GEO vs AEO vs SEO Optimization: How to Appear in AI Answers

Search engines are evolving rapidly. Instead of the traditional “10 blue links,” users are increasingly shown ready-made answers directly on the results page. Today, Google search results often include brief answers generated by AI. These do not list multiple links but instead provide a single, most relevant response — usually based on top-ranking sources. At the same time, more and more users are looking for answers directly in ChatGPT, Gemini, and Perplexity.

Under these conditions, traditional SEO methods are no longer sufficient. This is why new content optimization approaches have emerged: GEO (Generative Engine Optimization) and AEO (Answer Engine Optimization). In this article, we will examine in detail what these approaches are, how they relate to classic SEO and to each other, and what needs to be done for your website to appear in AI-generated answers.

What Is GEO

GEO (Generative Engine Optimization) is the optimization of websites and content for generative search engines and large language models (LLMs). These are systems that generate a synthesized answer to a user’s query rather than displaying a page with a list of links. Examples include Google AI Overviews, LLMs such as ChatGPT, and newer engines like Gemini, Claude, Perplexity. The goal of GEO is to adapt your content so that it is selected and used by these language models when generating responses.

Unlike traditional SEO, which focuses on improving rankings in standard Google search results, GEO targets generative engines that provide users with immediate answers based on information gathered from the web. In other words, GEO is about making your content the “building material” for AI-generated answers. If an algorithm considers your website reliable and well-structured, it will incorporate fragments of your text into its response and cite you as a source. The primary objective of GEO is precisely this: ensuring that your brand or page appears in AI answers.

In GEO, content is no longer the final destination for the user but rather a data source for the model. For example, when a user asks an AI assistant, “How to choose an air fryer”, the engine scans many pages and compiles a concise response. Whether AI extracts facts or wording from your material depends on how the content is written and structured. Generative algorithms prefer content that is well-organized, easy to parse, concise, and free of filler, while still containing the necessary details. Experts note that even markers such as phrases like “in summary” or the use of lists help LLMs identify the main idea and accurately rephrase it. As a result, GEO optimization often means writing content in a way that is convenient for machine understanding.

Users want to receive information instantly and in a convenient format. According to SparkToro statistics, the share of zero-click searches continues to grow. Their report shows that organic link clicks decreased by nearly 4% between March 2024 and March 2025, while the number of searches ending without any link click increased by approximately 3%. Zero-click search is significantly more convenient because it delivers instant information and in this context, it is critically important that your web page is the one providing that information.

AEO, AIO, and LEO Optimization

You have likely encountered the following abbreviations as well:

• AEO (Answer Engine Optimization)
• AIO (AI Optimization)
• LEO (LLM Optimization)

All of these are part of GEO optimization — or, essentially, different names for the same concept. Due to the rapid development of neural networks, a wide variety of terms has emerged in the marketing landscape. However, the main established term is GEO, which encompasses all of these approaches.

AEO (Answer Engine Optimization)

AEO focuses on optimizing content for systems that deliver a ready-made answer to a query without requiring an additional click. “Answer Engine Optimization” include AI tools (Google AI Overview, Bing Copilot, Perplexity) as well as classic search result features such as featured snippets, “People Also Ask” sections, and voice assistant answers (Google Assistant, Siri, Alexa). Simply put, AEO aims to make it easy for a search engine to extract a concise answer from your content and display it directly within its interface.

AIO (AI Optimization)

A broader term that refers to optimizing a brand and its content for AI systems in general — search engines, recommendation systems, and generative models (LLMs). This includes data structure, brand mentions, trust signals, and machine readability.

LEO (LLM Optimization)

A narrower subset of AIO focused specifically on optimization for large language models such as ChatGPT, Claude, and Gemini.

The goal is for LLMs to:

• be aware of the brand
• cite it correctly
• recommend it
• use its content when generating answers

AI Could Overtake Google Search by 2030

Gartner-style forecasts summarized by TTMS predict that traditional search engine query volume could decline by roughly 25% by 2026 as users shift part of their behavior to generative assistants, even though incumbents like Google and Bing are integrating these capabilities into their own experiences.

GEO vs AEO vs Traditional SEO

New approaches do not replace classic SEO — they complement it. However, their focus and metrics differ. Below is a comparison of SEO, AEO, and GEO.

