When a client asks for SEO and needs GEO

The brief says SEO. The problem is GEO.

A client emails: they “need SEO.” Rankings are down, or they want “more traffic.” The usual. But two minutes on their site and you see a different problem: search engines know they exist, but machines have no idea where they are, who they serve, or how to use their content.

They asked for SEO. What they actually need is GEO: structure and context so Google, Maps, and language models can connect their business to a place, a service area, and real questions people ask nearby.

For a three-person agency, this is where the sales conversation either gets interesting or gets messy. You are not pitching a bigger retainer. You are changing the shape of the work.

Reframing SEO without sounding like a zealot

Most clients have a mental picture of SEO that is ten years old: keywords, some blog posts, maybe “link building.” If you open with llms.txt and Schema.org you lose them in the first five minutes.

The conversation is easier if you start from their problem statement and then quietly rotate it.

Something like:

“You asked for SEO because you want more people to find you. Right now, search engines see your pages, but they do not really understand your business in your city. Before we talk about ‘more traffic’, we have to fix that understanding.”

That gives you permission to explain GEO as a precondition, not a competing service. You are not saying “forget SEO, here is my new thing.” You are saying “your version of SEO is missing the local and machine-readable part.”

Why llms.txt and Schema.org suddenly show up in the offer

Clients notice when your proposal includes unknown line items. They usually do not say “what is Schema.org?” They say “why is this more expensive than last time?”

So you explain the new pieces in business language, not protocol language.

llms.txt: who is allowed to learn from you

Instead of explaining the spec, you describe the behaviour:

I wrote about this earlier in Schema.org markup and entity recognition in 2026.

“Right now, large language models can scrape your content, learn from it, and answer people’s questions without ever showing your site. llms.txt is basically your rulebook at the front door. It tells these models what they can use and what they should not touch.”

Then you connect it to something they already care about:

• Risk: “If you publish documentation or pricing, you probably want some control over how that appears in AI tools.”
• Opportunity: “If you have helpful content, you want models to understand that you are the local expert, not just ‘a random website’.”

The key is to frame llms.txt as a governance and positioning tool, not a magic AI feature. It reduces the fear that you are experimenting on their site.

Schema.org: labelling a business for machines

Schema.org is even easier to describe without jargon:

“Your website is written for humans. Schema is the label set for machines. It tells Google and AI tools: this is a physical store here, these are the opening hours, this is the product range, these are the reviews. Without it, they guess.”

Then you link it back to GEO:

• “If someone nearby asks for what you sell, Schema helps search engines understand that you are relevant here, not on the other side of the country.”
• “If someone asks an AI assistant for ‘a store that does X near me’, Schema is one of the ways you get into that answer set.”

You are not selling markup. You are selling “stop being invisible to the part of search that is now machine-to-machine.”

Handling the cost objection without apologising

Once clients understand the idea, the next question is simple: “Why does this cost more?”

As a small agency, you cannot hide behind layers of account managers. You have to say the quiet part out loud:

• “We are doing work we did not do before: mapping your business into a structured model, not just writing copy.”
• “We have to coordinate with whoever manages your CMS, your locations, and your product data.”
• “We are taking responsibility for how your content appears in places you do not directly control.”

The useful move is to separate setup from ongoing. GEO work has a heavier first phase: deciding on entities, locations, and data sources, and implementing Schema and llms.txt. After that, the monthly work is more about keeping the model in sync with reality than endlessly “doing more SEO.”

That gives you a pricing structure that feels fair: a one-time GEO foundation, then a lighter retainer for updates and monitoring. Clients are used to paying more for implementation than for maintenance. You are just naming it explicitly.

“How do we measure this?” and the honest answer

GEO is measurable, but not in the neat way clients expect from a standard SEO dashboard. If you promise precise cause-and-effect, you trap yourself.

More on this in my article Replacing Three Client Dashboards With One Local n8n.

A safer way to frame it:

“We can measure whether search engines understand you better. We cannot isolate that from every other change in the market and say ‘this Schema tag gave you a specific number of extra leads’.”

Then you talk about the indicators you can watch:

• Search Console and analytics: more impressions and clicks for “[service] + [city/area]” queries.
• Branded search: more people searching for your name plus your location or category.
• Quality of leads: more inquiries from the right region, fewer from outside your service area.

You also acknowledge the messy part: AI answers are not fully trackable. If someone asks an assistant for “a company that does X in Utrecht,” and that assistant uses your content, you might get a phone call without a clear referrer.

Instead of pretending you can measure that perfectly, you position GEO as a defensive and foundational move:

• “If you do nothing, you stay invisible to these systems.”
• “If you do this, you give yourself a fair shot to be included when they answer.”

What you deliberately do not promise

The temptation when selling something new is to promise outcomes you cannot control. That is how you end up explaining yourself to an angry client twelve months later.

With GEO, there are a few promises you should avoid:

• “Top 3 in Google for [keyword].” You can influence relevance and clarity. You cannot control competitors’ budgets, Google’s experiments, or user behaviour.
• “Guaranteed AI assistant visibility.” LLM providers change their models and answer formats constantly. You can make content eligible, not guaranteed.
• “Exact ROI in X months.” GEO usually supports multiple channels at once: organic search, direct traffic, referrals, and AI tools. Attribution will always be fuzzy.

What you can promise is directional and structural:

• “Your business data and content will be machine-readable and location-aware.”
• “We will maintain a clear rulebook (llms.txt) for how AI tools may use your content.”
• “We will review and adjust the structure as your business or the tools change.”

That sounds less glamorous, but it is defensible. Clients who are serious about their business tend to prefer defensible over glamorous.

Keeping the conversation human in a three-person agency

In a big agency, this kind of shift becomes a slide deck and a new product name. In a three-person team, it is just a client and a video call. That is an advantage if you use it.

Read also The client code I will not paste into an AI tool.

First, you can be transparent about the learning curve without undermining trust:

“Search is changing fast. We are not pretending to control Google or OpenAI. What we can do is make sure your site speaks clearly to both humans and machines, and then adjust as we see what works.”

Second, you can tie GEO work to specific client types instead of abstract theory. For example, local businesses that invest in clear structure and geo-context tend to stop getting so many irrelevant leads and start seeing more of the right kind of traffic.

Finally, you keep the offer small and concrete. No “GEO transformation programs.” Just:

• Audit: “How findable and understandable are you now?”
• Foundation: “Implement Schema and llms.txt, fix obvious gaps.”
• Care: “Monitor, adjust, and keep your structure aligned with your real business.”

When a client asks for SEO and actually needs GEO, the hard part is not the technical work. It is having a sales conversation that upgrades their mental model without scaring them off. If you can explain why llms.txt and Schema.org are in the proposal in plain language, handle the cost and measurement questions honestly, and avoid promises you cannot keep, you are already ahead of most of the market.