Optimising AI Search: What Google, Microsoft, and Perplexity Really Say About AEO and GEO

The Silent Traffic Slump: When No One Clicks Anymore  

Imagine this: your website ranks in position 1. Your content is top-notch. Yet, traffic is declining – month after month, 40 to 60 per cent less in your analytics dashboard. No technical glitch, no penalty, no algorithm update. The answer is displayed directly within the search engine – before anyone even clicks on your website.

Google AI Overviews answer questions directly in the search results. ChatGPT and Perplexity deliver complete answers with source citations – instead of blue links. Microsoft Copilot summarises what users would have previously read on your website.

The numbers prove the scale of this shift: according to SparkToro (2024, Similarweb clickstream data), 58.5% of US search queries and 59.7% of EU search queries end without a single click on an external website. Out of 1,000 Google searches, only 360 clicks actually reach the open web. For search queries involving AI Overviews, the zero-click rate rises to 83% according to Seer Interactive.

This doesn't just change traffic – it destroys the central control metric: visitor numbers as a measure of success. Your content is cited, summarised, and consumed – but not a single page view appears in Matomo or GA4. You are providing value, but you cannot prove it.

The scale of this shift is highlighted by three key figures:

In the midst of this shift, two terms have established themselves: AEO (Answer Engine Optimisation) and GEO (Generative Engine Optimisation). The industry suggests that a completely new strategy is needed. But what do the platform operators themselves say?

Table of Contents  

1. The Blind Spot: Why Your Analytics Data Lies  

The decline in traffic is merely the symptom. The problem: your analytics dashboard no longer reflects reality. Your content is still being consumed – only invisibly, within AI platforms.

Mike King (iPullRank, AI Search Marketer of the Year 2025) coined the term "Measurement Chasm" – a growing gap between reality and what your tools actually capture. In traditional SEO, the feedback loop was clear: Keyword → Ranking → Click → Conversion. In AI search, this chain breaks down. Your content is retrieved, synthesised, and integrated into an AI response – without a single entry in Matomo or GA4.

What exactly has become invisible?  

What do the experts say?  

Rand Fishkin (SparkToro) speaks of the end of click-based attribution. The traditional principle: click on Google result → website visit → contact form → customer. Every step is traceable – this is what all attribution models in GA4, Matomo, or HubSpot are based on.

Fishkin's argument: when AI systems cite and summarise your content, this chain breaks. No referrer, no page view, no conversion path – your content provides value, but no analytics tool captures it.

His demand: "influence-based marketing measurement". No longer "Which click led to the purchase?", but rather: "How often is our brand mentioned in AI answers – and how does that influence subsequent searches and purchasing decisions?"

Mike King (iPullRank) developed a concrete framework for this: moving away from "Do we rank?" towards "Are we cited?" He recommends a three-tier measurement approach:

1. Input metrics – Is your content structured in a way that AI systems can understand and retrieve it?
2. Citation tracking – Is your content actually being cited in AI answers? In what position? In what context?
3. Business outcomes – What measurable business results (conversions, revenue) are generated from AI referral traffic?

Eli Schwartz (Product-Led SEO) warns against being blinded by visibility metrics alone: “Stop celebrating LLM visibility scores as if they pay your bills.” He demands that all metrics be consistently traced back to revenue – AI citations are only relevant if they demonstrably bring in customers.

What follows from this?  

The core question shifts: no longer “How many visits are we getting?”, but rather “How often is our brand cited in AI responses?”

• New metrics: Citation Rate and Share of Voice instead of Click-Through Rate
• Active tracking: Regularly querying AI platforms yourself instead of relying on passive dashboards
• Indirect revenue correlation: Brand influence in AI responses → subsequent direct search → conversion

2. AEO and GEO: Definitions, Differentiation, and Context  

Answer Engine Optimisation (AEO) optimises content so that it is cited as a direct response in AI-powered platforms – instead of just appearing in traditional search engine results pages (SERPs).

Generative Engine Optimisation (GEO) describes the overarching discipline: maximising visibility in AI-generated search results, i.e., anywhere answers are synthesised from multiple sources.

Why Traditional SEO Still Matters: How AI Search Works Behind the Scenes  

The table shows that AEO/GEO has different goals compared to traditional SEO. But how does an AI decide who to cite? The answer is surprisingly simple – and explains why your existing SEO work remains the ultimate prerequisite.