Criterion	SEO (Search Engine Optimization)	AEO (Answer Engine Optimization)	GEO (Generative Engine Optimization)
Primary goal	Improve page rankings and drive clicks (traffic).	Provide content that a search engine or assistant can immediately display as an answer, mentioning your website.	Ensure your content is included in AI-generated answers so that the brand is cited within them.
Where it appears	In organic search results (standard SERP listings).	Answer blocks such as featured snippets, knowledge panels, FAQ sections in SERPs, and voice assistant responses (Google Assistant, Siri, Alexa).	AI-generated outputs such as Google AI Overviews, chatbot answers (ChatGPT, DeepSeek, Perplexity), and voice AI assistants providing extended responses.
Content format	Full-fledged pages optimized for keywords and user intent. Text is written for humans but takes SEO factors into account (keywords, headings, meta tags).	Highly structured content with clear questions and short answers. Uses lists, tables, and definitions. Language is close to conversational to match real user queries. Semantic markup (FAQ, HowTo, etc.) is actively used to highlight knowledge.	Deep, contextual content that comprehensively covers a topic (topic clusters). The text is structured by subtopics and written in clear, natural language. Includes data, examples, and expert citations to increase authority.
Key KPIs	Search rankings for target queries; organic traffic volume; CTR (click-through rate); organic conversions.	Number of featured snippets and answer box placements; share of traffic coming from these elements (including voice assistants or chatbots); brand mentions in AI answers; traffic quality (user behavior, conversion).	Frequency of citations in AI overviews and answers; share of voice in generative results (the proportion of your brand mentions compared to competitors); traffic from AI blocks; audience reach gained through AI answers — even without a click.

Is Traditional SEO Dying?

No, classic SEO is not dying. Without it, GEO website promotion will not succeed.

Krishna Madhavan, Principal Product Manager at Microsoft Bing, states:

“Whether you call it GEO, AIO, or SEO, one thing hasn’t changed: visibility is everything. In today’s world of AI search, it’s not just about being found, it’s about being selected. And that starts with content.

Traditional SEO fundamentals still matter. Crawlability, metadata, internal and external links are still critical to making your content accessible. But they are only the starting point.

In AI-powered search, ranking still exists, but instead of ordering entire pages, it determines which fragments of content are included in the final answer. Assistants like Copilot break content into smaller, structured fragments that can be evaluated for authority and relevance. These fragments are then assembled into answers, often drawing on data from multiple sources to create a single coherent response.”

How a Website Gets into AI Answers

Let’s examine the factors that determine whether your content is selected for AI-generated answers.

Factor #1: The Page Must Be Well Indexed

Most generative systems — whether Google SGE or third-party LLMs — are based on search indexes and the Retrieval-Augmented Generation (RAG) method. This means that before generating an answer, AI performs research across search results: it finds relevant pages in the index, extracts information from them, and then generates a coherent response based on that data. This approach allows AI to work with relatively up-to-date information, rather than relying solely on its training data.

Hence the first requirement: the page must be well indexed by search engines. If Googlebot or another crawler cannot access your content, does not understand its structure, or considers it low quality, it will not appear in AI overviews. Make sure there are no technical barriers: pages should return a 200 OK status code, not require authorization, not be blocked in robots.txt, and have a clear link structure and sitemap.

As Google representative John Mueller has emphasized, even in the AI era, answers are still generated based on crawlable web content — so technical SEO “does not lose its relevance.”

Factor #2: Machine Readability and Content Structure

The ranking algorithm that selects sources for AI answers considers many of the same signals as traditional search: content quality, topical keywords, heading structure, lists, emphasis, and more. However, generative models also require that the core meaning can be extracted quickly.

Practical advice: keep paragraphs and sentences short, use subheadings to create a logical hierarchy, and place direct answers next to the questions they address. Studies show that breaking content into small, meaningful chunks (2-3 sentences per paragraph) simplifies AI processing and increases the likelihood of being cited. Clear headings (H2, H3) that summarize the following section help the model understand what each block of text is about.

For example, Stephanie Yoder, Director of Content at Rebrandly, advises:

“To make text AI-friendly, use clear headings with relevant key phrases — AI scans them to understand the content.”

Factor #3: Authority and Expertise of the Resource (E-E-A-T)

The reputation and expertise of your resource play a crucial role. In essence, E-E-A-T principles (Experience, Expertise, Authoritativeness and Authoritativeness) come to the forefront. Google and other search engines aim to supply their AI answers with information from reliable sources. As a result, content created by expert authors, websites with strong niche authority, and regularly updated materials gain a clear advantage.

