Short answer
Is finding customers with AI actually possible?
Finding customers with AI has two readings and only one of them works. The first is generating messages with AI and sending them in bulk — scalable, with returns that fall fast. The second is the brand appearing in the answer when a customer asks a language model; nothing is paid per click and the person already has the need. Galata Media's GEO service builds the second.
Key takeaways
- The durable form of finding customers with AI is not having a model write your outreach but being named in the model's answer.
- When a user asks a language model, they get three names rather than ten links; being one of those three is work separate from classic ranking.
- Getting into a model's training data cannot be targeted; what can be is being readable and quotable when the model searches.
- Access is the precondition for appearing in the answer: if OAI-SearchBot and its equivalents are blocked in robots.txt or at the firewall, no content work produces a result.
- Traffic from this channel is small but mature: the comparison already happened inside the model and the visit arrives at the decision stage.
Which reading of this actually works?
The phrase describes two different jobs, and confusing them sends the budget to the wrong place. The first reading is generating bulk messages and proposal copy with AI: it scales, but the same tools exist on the receiving end, so response rates fall quickly and the channel exhausts itself. The second is the brand appearing inside the answer when a customer asks a model.
The economics of the second reading run differently: the person is asking because they already have a need, they see three names in the answer, and the comparison happens there. Nothing is paid per click, and being inside the answer works again every time the same question is asked. Competition is still far below what classic search carries — because most companies do not even measure this surface.
You cannot know what is happening on a surface you do not measure. Typing your brand name into ChatGPT is not a measurement — a brand named in the question appears in the answer. Measurement uses questions that never name the brand, repeated across runs.
What does it take to be inside the answer?
Open crawler access
If OAI-SearchBot, Claude-SearchBot and PerplexityBot are blocked in robots.txt or at the firewall, appearing in the answer is impossible. This is the cheapest fix with the fastest effect.
Put the answer in a standalone passage
A model does not read the page end to end; it scores passages one by one and takes the strongest. What is needed is a 40–75 word paragraph that names its subject and still makes sense lifted out of context.
Make the brand resolvable
Before recommending a name a model wants to tie it to an organisation. The schema.org Organization markup, the details on the site and the Google Business Profile have to carry the same name, address and phone.
Take the baseline measurement
Run twenty questions that never name the brand across four engines, in clean sessions, repeated; record in how many your name appears. On the first run that number is zero for most organisations, and it is a baseline.
Frequently asked
Can I pay ChatGPT to appear?
- For organic answers there is no such channel: there is no submission form and no directory, and visibility depends entirely on crawler access and page structure. A service sold as "we get you added to ChatGPT" should not be bought — the only sellable work is opening access and making the content quotable.
Can traffic from this channel be measured?
- Partly. When a user clicks the link in an answer it shows in analytics as a referral from chatgpt.com or perplexity.ai; someone who merely reads your name and then contacts you directly cannot be attributed. So measurement runs on two legs: the referral in analytics, and a regular test recording whether the brand appears in the answers at all.
Sources
- 01OpenAI crawlers: GPTBot, OAI-SearchBot, ChatGPT-User, OAI-AdsBot — OpenAI, 2026
- 02Does Anthropic crawl data from the web, and how can site owners block the crawler? — Anthropic, 2026
- 03PerplexityBot and Perplexity-User — Perplexity, 2026
- 04GEO: Generative Engine Optimization — Aggarwal, Murahari, Rajpurohit, Kalyan — KDD 2024, 2024
- 05AI features and your website — Google Search Central, 2026
Queries this page answers
- finding customers with ai
- ai lead generation
- get cited by chatgpt