Why does AI call our licensed exchange or casino “unverified”?

Because the licence exists in a form machines cannot read, while the negative sources about you are perfectly readable text. We have audited brands holding real, verifiable licences whose sites served 403 to every AI crawler — so the models answered using review aggregators and complaint threads instead. The fix is not a better licence page for humans; it is machine-readable trust: the registration in schema, a trust hub a crawler can fetch, and the crawlers unblocked.

Our token or protocol has huge search volume. why are we invisible in AI answers?

Volume and AI presence are different channels. Across our published dataset 93% of brands were known to AI and described accurately, while only 4.5% were named when a buyer asked who to pick. In crypto the split is wider than average, because the category answers are assembled from lists and comparisons that most protocols never appear on.

We are anonymous by design. does GEO even work for us?

Partly, and you should know which part. Anonymity costs you the entity and expertise layers: no named humans, no verifiable organisation record, so a model has less reason to quote you over a competitor who has both. What still works is readability, structured data about the product rather than the people, and third-party sources — audits, listings, documentation others cite. We will tell you what the ceiling looks like before you pay for anything.

Do you work with competitors in the same niche?

Not without a firewall, and we say so out loud. When two clients overlap, we agree in writing what is not touched for either — no shared target lists, no cross-promotion, no reusing one client’s positioning against the other. If the overlap is head-on we decline the second engagement.

What does the crypto bench actually consist of?

48 audits inside web3 and crypto and 960 companies mapped in that vertical, across protocols, DeFi, exchanges, infrastructure, compliance, security, trading and education. Two web3 development clients and one crypto-law client are on retainer; their names stay under NDA until they say otherwise, and the numbers we quote from their programs are marked as single measured runs, not averages.

Crypto SEO agency - GEO for web3, measured across six engines

009 Agency is a crypto SEO and web3 marketing agency built around Generative Engine Optimization: we make protocols, exchanges, DeFi products, infrastructure and crypto-law firms visible, cited and chosen inside AI answers across ChatGPT, Claude, Gemini, Perplexity, Grok and DeepSeek — and we re-measure it every two weeks against a logged baseline.

Web3 is the deepest bench in our dataset: 48 audits and 960 companies mapped in this vertical alone. That matters less as a credential than as a practical advantage — in most crypto categories we already know which competitors the engines name, and which sources they name them from, before your audit even starts.

why crypto loses in AI answers specifically

The general failure pattern holds everywhere: known but not recommended. In our published dataset 93% of brands were described accurately by AI and only 4.5% were named when a buyer asked who to pick. Crypto adds three problems on top of it.

trust is the category’s bottleneck, and it is unreadable

Buyers ask AI whether you are safe before they ask whether you are good. We have audited licensed operators with genuine regulatory standing whose sites returned 403 to every AI crawler — the licence was real, none of it was reachable, and the models answered with review aggregators and complaint threads. In the same audits engines volunteered warnings like “verify before depositing” about brands that had done nothing wrong except leave the trust layer machine-unreadable.

the average is low, so the shelf is cheap

The average GEO health score across web3 and crypto in our report was 37 out of 100 — below even the market-wide 35 in places, and nowhere near the 60 nobody in the sample cleared. Categories are decided by a handful of lists and comparisons, and most of those lists were published by people who are not you.

JS-heavy stacks delete you from the corpus

Crypto sites lean on app-like frontends. Half of all audited sites were unreadable to AI crawlers to some degree and a third were a full blackout — the engine receives a container and no content. One site scored 100/100 for SEO in Lighthouse while GPTBot got 1,133 bytes of empty shell from the same URL. If that is your stack, everything above it is wasted spend.

where we work inside web3

the 48-audit bench, by segment
segment	what the audits keep showing
protocols & DeFi	Documentation is citable, marketing pages are not. Engines quote the docs and name someone else as the product.
exchanges & trading	Trust queries dominate. Without machine-readable licence and entity data the answer defaults to aggregators.
infrastructure & dev shops	Category answers are built from roundups and directories. Being absent from four lists costs more than any on-page fix.
crypto compliance & law	The strongest citation upside we have measured: expertise content plus named authors moves recall fastest.
security & audits	Technical credibility exists in PDFs. Nothing a model can parse.
education & media	Volume without entity: high traffic, no organisation record, rarely named as a source.

what the first 90 days look like

The order is the same as everywhere, because it has to be: readable, then verifiable, then quotable.

Audit. Your buyers’ real prompts run live through six engines and logged verbatim, plus crawler checks with actual AI user-agents. You get a health score, the verdicts engines currently give about you, and the three most expensive gaps. Complimentary. what is inside →
Unblock and mark up. Crawler access, rendering, llms.txt, the JSON-LD entity graph, machine-readable trust. Shipped as files your developer applies. technical GEO →
Publish what gets quoted. Comparisons, definitions, security and licence pages written to be liftable in one paragraph.
Get onto the lists. The roundups, directories and communities the engines actually cite in your category — identified from your own panel, not from a generic outreach list.
Re-measure. Same panel, every review, reported as a delta. Series, not screenshots.

proof, and its limits

One measured result we can state: a crypto-law client went from 0/5 to 4/5 engines naming the firm on its category prompt by day 17 of the program (July 2026). That is one measured run on one brand, not an average and not a promise — we re-test every review precisely because single runs are volatile.

Client names stay under NDA until clients decide otherwise, so nothing on this page is a named case study. Ask on the call and we show the raw logs, including the answers where we did not move the needle.

one thing we will not do

We will not sell you bought placements as authority. In one audit the only readable corpus about a brand was its own paid promotion, and the engines flagged the sources as promotional rather than independent, then issued a high-risk verdict. In crypto that failure mode is common and expensive: volume without independent confirmation makes the answer worse, not better.

crypto SEO agency: GEO for web3