Make sure that key pages clearly list authors and their qualifications, include an “About Us” page, and provide contact information — all of this increases trust. Support your statements with references to sources, research, and data. For example, when you cite statistics or make factual claims, AI algorithms see it as a plus if those statements are backed by links to reputable sources or your own research. Content that is rich in facts and evidence is more likely to be selected for AI-generated summaries.

Factor #4: Content Relevance and Freshness

Information freshness also matters. Generative systems are capable of verifying data relevance through search, so content that is regularly updated, includes new insights, and reflects current trends is more attractive for AI answers. SEO practice shows that regular content updates improve both traditional rankings and the likelihood of appearing in AI overviews.

If the topic is dynamic (e.g., technology, legislation, pricing), develop a habit of frequently reviewing and refreshing relevant pages.

Factor #5: Mentions and Brand Strength

An interesting effect has been observed with branded queries: when a user’s question implies a specific brand or product, AI engines strive to cite official or authoritative sources for that brand as accurately as possible. Reports indicate that when a major, well-known brand appears in an AI Overview, click-through rates can even be higher than in traditional search — users are more inclined to click familiar names.

This gives strong brands an advantage: their information is more frequently included in AI answers, and in some cases the model may mention the brand as a trusted solution even without a direct brand query. For lesser-known websites, this is a challenge — brand awareness and expertise must be developed to earn a place among cited opinion leaders.

Core SEO Factors Still Matter

It is important to understand that generative algorithms rely on traditional search. Advanced GEO/AEO practices do not eliminate the need for classic optimization factors. If a page does not rank at least within the top 20 results for a given topic, the likelihood of it being selected for an AI answer is low.

Therefore, while working with new formats, do not forget foundational ranking signals: relevance, uniqueness, high-quality backlinks, and user experience. All of these contribute to the overall quality index used by both traditional search and AI systems.

Content Approaches for GEO and AEO

Content creation strategies differ depending on whether the goal is a concise answer (AEO) or a comprehensive AI-generated overview (GEO). Below are recommendations for both cases.

For AEO (Specific Answers and Snippets)

• Question–Answer Structure. Structure your content so that you have a separate answer for each significant question on the topic. Ideally, put the question in the headline or at the beginning of the paragraph, and immediately follow it with a clear answer in one or two sentences. This block can be formatted as an FAQ or simply as the first paragraph under the corresponding subheading. Example: Question: “How do you clean an air conditioner?” – Followed by a short answer listing 2-3 steps. The key is that the answer should be self-contained, without requiring the reader to go through the rest of the article. These 2–3 sentences are exactly what a search engine can extract and show directly to the user. Detailed explanations can follow later, but the core summary must come first.
• Focus and Clarity. When answering a question, stay on point and avoid digressions. For example, if the user asks, “Can a refrigerator be placed next to a radiator?” A poor answer: “This is a common question among many homeowners, as kitchen space is often limited…”
  A correct approach: “No, it is not recommended to place a refrigerator directly next to a radiator. Heat from the radiator forces the compressor to work under increased load, which reduces the appliance’s lifespan.” The very first sentence delivers a direct “no,” followed by a brief explanation. Aim for this structure consistently: answer first, explanation second, not the other way around.
• FAQs and Semantic Markup. Group the most common questions related to your topic and create an FAQ section on the page. Each question should be a separate subheading (H2/H3) or expandable block, followed by a concise answer. Add FAQPage structured data so that search engines clearly understand the question–answer format.
  Similarly, when appropriate, use HowTo, QAPage, and other schema types. While markup does not guarantee inclusion in snippets, it significantly increases the chances that Google or voice assistants will feature your content in AI answers.
• Conversational Language. As mentioned earlier, write in a style close to natural speech. Avoid overly bureaucratic or complex sentence structures — no one speaks to voice assistants in dissertation language. Research into voice queries shows that approximately 80% of them are phrased naturally using question words (“how,” “what,” “why,” etc.). Include real phrases that users are likely to say. This does not mean lowering language quality — it means making it more natural and lively.

