009 Agency - a generative engine optimization (GEO) agency
235+ manual audits3,900+ companies mapped93,650+ answers analysed
backed by our own research and data: a published dataset of 100 audits across startups, scaleups and enterprise, and 135 more since
protocols, trading, tokenization, prediction markets
1 public case: Binaryx
B2B platforms, operator brands and their martech
1 public case: NDA casino
development agencies, consulting, legal, accounting and tax, recruiting, marketing agencies, corporate services
payment rails, card issuing, payroll and the compliance around them
AI labs and applied-AI studios: agents, infrastructure, AI for fintech
software products sold to teams: the most buyer questions, the fewest measured answers
brands that sell direct, where the assistant is already the shortlist
brokerages, developers, manufacturers and the studios that visualize for them
last snapshot: sep 5, 2026
Manimama: from absolutely invisible to the #2 answer in Europe
1,004 published articles the engines had read and never once recommended. twenty buyer prompts, four engines, every run dated.
0/20 → 20/20buyer prompts naming the firm
#2 in EuropePerplexity, citing the firm’s own page
six engines called it a scam. now all six recommend it
a licensed casino every engine warned German players away from, because its own site never said who ran it.
6 → 0engines calling it a scam
24 → 47health score, two re-audits
INC4: from a mistaken identity to the source the engines quote
a Kyiv and Lisbon AI studio whose brand name resolved to a conference before the company, while the engines quoted pages that had stopped existing.
27 → 53health score in seven weeks
20/20 + 8/52first set, then the fintech panel, #1 in 4
Binaryx: ×238 more visits from AI assistants
a tokenized real estate platform whose fresh migration left hundreds of broken URLs and a blog Bing had never indexed, named in 3% of AI answers about its category.
3 → 713visits a month from AI assistants
13th → 4thshare of voice of 14 rivals, 50 prompts
009 × your brand · GEO program · last snapshot: review 4 demo data, illustrative
frozen prompts · the same questions every review click a cell to read the engine’s words
| prompt | chatgpt | perplexity | gemini | claude | grok | deepseek |
|---|---|---|---|---|---|---|
| “Your Brand vs two rivals: who should I hire?” branded | ||||||
| “top companies in your category” | – | – | – | |||
| “recommended partners for your service” | – | – | – | – | – | – |
and 13 more prompts to track
all you need to do is copy-paste. an organic traffic source has never been this easy to launch.
dashboard / sprint 4 of 6 demo data, illustrative
how checkboxes work:To doDone, awaiting verificationVerified by 009
all tasks are copy-paste. 009 writes the file, the text and the exact place to put it. you paste and tick.
tech · 3 open
Context: the file tells LLM crawlers what the company is and which pages matter. without it the engines guess from third-party profiles.
Action: copy the file below, serve it at /llms.txt as text/plain, tick this task.
# Your Brand > Cross-border payment infrastructure for marketplaces and SaaS platforms. Licensed EMI, London & Vilnius. Since 2017. ## Products - [Payouts API](https://yourbrand.com/payouts-api) - [Multi-currency accounts](https://yourbrand.com/accounts) ## Company - [Licences and compliance](https://yourbrand.com/compliance) - [Case studies](https://yourbrand.com/customers)
Unlocks: a canonical fact-card for the whole domain, read by every engine.
Context: 23 of your URLs resolve through two to four chained redirects. AI crawlers give up after the first hop.