All major AI search platforms use Retrieval-Augmented Generation (RAG). The principle: the AI does not just invent answers out of thin air. First, it searches the traditional index for the best sources – and then formulates an answer based on them. The consequence: if you rank poorly in the search index, you won't even be found by the AI in the first place.

3. Straight From the Source: What the Platforms Say About AEO and GEO  

The industry is full of new buzzwords. However, the platform operators themselves speak a surprisingly consistent language. The following compilation is based on Glenn Gabe's analysis from 3 March 2026 – and the core message is the same across all of them.

Google: "It is SEO."  

Google's leaders have repeatedly and unequivocally positioned themselves on this in 2025/2026:

Microsoft: SEO Fundamentals Plus "Snippable" Content  

Perplexity: Brand Building as the Key  

Platform Conclusion: Good SEO IS Good AEO/GEO  

The message is clear: AEO/GEO does not replace traditional SEO – it is a subset of it. A targeted expansion of proven practices tailored for AI-specific requirements. If you do solid SEO, you already have the best possible foundation.

4. What Still Changes: 6 AEO/GEO Optimisations That Make the Difference  

While good SEO is the foundation, six areas of action will differentiate whether you are simply "found" or actually "cited":

Content Structure: The Inverted Pyramid Principle  

AI systems prefer to extract the first 1–2 sentences of a section. Therefore, structure your content according to the Inverted Pyramid principle:

1. Direct answer (first 1–2 sentences) – this is what the AI extracts
2. Key facts & context (bullet points, data, citations) – supporting evidence
3. Detailed explanation (background, methodology, case studies) – comprehensive depth
4. Related topics (links to further content) – topical authority signals

Schema Markup: Convincing Figures  

E-E-A-T: Trust Is Mandatory  

AI systems prioritise demonstrably trustworthy sources. Implement the following:

• Author profiles featuring credentials, experience, and social media links
• Source citations linking to authoritative studies and official documentation
• Visible update timestamps on every page
• HTTPS, privacy policy, legal notice (Imprint), and contact information
• Original research: in-house data, case studies, and expert quotes

robots.txt: Explicitly Allow AI Crawlers  

Without access, AI platforms cannot index your content. These are the bots you need to know:

5. TYPO3 Implementation: AEO/GEO in Practice  

The previous chapters established the what. Now, let's look at the how – with concrete code examples for TYPO3 v13 and v14 (v14 preferred) that you can deploy straight into your project.

Required Extensions  

The following extensions are required for AEO/GEO implementation in TYPO3:

robots.txt via Site Configuration  

Configure the robots.txt in your TYPO3 site configuration to grant access to AI crawlers:

# config/sites/main/config.yaml
routes:
  - route: robots.txt
    type: staticText
    content: |
      # Search engines
      User-agent: Googlebot
      Allow: /

      User-agent: Bingbot
      Allow: /

      # AI search platforms (real-time citation)
      User-agent: GPTBot
      Allow: /

      User-agent: ChatGPT-User
      Allow: /

      User-agent: PerplexityBot
      Allow: /

      User-agent: ClaudeBot
      Allow: /

      User-agent: anthropic-ai
      Allow: /

      User-agent: Google-Extended
      Allow: /

      User-agent: CCBot
      Allow: /

      # Default
      User-agent: *
      Allow: /
      Disallow: /typo3/
      Disallow: /typo3conf/
      Disallow: /typo3temp/

      Sitemap: https://example.com/sitemap.xml

Schema.org with EXT:schema – FAQPage via Fluid  

{namespace schema=Brotkrueml\Schema\ViewHelpers}

<schema:type.fAQPage>
    <f:for each="{faqItems}" as="faq">
        <schema:type.question -as="mainEntity" name="{faq.question}">
            <schema:type.answer -as="acceptedAnswer" text="{faq.answer}" />
        </schema:type.question>
    </f:for>
</schema:type.fAQPage>

Article Schema with Author via Fluid  

{namespace schema=Brotkrueml\Schema\ViewHelpers}

<schema:type.article
    -id="https://example.com/artikel/{article.uid}"
    headline="{article.title}"
    description="{article.teaser}"
    datePublished="{article.crdate -> f:format.date(format: 'c')}"
    dateModified="{article.tstamp -> f:format.date(format: 'c')}"
>
    <schema:type.person -as="author"
        name="{article.author.name}"
        url="{article.author.profileUrl}"
    >
        <schema:property -as="sameAs" value="{article.author.linkedIn}" />
    </schema:type.person>