For GEO (Comprehensive AI Overviews)

• Topic Clusters and Full Context. Generative answers aim to cover all aspects of a question so the user receives a complete explanation. When preparing content, think beyond a single narrow query. Fully cover the topic within the user’s intent.
  The topic cluster methodology works particularly well here. Choose a core topic (the cluster) and cover it with multiple related articles or one large, well-structured piece. For example, the topic “website optimization for AI” can include subtopics such as SEO, AEO, GEO, technical factors, and analytics — exactly as in this article.
  By linking these subtopics internally and maintaining a consistent style, you become an authoritative source for the broader context. Studies show that websites built around connected content clusters appear in AI answers more often because the model recognizes deep topic coverage.
• Examples, Case Studies, and Data. Generative AI values substance — not just general statements but concrete details. Real-world examples, short case studies, statistics, and tables enrich the content and make it more useful. The presence of numbers or facts increases the likelihood that AI will choose your content for citation. If you provide unique insights or fresh market data, the model is more likely to include those facts in its response and attribute them to you. Support key claims with references — research findings, expert quotes, or user feedback.
• Tone and Narrative Style. GEO content benefits from a tone similar to that of a qualified consultant: friendly, but information-dense. Avoid both excessive promotional language and heavy jargon. Write as if explaining a complex topic to a real person, breaking it down into understandable parts. Consider using subheadings phrased as long, conversational search queries — for example, headings starting with “What should you do if…”. These directly match potential user questions and help your content align with real-world phrasing. Such headings also increase the chances of appearing both in People Also Ask blocks and in generative AI answers.

Key Difference Summary:

• AEO content is more fragmented and highly targeted — small, precise answers for specific questions.
• GEO content is broader and more cohesive, aiming to fully cover a topic and become a trusted source for AI summarization.

Ideally, combine both approaches: start with a direct answer, then expand and deepen it.

Technical Factors and Markup

No matter how strong your content is, without a solid technical foundation it will not realize its full potential. Below are the key technical aspects that influence inclusion in AI-generated answers.

1. Page Performance and Accessibility

AI-powered search engines often generate answers in real time, accessing your website at the moment of the query as part of the RAG process. Therefore, page load speed and content delivery performance matter.

Optimize Core Web Vitals: improve loading speed, remove render-blocking scripts, enable compression — just as you would for classic SEO, but with the understanding that your “user” may now be the search engine itself.

Mobile responsiveness is mandatory. Google AI Overviews are primarily displayed on mobile devices, and traditional search has long been based on mobile-first indexing. If a page renders poorly or functions incorrectly on smartphones, its chances of being surfaced decrease. Make sure layouts do not break, fonts are readable, and no elements overlap or obscure content.

2. Crawlability and Indexability

As mentioned earlier, pages must be freely crawlable by bots. Conduct a basic SEO audit to ensure:

• no important sections are blocked in robots.txt;
• canonical URLs are set correctly;
• there is no duplicate or “thin” content that could lower the site’s overall quality score.

Site structure should help search engines understand content hierarchy. This is especially important when implementing topic clusters: interlink cluster materials, add a central hub or cluster landing page. When search engines see logical structure and internal connections, it becomes easier for them to select the most relevant page from your site for a specific AI-driven query.

3. Structured Data (Schema Markup)

Semantic markup is one of your strongest tools for answer optimization. It provides a machine-readable summary of page content.

FAQ and HowTo schemas have already been mentioned, but other important schemas include:

• Article / BlogPosting – Mark up core articles, specifying author and publication or update dates. The author field is particularly valuable for E-E-A-T, as it explicitly ties content to an expert.
• Organization / Person – Describe your organization and key individuals. This helps search engines connect your site to entities in the knowledge graph and increases trust.
• Product / Review / Rating – For commercial websites, product and review markup is critical. Structured reviews and ratings signal that content is validated by real users, which is especially important for YMYL queries.
• BreadcrumbList – Marked-up breadcrumbs clarify page hierarchy and often appear in SERPs, improving brand visibility.
• Dataset / Statistics – If you publish data, consider marking it up as a Dataset. For example, reports containing figures can include a dataset schema. While still relatively rare, search engines are moving toward using such data for precise answers.

Overall, structured data helps algorithms identify key elements and entities on your page. Avoid overuse — only mark up content that genuinely exists, but do not miss the opportunity to highlight important information for AI systems.

4. Multimedia and Emerging Formats

Generative AI in search is increasingly capable of working not only with text but also with images, video, and other media. AI Overviews already include visuals and interactive elements.

