same event. The exact deltas live inside our internal report rather than the marketing page because we do not have permission from the client to publish them.

FAQPage schema with speakable extension on cost, timeline and qualification questions earns rich-result eligibility within 30–60 days when the rest of the markup graph is clean. ItemList with Person references is the correct pattern for ranking pages (CryptoLawIndex-style) where naive Organization-per-firm markup creates entity confusion.

llms.txt is designed during discovery — intent-led, not boilerplate — and quarterly we verify the file is still served at 200 OK and that AI crawlers are actually fetching it (GPTBot user-agent log analysis on the access log). The verification step is the part most agencies skip; without it, publisher robots.txt changes silently strip the AI-fetch coverage.

MiCA Article 66 and the UK FCA financial-promotion regime both treat the original publication as the responsible party when AI quotes your source paragraph in an answer. When ChatGPT quotes a sentence from your site that says "guaranteed APY" or "regulator-endorsed", that sentence is appearing in an AI answer with attribution back to you. The regulator does not care that the AI reformatted it.

Our regulatory review workflow runs before publication for every YMYL piece. The compliance lead — your in-house compliance lead, your retained counsel, or our contracted reviewer — passes 2–4 days per piece flagging the four banned-claim categories (guaranteed returns, risk-free framing, implied regulator endorsement without specific licence reference, universal-suitability suggestions). For MiCA-bound clients we maintain a per-jurisdiction copy-rules document so the German variant satisfies BaFin and the French variant satisfies AMF.

The output of that workflow is what produced the 100% compliance approval rate on the MiCA exchange engagement referenced above. Generalist AEO agencies do not run this step because the AEO category has not yet had to pay the regulatory tax that the SEO and content categories already paid. We expect that to change inside 12 months as MiCA enforcement compounds.