    <schema:type.organization -as="publisher"
        name="{settings.siteName}"
        url="{settings.siteUrl}"
    >
        <schema:type.imageObject -as="logo" url="{settings.logoUrl}" />
    </schema:type.organization>
</schema:type.article>

Organization Schema via PSR-14 Event  

<?php

declare(strict_types=1);

namespace Vendor\SitePackage\EventListener;

use Brotkrueml\Schema\Event\RenderAdditionalTypesEvent;
use Brotkrueml\Schema\Type\TypeFactory;
use TYPO3\CMS\Core\Attribute\AsEventListener;

#[AsEventListener(identifier: 'site-package/add-organization-schema')]
final readonly class AddOrganizationSchema
{
    public function __construct(
        private TypeFactory $typeFactory,
    ) {}

    public function __invoke(RenderAdditionalTypesEvent $event): void
    {
        $organization = $this->typeFactory->create('Organization')
            ->setProperty('name', 'Ihre Firma')
            ->setProperty('url', 'https://example.com')
            ->setProperty('logo', 'https://example.com/logo.png')
            ->setProperty('description', 'Short description for AI understanding')
            ->setProperty('sameAs', [
                'https://www.linkedin.com/company/ihrefirma',
                'https://github.com/ihrefirma',
            ]);

        $event->addType($organization);
    }
}

Content Freshness with SYS_LASTCHANGED  

# Visible "Last updated" date
lib.lastModified = TEXT
lib.lastModified {
    data = page:SYS_LASTCHANGED
    strftime = %d.%m.%Y
    wrap = <time datetime="|" itemprop="dateModified">Last updated: |</time>
}

FAQ Content Block with Automatic Schema  

# ContentBlocks/ContentElements/faq-accordion/config.yaml
name: vendor/faq-accordion
typeName: faq_accordion
title: FAQ Accordion
description: FAQ with structured data for AI search
group: common

fields:
  - identifier: faq_items
    type: Collection
    labelField: question
    fields:
      - identifier: question
        type: Text
        required: true
      - identifier: answer
        type: Textarea
        enableRichtext: true
        required: true

<!-- Frontend.html -->
{namespace schema=Brotkrueml\Schema\ViewHelpers}

<section class="faq-accordion">
    <schema:type.fAQPage>
        <f:for each="{data.faq_items}" as="item">
            <schema:type.question -as="mainEntity" name="{item.question}">
                <schema:type.answer -as="acceptedAnswer">
                    <schema:property -as="text"
                        value="{item.answer -> f:format.stripTags()}" />
                </schema:type.answer>
            </schema:type.question>

            <details class="faq-item">
                <summary>{item.question}</summary>
                <div class="faq-answer">
                    <f:format.html>{item.answer}</f:format.html>
                </div>
            </details>
        </f:for>
    </schema:type.fAQPage>
</section>

llms.txt: Two Methods  

ddev composer require web-vision/ai-llms-txt

# config/sites/main/config.yaml
imports:
  - resource: 'EXT:ai_llms_txt/Configuration/Routes/RouterEnhancer.yaml'

6. How Do I Know If My AEO/GEO Is Working?  

In Chapter 1 we showed why traditional analytics fail. Here is the remedy: a concrete set of KPIs you can start using today – ranging from a free spreadsheet to enterprise tools.

KPIs in Three Tiers  

Citation Rate & Share of Voice: How to Measure It Concretely  

No tools, no budget required. A Google Sheet and 60 minutes per month are enough to get started.

What exactly is the Citation Rate?

The Citation Rate measures how often your brand appears as a source in AI responses – relative to the number of tested queries. Core question: When someone asks a question vital to my business, do I get cited?

What exactly is Share of Voice?

Share of Voice (also known as "Share of Model") measures your share of all citations compared to your competitors – not just if you are cited, but how large your share is. According to KnewSearch, market leaders achieve a 31% Share of Voice, with the top 3 of a category together commanding 67%.

Step-by-Step: Manual Measurement with a Spreadsheet

You need a Google Sheet (or Excel) and 60–90 minutes per month. Mike King (iPullRank) offers a free template with a Looker Studio dashboard – or you can start with your own sheet.

