Building a GEO strategy in 30 days: The step-by-step plan

What GEO is and why 30 days is realistic

GEO stands for Generative Engine Optimization, meaning the optimization of your visibility in AI answer engines rather than in classic ranking lists. When someone asks ChatGPT "Which tax advisor in Leipzig specializes in freelancers?", it is no longer a ranking list that decides but a generated text with only a few names in it. Your goal: to appear in these answers correctly, completely and positively. That is measurably different from SEO, but it overlaps in many fundamentals.

Why 30 days? Because GEO is not a one-off major project but a cycle of observing, adjusting and measuring again. In one month you complete a full first round: you know how AI systems talk about you today, you have closed the biggest gaps and you have a measurement system in place. After that you repeat the cycle monthly. Perfection is not the goal of the first 30 days, but a working, repeatable process is.

The plan works across industries. A trades business, a SaaS company, a law firm and an online shop all go through the same four waves, but fill them with different content. What matters is the order: first understand where you stand, then act. Anyone who immediately rewrites texts without knowing the starting point is optimizing in the dark.

Week 1: The audit – how AI talks about you today

Start with honest measurement. Ask the relevant AI systems the 15 to 20 questions your customers actually ask. For a physiotherapy practice these might be "Where can I get a short-notice appointment for manual therapy in Cologne?". For each question, note: Are you mentioned? Are the facts correct? Which competitors appear? Use several systems, because ChatGPT, Perplexity and Google AI Overviews answer differently and sometimes draw on different sources.

Document the answers verbatim in a table, with the date. This is your baseline. Without this starting point you cannot later prove progress or make the case to management. Pay particular attention to false statements: outdated opening hours, wrong locations, confused services. Such errors weigh more heavily than simply not being mentioned, because they actively cause harm and cost trust.

At the end of the week, you prioritize. Distinguish between "misrepresented" (immediate action), "not mentioned at all" (needs building up) and "mentioned, but weak" (fine-tuning). This prioritization steers the next three weeks. Anyone who works thoroughly here saves weeks of unfocused busywork later.

Week 2: Sharpen content – deliver answers, not brochures

AI systems prefer content that answers a specific question directly and completely. Go through your most important pages and rewrite them so that the first paragraph clears up the question immediately, before any details follow. An online shop for running shoes should offer a clear, self-contained and quotable paragraph in response to "Which shoe is suitable for overpronation?", not an advertising text. Think in question-and-answer blocks rather than marketing prose.

Create an honest FAQ that takes up real questions – including uncomfortable ones like prices, waiting times or limitations. It is precisely these passages that AI systems like to quote, because they are unambiguous. State facts explicitly: location, certifications, areas of expertise, target groups. Vague phrases like "leading provider" do not help a machine; "specialized in employment law since 2011, three certified specialist lawyers", on the other hand, does a great deal. Concreteness is the currency.

Add structured data (Schema.org) where it fits: organization, location, products, FAQ. These machine-readable details help systems capture your facts cleanly. They are not a magic trick, but they significantly reduce the risk of mix-ups and false statements.

Week 2 in detail: The most important content building blocks

Not every page is equally important. This week, concentrate on the handful of pieces of content that describe your core offerings, and bring them to a level a machine can process without difficulty. Clear headings, short paragraphs, unambiguous facts and few filler words beat any glossy phrasing.

The following list helps you structure the week's workload. Work through it from top to bottom and tick off what is done. Whatever you don't manage in week 2 deliberately moves to the following month rather than into late-night shifts.

• Core service pages with a direct answer in the first paragraph
• FAQ with 10 to 15 real, including uncomfortable, questions
• Explicit facts: location, year founded, specialization, target group
• Schema.org markup for organization, products and FAQ
• Consistent details across all pages, no contradictions

Week 3: Strengthen mentions – trust is built outside your site

AI systems rely heavily on what others write about you, not just on your own website. This week is about external mentions: industry directories, specialist portals, press, review platforms and thematically relevant blogs. First make sure you have clean, consistent entries in the important directories of your industry – for a law firm, lawyer-specific portals; for a restaurant, reservation and review services; for a SaaS product, relevant comparison sites.

Consistency is decisive here. If your company name, your address or your specialization appear differently in various places, uncertainty arises, and machines then tend toward vague or false statements. Standardize these details systematically. Wherever you can place genuine specialist articles, interviews or guest posts, do it – not because of backlinks in the old sense, but because real substance about you is created there that can be quoted.