Action: paste the map into the nginx config or the Cloudflare rules panel, both versions are in the task pack. deploy, tick.
polymarket
aave
kalshi
hacken
stobox
legion
wallchain
core3
altcoinomy
adi foundation
cryptology
flipper
cedra
claimr
pagga
bitmaker
limitless
voice
1xbet
parimatch
pin-up
ggbet
stake
sprut gaming
bukawin
boomzino
ideasoft
4ire labs
redduck
evacodes
boosty labs
solus
karpinska
manimama
amlbot
bsa global
dream view
kaizen
input global
bswan
inc4
grit leaders
stablerail
c-payroll
leadingcards
kolo
cryptoprocessing
moonpay
zero hash
coinspaid
brite
alaan
redotpay
transfi
gnosis pay
trustee
kast
moduledge
elevenlabs
patsnap
wiz.ai
qeen.ai
retell ai
hedra
photoroom
pipedrive
softswiss
vanta
oneleet
documenso
calendly
reply.io
tuum
eterna sports
secretlab
christopher ward
velasca
jolie
nnormal
lyka
koala
lbi properties
grandiora
layan
omega render
cylind
visengine
importa# redirect map (excerpt, full file attached)
location = /products/payouts { return 301 /payouts-api; }
location = /solutions/marketplaces { return 301 /marketplaces; }
location = /blog/tag/compliance { return 301 /blog; }
# 36 more rulesUnlocks: every legacy backlink passes authority to the right page in one hop.
Context: the engines read structured data before prose. your product pages carry none.
Action: one JSON-LD block per page is written. paste each into the page head, run the rich results test, tick.
{
"@context": "https://schema.org",
"@type": "Service",
"name": "Cross-Border Payouts API",
"provider": { "@type": "Organization", "name": "Your Brand", "url": "https://yourbrand.com", "foundingDate": "2017" },
"areaServed": ["GB", "EU"]
}Unlocks: the engines stop guessing what each page sells.
content · 2 open
Context: on the last review Perplexity rebuilt the top-5 payout providers for marketplaces from two third-party lists. your own list on your own domain is the strongest card in that deck.
Action: the article is ready on the articles page. fact-check it, keep the locked phrases exactly as written, paste into the CMS, publish, request indexing the same day.
Top 8 Cross-Border Payout Providers for Marketplaces in 2026 We compared eight providers on licences held, settlement speed, FX markup and published pricing. 1. Your Brand: FCA-authorised EMI, 38 currencies, same-day settlement in 24 markets, since 2017 [locked] ...
Hand-off: tick this task with the live URL. we re-measure the two prompts it feeds at the next review.
dialog with 009 regarding this task
Context: ten question and answer pairs written from your real sales-call questions, FAQPage schema included.
Action: read once, edit anything except the locked lines, paste into /faq, tick.
Q: How long does onboarding take for a marketplace? A: Five to ten business days including KYB. Your Brand holds an EMI licence in the UK and one in the EEA, with pricing published per corridor. [locked]
Unlocks: the exact questions buyers ask the engines get an answer from your own domain.
growth · 2 open
Context: your profile still says 12 currencies and the 2023 fee table. on the last review Claude quoted exactly those numbers to a buyer.
Action: paste the canonical description below into the profile, the same text goes to LinkedIn. tick when saved.
Your Brand is a cross-border payment infrastructure company founded in 2017, headquartered in London with an office in Vilnius. Licensed as an EMI in the UK and the EEA, it runs payouts in 38 currencies for marketplaces, SaaS platforms and gig-economy apps, with pricing published per corridor.
Unlocks: the #1 source the engines cite for vendor facts stops feeding them old data.
Context: the engines build vendor answers from third-party top-X lists. the six letters target the exact lists they cited in our measurements.
Action: letters are written, facts verified, article URL inserted. send from your own domain, two or three per week, tell us which went out.
Subject: one addition to your 2026 marketplace payout providers list Hi Anna, your list is the one buyers keep quoting. one provider is missing: Your Brand, FCA-authorised EMI, 38 currencies, same-day settlement in 24 markets, public pricing. fact-pack attached, happy to answer anything within a day.
Unlocks: every acceptance is a permanent retrieval source.
AI knows the brand by name and describes it accurately.