Step 1 – Create your query list. Gather 25 questions that your target audience would ask an AI – natural queries, not SEO keywords. Distribute them across 5 categories, with 5 queries each:

Step 2 – Test systematically. Enter each of the 25 queries into 5 AI platforms – yielding 125 data points per month. Always use incognito mode and log out of all accounts.

Step 3 – Record the results. For each query × platform, log the following in your sheet:

Step 4 – Calculate KPIs. Calculate KPIs from raw data using simple spreadsheet formulas:

Citation Rate       = ZÄHLENWENN(Zitiert;"Ja") / ANZAHL(Zitiert) * 100
Share of Voice      = Eigene_Zitationen / (Eigene + Wettbewerber) * 100
Query Coverage      = Queries_mit_Zitation / 25 * 100

Enter your results monthly into a trend sheet. After 3 months, clear patterns will emerge. Crucially, according to Otterly.AI (1 million+ data points), only 30% of brands maintain their visibility from one AI response to the next – so regular measurement is vital.

Practical Example: A TYPO3 Agency Measures Its AI Visibility

A concrete example: the fictional TYPO3 agency "AlpineWeb" based in Salzburg wants to find out if they appear in AI answers when potential clients search for TYPO3 services.

Query list (excerpt – 5 out of 25):

Results after testing on 5 platforms (excerpt):

KPI Calculation for AlpineWeb (Month 1):

With 25 queries × 5 platforms = 125 data points, AlpineWeb achieves the following result:

What AlpineWeb deduces from this:

• Immediate action: expand expertise content – the accessibility articles are cited well, so create more of this type (e.g., TYPO3 security, TYPO3 performance)
• Weakness: AlpineWeb is barely found for brand queries ("TYPO3 agencies Austria") → supplement Organization Schema with areaServed, and set up llms.txt
• Tracking: ChatGPT almost never cites → check if the content is approved for GPTBot in robots.txt

Benchmark Data by Industry  

Comparison of Monitoring Tools  

Tools like Semrush, Brand24, Otterly.ai, Gauge and SE Ranking support the AI visibility measurement.

Tracking AI Referral Traffic (Matomo & GA4)  

7. TYPO3 AEO/GEO Checklist  

All measures at a glance – sorted by impact, linked to their respective sections.

Extensions & Configuration  

Schema Implementation  

Content & E-E-A-T  

Monitoring & Measurement  

8. Traffic Magnets: Why Micro-Tools Are the Best AEO/GEO Strategy  

All previous optimisations make you visible in AI answers. But they do not solve the core problem: the click to your website is missing. The AI provides the answer – so why should anyone click your link?

The answer: because your website offers something no AI can replicate. An interactive tool, a calculator, an analysis, a quiz – something users have to interact with, not just read.

At webconsulting.at, we deploy this strategy systematically: where appropriate, we embed specialised tools directly inside our key articles – no download, no sign-up, fully usable right on the page. While not every article requires a tool, where an interactive element offers genuine value, the results are clear: longer dwell times, more backlinks, and higher AI citations compared to pure text content.

Examples From Our Practice  

Deepfake Analysis Tool – analyse images and videos across 4 forensic levels (metadata, C2PA, signal, semantics). No registration, no installation.

Screenshot

Public Administration Directory – searchable directory of 5,600+ Austrian public sector websites by category, state, and domain.

Screenshot

Content Optimisation Checker – optimise copy based on three principles: Concise, Scannable, Objective. Side-by-side comparison of original and optimised text.

Screenshot

AI Content Estimator – submit your own estimate of AI content share per industry and instantly compare it with study data.

Screenshot

AI Compendium With Downloads – 100 Q&As on AI featuring chapter downloads (PowerPoint, PDF, ZIP), videos, quizzes, and flashcards.

Screenshot

Conclusion: SEO Remains the Foundation – AEO/GEO Sharpens the Focus  

The platform operators' message is unequivocal: those who do solid SEO have the best possible foundation for AI search visibility. AEO and GEO are not a revolution – they are a targeted expansion incorporating Schema markup, E-E-A-T signals, content freshness, and technical accessibility for AI crawlers.

The critical difference lies in the goal: not just being found, but being cited. And not just being cited, but driving traffic back to your site – through interactive content that AI answers cannot replace.