We combine SEO, GEO and AEO with code so that brands show up on Google and in AI answers, and we build the work behind it together.
Nine areas; the second brain is shown above, the other eight are below. Each has its own page.
Every page has three parts: what we do, what we deliver, how we measure.
Business model, pricing logic, partnerships and new market decisions; each one weighed with numbers.
Delivered
Visibility in Google search and in ChatGPT, Gemini and Perplexity answers. A page is found first, then cited, then mentioned by name.
Delivered
iOS, Android and web apps; from the first version to the store release.
Delivered
Measurement setup, experiments and conversion; the data shows what works.
Delivered
Agents built for the project, for project tracking, social media, content and code.
Delivered
Reviewing existing code, fixing bugs, speeding up the slow parts.
Delivered
The store and marketplace side: product pages, catalog, checkout.
Delivered
Name, voice and visual language; the brand reads the same on every channel.
Delivered
Twelve work items. Each project picks the ones it needs.
We named them ourselves; the content is rewritten for every project.
We open the page side by side as a visitor and as a search bot sees it, and list the differences.
List of differencesWe request URLs that do not exist on purpose, and check whether the site returns the right status code.
Status code tableStructured data matches the visible content of the page one to one. Nothing that is not on the page goes into the schema.
Schema setWe decide which AI bots get into the site and which do not, based on the goal of the work.
Bot access decisionWe write short answer, definition and table blocks that AI answers can lift directly.
Block setWe measure what AI says about your brand with fixed questions, and fix what is wrong at its source.
Right / wrong tableWe find queries in Search Console that get impressions but no clicks, and match them to the right page.
Gap listWe find pages that no other page links to, and connect them to the internal link network.
Link mapWe find places where key information exists only in an image, and turn it into text and alt text.
Alt text listWe measure to find the step that slows the page down, and remove it with a before and after measurement.
Before / after measurementThe design system and brand guide are written so that machines can read them too; style rules are checked automatically.
Guide and check scriptWe set up agents with written rules, where a separate reviewer looks at every output.
Agent setupSeven steps from the first call to regular reports.
You can also tap the steps to move between them.
01 / 07 Foundation
We talk through the business, the goal and the limits. What will be measured, and how, is written down in the very first call.
Short note
02 / 07 Foundation
The site, search results and AI answers are audited. The starting score that comes out is the baseline for every later measurement.
Starting score
03 / 07 Foundation
Two pieces of work run at once: competitor research, and an independent audit by someone who did not do that research. At the end the two are compared.
Competitor researchIndependent audit
04 / 07 Foundation
Every finding is opened in a real browser and checked by a person. A finding that cannot be verified goes into the report as "could not be verified".
Verified findings
05 / 07 Every phase
The priority order and measurable goals are written down. High-impact, low-effort work goes first.
Plan
06 / 07 Every phase
Three handover points in order: a preview first, then the final version, then delivery. Approval is given at each point.
PreviewFinal versionDelivery
07 / 07 Every phase
The score is measured again with the same method; what changed and what did not is written down. The new measurement updates the plan, and the loop goes back to step five.
ReportNew score
Foundation Technical problems are closed, measurement is set up.
Growth Content, experiments and agents come in.
Scale What works is carried over to new pages, channels and languages.
An agent that does the work, a separate agent that checks it, a human who decides.
Research, content, code, competitor tracking, social media, reporting.
Work such as research, content, code and competitor tracking goes to agents built for the project. Every output is checked by a separate agent that did not write it; anything unsourced or wrong goes back. A person decides what goes live.
Work channel
SimulationApproval gate
The check passed. Whether it goes live is a human decision; in this example, the decision is yours.
Agents work fast. A person checks what is true.
Eight examples generalized from our own work; each one produced a rule.
AI"Change made, operation successful."
HumanThe page was opened: nothing had changed.
Rule: every write goes through human review.
AI"The product supports this feature."
HumanChecked against the documentation: the product had no such feature.
Rule: every claim is matched to its source.
AI"This keyword gets a lot of searches, add it to the page."
HumanIt did get searches, but it was not true of the product.
Rule: demand does not make something true.
AI"Best practice: add this to every page."
HumanIt went against this site’s context; it was rolled back by a human decision.
Rule: general advice is weighed against the context.
AI"The wrong sentence has been fixed."
HumanThe sentence written in its place was made up too.
Rule: no data, no claim.
AI"The script ran without problems."
HumanPart of the output was wrong; the review caught it.
Rule: automated output is read by sampling.
AI"I updated the crawl setting."
HumanRobots and the sitemap are locked; the change waited for approval.
Rule: critical settings need human approval.
AI"The market will definitely grow this year."
HumanThe figure came from a sales report; it had no method.
Rule: take the method, not the sales figure.
Eight rules that stay the same in every project.
Every line in our reports is written by these rules.
Anything that cannot be verified is written as "could not be verified".
Goals are written so they can be measured; results are not guaranteed.
Every finding is checked in a real browser before anyone says "done".
Each sits on its own line; every data point is dated.
Whoever wrote a piece of work does not approve it.
Agents suggest and carry out; a person makes the final call.
Passwords, keys and personal data are not collected.
Risks are closed before they grow.
Eight subtle points we come across in our work. The details come in the project.
BeforeUser-agent: *
Disallow: /panel/
AfterUser-agent: *
Disallow: /panel/
User-agent: AI-bot
Allow: /
/panel/ may now be open to AI-bot
Once you give an AI bot its own group in robots.txt, the rule you wrote for everyone may no longer apply to it.
Google cannot see the noindex on a page blocked by robots.txt. The order matters: noindex first, then close the door.
If the URL the page loads its data from is blocked to bots, Google may see the page as empty or as a duplicate. We also open the page the way the bot sees it.
<meta name="description">not foundThe description that looks filled in on the admin panel may never be printed on the page by the theme. We look at the source.
AI visibility has four steps. Being read is the first; being recommended is at the top.
The same question, five times
Brand mentioned in 3/5 answers
SimulationAI gives a different answer to the same question each time. That is why we measure by asking a fixed set of questions again and again.
Opening hours?Correct
Delivery time?Outdated
Store address?Doesn’t know
AI may still repeat old information you removed from your site. That is why we measure regularly.
A sitemap is a hint. The path Google really follows is internal links; we find the pages that get none.
The plan is written for the industry, the audience, the country and the goal.
Change the choices and see how the weight of the plan shifts.
E-commerce · Consumers · Türkiye · Be found in search
SimulationWeight of the work
First three steps
The real plan is written after the first call and the audit.
The scope is wide. In every area we work with a method tested on our own brands and on the projects we build.
Two ways of working side by side; the diagram is general, and agencies differ from one another too.
A model that carries the work through many hands
Work that stays at the same table from start to finish
Write about your brand or project and what you want to do. The message goes straight to Rues.