…and names it when a buyer asks who to pick.
six engines are already answering your buyers’ questions. see what they say about you.
a buyer asks. the model answers. your name is not in the answer.
right now what a model fetches from your site comes back empty or malformed.
no named people, no facts that agree with each other, nothing to match you against.
models quote comparisons, breakdowns and prices: the pages you do not have.
the answer is assembled from directories, roundups and communities you are not in.
otherwise nobody knows whether anything moved, and the lift stays a claim.
product reality
search reality
12 buyer prompts × 2 engines = 24 live answers · your market, in its language · every prompt sits on live demand (70 to 480 searches a month)
| # | buyer prompt, verbatim | type | chatgpt | perplexity |
|---|---|---|---|---|
| 01 | which payout provider would you recommend for a marketplace in our market? | recall | not named | not named |
| 02 | we need same-day payouts in 30 currencies. which platforms can do it? | recall | not named | not named |
| 03 | what does a cross-border payout cost per corridor? | citation | 2 sources | cited |
| 04 | how long does marketplace onboarding take with an EMI? | citation | 0 sources | 0 sources |
| and 8 more prompts, the same shape | ||||
the recall answers send buyers to directories and comparison lists more than to any provider. those slots are exactly what the planned pages and listings take. named or cited: 1 / 24
| domain | visits / mo | ranked kw | how they win | named by AI |
|---|---|---|---|---|
| your brand | 16 | 6 | target · 0 top-3 keywords | cited 1 of 24 |
| competitor a | 61,875 | 921 | the category giant · 93 top-3 | not named |
| competitor b | 8,750 | 803 | 42 top-3 keywords | not named |
| competitor c | 755 | 150 | a page for each buyer question | 2 of 24 answers |
| competitor d | 580 | 199 | #3 for the pricing query | 1 of 24 answers |
competitor c is named twice on 755 visits a month; competitor a never, on 61,875. placement, not traffic, decides who gets named.
critical01 · entity layer
two addresses across the knowledge panel and the privacy page; Organization schema with no sameAs, no registration number; one engine resolves your name to another company.
critical02 · proof and trust
one case study with a quote and no figure; no number on any product page; no price or corridor table; team without bios; empty legal pages left indexable.
high03 · content architecture
absent for the three buyer queries with the most demand; the only #1 is a 2025 blog post; the pricing query lands on a pitch, not a table.
plus the second-tier list: 27 of 41 URLs without a meta description, duplicate H1s, llms.txt missing, security headers.
/payouts-api/ · 320 / mo
what the API does, corridors, settlement times, indicative price bands and an FAQ. answers the two pricing prompts in our panel.
/marketplace-payouts/ · 170 / mo
the engines answer this query with comparison lists and a competitor at #3. you have one heading on a generic page. one page: options, rate context, a case.
/cross-border-fees-guide/ · 480 / mo
the citation prompts want a guide with a table and a definition block. today the topic is a pitch without a template or FAQ.
the keyword map behind it: 6 hubs, 48 queries, volumes and difficulty per query, the first three pages to ship marked.
goal
target by review 3: named or cited 1 / 24 → 6 / 24 health score 55 → 70
the sprint-1 fix specs ship with the audit, yours whether or not we work together. each one is ready to paste, with the exact place to put it.
no call needed to use them. if we do work together, sprint 1 starts from these files on day one.
benchmarked against our own dataset, not a borrowed industry average. seven chapters shown here, the full audit runs to about 30 slides.
how the audit works, section by section and what we fix first: technical GEO
Real experts in GEO
“We worked with 009 on our AI search visibility, and this cooperation made a great impact on how our company appears in Claude, ChatGPT, and Gemini. We worked for 3 months and felt the results of their work at 3rd month.”
Sofia PidturkinaChief Marketing Officer
AI chat visibility for a performance marketing agency
“For every prompt we cared about, we are now the first or one of the first names the assistants give.”
Kevin MartinezChief Marketing Officer
Marketing strategy for a real-estate platform
“The audit at the start was very detailed, and they came up with a lot of useful, interesting tasks from it. We got our first leads from ChatGPT, and for us that’s a completely new traffic channel.”