Reviews and testimonials count as well. Actively ask satisfied customers for feedback on the relevant platforms. The quantity and recency of genuine reviews shape how confidently an AI system positions you. Manipulated reviews are risky and short-sighted; rely on authentic voices.

Week 4: Measure and adjust – turning a one-off into a system

Now you repeat the measurement from week 1 with exactly the same questions. This shows you directly what has moved: Are you mentioned in more questions? Have false statements disappeared? Has the tone improved? Record the new answers alongside the old ones. This comparison is your most important proof that GEO works, and the basis for budget decisions in the company.

Don't expect miracles after four weeks. Some changes to your website only take effect once AI systems update their data, which takes days to weeks. External mentions take longer anyway. So evaluate trends, not individual moments. If three of ten previously false statements have been corrected and you now appear for two new questions, that is a solid first month.

Finally, you set the rhythm for the following months. Define a small, fixed set of metrics – mention rate, share of correct facts, number of relevant mentions – and measure it monthly on the same day. The 30-day sprint thus becomes a lasting process that runs with manageable effort.

Typical mistakes and how to avoid them

The most common mistake is starting without measurement and hectically rewriting texts. Without a baseline you never know whether your work has had any effect, and you get lost in detailed optimizations of unimportant pages. Second classic: feeding AI systems advertising language. "Innovative, leading, unique" is meaningless to a machine. Concrete, verifiable facts win out, flowery adjectives do not.

Another stumbling block is contradictions between sources. If your website says something different from your industry directory, uncertainty arises, and you cannot control which version the system believes. Equally risky: optimizing everything once and then stopping. GEO is cyclical, because both the AI systems and your environment are constantly changing. Anyone who stops after 30 days quickly loses the head start again.

Finally, many underestimate the role of external voices and rely solely on their own website. Your site is the foundation, but trust is built from the sum of many sources. So plan in time for directories, reviews and specialist articles from the start, not just for your own text.

What a realistic 30-day plan looks like in the calendar

A plan only becomes reliable once it is set in concrete time blocks. Reserve two fixed slots of 90 minutes each per week and treat them like a client appointment. In week 1, the first slot belongs to the audit, the second to collecting your most frequent customer questions. This way you build up a base of material early on that you can draw from in the following weeks, instead of starting from scratch each time.

In weeks 2 and 3 the focus shifts to writing and distributing. Plan one slot for creating concrete answer pages and one for maintaining directories and mentions. What matters is that you produce small, finished units rather than one big text that never goes online. A published FAQ page with five answers is worth more than a perfect draft in the drawer.

Week 4 is deliberately more lightly scheduled. One slot for measuring, one slot for improving the two weakest spots. If you keep up this rhythm over the month, you will have invested around twelve hours in the end and built a system that can then be continued with one to two hours per week.

A worked example: from empty profile to cited source

Take a regional trades business with a plain website and no structured answers. On day 1 no AI answer names the business, simply because no clear, quotable statements exist. In week 2 six concrete answer pages are created: services, catchment area, price range, process, emergency service and qualifications. Each page answers exactly one question in two to three clear sentences, right at the start.

In week 3 three mentions are added: an updated business directory entry, a guest post on the local portal and a correct entry in the map service. These external signals confirm what is stated on your own site. It is precisely this agreement between your own statement and outside confirmation that AI systems respond to when deciding which source to quote.

By the end of the month the business appears in AI answers for two of five test questions, mostly for the questions about emergency service and catchment area, because these answers were worded most unambiguously. That is not a lucky hit but the direct result of clear wording plus confirming mentions.

Industry differences: Why your plan doesn't look the same everywhere

The 30-day framework stays the same, the emphasis shifts depending on the business. In local services, catchment area, availability and concrete services count. Here it pays off to work on map services and regional directories early, because AI answers rely heavily on these sources when there is a location reference.

In B2B with products that need explaining, it is more the specialist questions, comparison criteria and use cases that get quoted. Instead of opening hours you need clean answers to questions like suitability, integration and the limits of your solution. Specialist articles and references have more impact here than entries in general directories.

In e-commerce, in turn, product details, comparability and independent reviews decide the outcome. Make sure that information on material, dimensions and use is clearly and machine-readably available. Whatever the industry, the rule is: don't transfer the four weeks rigidly, but weight the slots toward where your customers actually ask their questions.