A client asks ChatGPT which company to hire for their next software project. In a few seconds, five or six names come up. Yours isn’t one of them.
You look for what to do about it, and almost everything you find about “GEO” sounds like a new, magical discipline that came to replace SEO. It’s neither. And believing the hype is the fastest way to waste time and money.
GEO doesn’t replace SEO, it extends it
Appearing in an AI’s answer isn’t a trick separate from SEO. It’s, in large part, the same work as always plus a handful of specific steps. A model like ChatGPT relies on searching, reading and understanding web pages, so almost everything that helps Google find you, the same SEO for software development companies as always, also influences whether a model can name you. GEO exists, but it’s SEO with a few extra steps, not a formula that erases what came before.
First, the bot has to be able to read you
Before any strategy, there’s a baseline condition, and it’s pure technical SEO. If your site builds its content with JavaScript and what reaches the server is an almost empty page, today you’re making it hard for LLMs to crawl you. You can have the best team and the best case studies, but if the bot can’t read your page, to it you don’t exist. It’s not GEO and it’s not magic. It’s technical SEO, and it’s where most companies fall short.
LLMs repeat what your site says clearly
The second point is also good old SEO. Your page has to say, in clear text, what you do, for whom and where. If it doesn’t say that you build custom software for companies, in the language and market of your client, the model has nothing to deduce it from. There are solid companies that stay invisible in Spanish because they describe themselves in English aiming at another market, or that talk about “digital solutions” without ever saying what they build. The bot doesn’t guess. It repeats what it finds written.
Here GEO does add something of its own, and it isn’t appearing everywhere
Up to here, everything was SEO. What GEO adds on top is finer and still being figured out. It has to do with existing as a clear entity for the models, with a single name and a single domain everywhere, your company’s data declared on the page and in its code, and a presence in the databases the models consult. And careful, this isn’t “getting mentioned in a lot of places”. Being in twenty directories doesn’t seem to move the needle. What counts is that all those signals point to one same recognizable company. They’re bets that raise the probability of being recognized, not guarantees, and it’s best to treat them that way, testing them and measuring whether they move anything.
Nobody has the formula, and whoever sells you one is selling hype
The uncomfortable truth is that nobody has a guaranteed formula for appearing in AI. It’s a system that reassembles its answer on every query, so you work with probabilities, not certainties. What we do know is that good SEO raises those probabilities, and that there are GEO steps worth testing and measuring. When we measured the software development sector in Mexico, we found that eight out of ten companies don’t appear anywhere, so the cost of entry is low and the window is open. If someone promises you a fixed spot in ChatGPT with a trick, be skeptical.
That query from the beginning, the client asking AI who to hire, isn’t won with a new trick. It’s won by getting the bot to read you, understand what you do and recognize you as a real company. And that is, almost all of it, good SEO.