Chief Operating Officer
Strong expertise in GEO and AI visibility
“We had a great experience working with 009 Agency. The team helped us understand how Manimama appears in AI search and what we could improve to get better visibility. Everything was clear and practical, without overcomplicating things. Communication was easy, and we always understood what was being done and why. Really enjoyed working with the team and would definitely recommend them.”
Vladislav IvanovChief Marketing Officer
complimentary
a ten-specialist audit protocol: AI-visibility scorecard across six engines, entity and schema review, competitor benchmarks on one scale, and a dated 90-day plan. you keep the scorecard and the plan whether or not we work together.
from $3,000/mo
bi-weekly sprints on the full loop: strategy, technical GEO shipped as ready hand-offs, citable content with QA before you ever see a draft, authority building. every review re-measures your AI visibility against the baseline: you see the scoreboard, not a slide deck. The final price is set after the audit.
what moves the price, what we never charge for, and when not to hire us: the full pricing page
GEO (Generative Engine Optimization) is optimization for AI engines: ChatGPT, Claude, Gemini, Perplexity, Grok, DeepSeek. SEO fights for Google rankings; GEO fights for your name inside the AI's answer to "who should I pick?". Across the 100 audits in our published dataset (July 2026), the two show no correlation: great SEO with zero AI presence is common, and so is the reverse.
No, it runs alongside it. Google still matters, and many of the sources LLMs cite are themselves ranked pages. But they are two channels with separate scoreboards, so we run and measure both: rankings in Google, recall and citations in the engines.
It depends on where you start, which is exactly what the audit measures. The order is always the same: machine readability first, then the entity layer, then citations. The earliest movers are llms.txt, schema and entity fixes. We re-measure every two weeks against your baseline and track series, not single runs, because AI answers are volatile. What we target is the commercial outcome: being recommended in the answers to the target prompts we agree with you, measured engine by engine.
The full-cycle GEO retainer starts from $3,000/mo (September 2026); the final price is set after the audit. The manual GEO audit that opens every engagement is complimentary: you keep the scorecard and the 90-day plan whether or not we work together. Book the audit call below - the audit starts right on it - or send the request form and we come back with the audit ourselves.
The report shows the patterns; the audit is the other 80%, run on you. Live category and branded prompts across six engines with every answer logged verbatim, crawler checks with real AI user-agents, the full entity and schema review, and competitor benchmarks on the same scale. You get a health score, the verdicts AI currently gives about you, your three most expensive gaps, and a dated 90-day plan in priority order.
Some do, and we are glad when it works. But the report is roughly 20% of the job: it names the patterns and the mistakes that repeat across 100 companies. It cannot tell you which of them are costing you, in what order to fix them for your category, which prompts your buyers actually type, or whether anything moved after you shipped. Those four answers only exist once someone measures your brand, and that is what the audit and the sprints are.
It is the biggest one we see. Blocking GPTBot or ClaudeBot doesn't remove your brand from AI answers, it removes your version of the story. The model still answers questions about you using whatever it can read: review aggregators, complaint threads, competitors. In our audits, blocked sites consistently got the harshest verdicts. The fix is usually a day of work: allow the crawlers, publish a clean llms.txt, and let your version of the story back into the corpus.
Eight niche groups: web3 & crypto (incl. prediction markets), iGaming, B2B services (development agencies, consulting, legal, accounting and tax, recruiting, marketing agencies, corporate services), fintech & payments, AI / AI tools, B2B SaaS, e-commerce & DTC, and real estate & construction. Those are the categories our audit dataset covers, so we know what the AI shelf already looks like in each. Engagement is referral-first, so request the audit below and we reply when a slot opens. No pitch, no spam. We also take other high-competition niches when the audit shows a real gap to take.
request a manual GEO audit
no discovery call before the real one.
book it and we show where you lose and how you win,
or send the domain and we return with the audit.
manual GEO audit, complimentary
We run the audit by hand before we reply, so what comes back is about your domain, not a brochure.