Optimize all media assets:

• add clear, descriptive alt text to images;
• surround visuals with explanatory text;
• provide transcripts or detailed descriptions for videos.

Search engines can extract answers from these text layers. Any content format — charts, infographics, audio — must be machine-understandable. Schemas such as VideoObject, ImageObject, and AudioObject are highly recommended.

The richer your multimedia presence, the greater the chance that an AI-generated answer will be multimodal and that your content will be part of it.

Technical Summary

Site speed, mobile usability, proper indexing, and semantic markup form the foundation of GEO and AEO success. Without this base, even the best content may remain invisible to AI systems. Address technical issues early so you can compete on ideas and expertise rather than removing algorithmic barriers.

Metrics and Analytics in the New Era of Search

Evaluating GEO & AEO effectiveness requires looking beyond traditional SEO metrics. Below are key indicators to track and tools that can help.

Visibility Share in AI Answers

Because traffic no longer always reflects visibility (a user may receive an answer without clicking), introduce a metric that tracks how often your site appears in AI-generated answers.

For example: Out of 100 target questions, your website appears in 20 AI answers — this means 20% visibility.

Manual tracking is difficult, so specialized tools are increasingly necessary. AI citation monitoring solutions are already emerging. Major SEO platforms are introducing relevant features:

• Ahrefs has launched a Brand Mention Tracker for Google AI Overviews, showing how often your brand is cited.
• Semrush released an AI Toolkit that allows you to enter keywords and see which domains are most frequently cited by AI on a given topic.

Startups such as Profound.ai, Goodrank, and Daydrm focus specifically on analyzing how AI models mention brands.

AI Referral Traffic

Currently, direct traffic from AI answers is still relatively small (Ahrefs reported around 0.5% of visits from AI search), but it is growing.

In web analytics, segment traffic coming from known AI platforms such as:

• chatgpt.com
• claude.ai
• perplexity.ai

Track not only volume but also quality. According to multiple companies, users arriving from AI tools convert significantly better than average. This makes sense: if a user clicks through from a detailed AI answer, their intent or purchase readiness is usually very high.

Some case studies show AI traffic converting 20+ times better than traditional search traffic. This is a strong argument for investing in GEO and AEO — fewer clicks, but each one is extremely valuable.

Brand Lift in the AI Environment

Monitor how often and in what context AI mentions your brand without direct prompting. This is harder to measure quantitatively, but indirect methods can help.

One approach is to regularly ask different AI services similar topical questions and track how often your brand or products appear in recommendations. Repeating this exercise every 1-2 months reveals brand visibility trends in AI systems.

Top 5 AI Visibility Monitoring Tools

In addition to Ahrefs and Semrush, notable tools include:

1. Peec AI — Best All-Around LLM Visibility & Prompt Tracking

Peec AI is a dedicated Large Language Model (LLM) visibility and monitoring platform that tracks how your brand appears within AI-generated responses across major engines like ChatGPT, Gemini, Claude, and Perplexity. It focuses on prompt-level visibility, competitor comparisons, and sentiment context rather than just counts of mentions.

How It Works: You input key prompts or topics, and Peec AI runs them across supported LLMs to show where your brand appears, which prompts generate the best visibility, and how competitors compare.

Features:

• Multi-engine brand mention & citation tracking
• Competitor gap analysis
• Sentiment & context reporting
• Dashboards showing visibility trends

Pricing: Starts around $89/month for basic prompt tracking; higher tiers for more prompts and enterprise features.

2. Profound — Enterprise-Grade AI Visibility & Optimization

Profound is a full-featured LLM visibility platform that combines monitoring with optimization recommendations, prompt exploration, and citation tracking across multiple AI models. It’s well suited for larger brands or those seeking deeper insights into how AI engines interpret and cite their content.

How It Works: Profound uses its own prompt database and analytics to track brand mentions, uncover new opportunities for being surfaced in AI results, and even suggest prompts to increase visibility.

Features:

• Cross-model visibility tracking (ChatGPT, AI Overviews, Perplexity, etc.)
• Prompt ideation & trends analysis
• Citations & source performance mapping
• Optimization suggestions

Pricing: Typically starts around $499/month with enterprise options available.

3. Scrunch AI — Insight-Driven AI Search Monitoring

Scrunch AI focuses on AI search engine visibility and optimization guidance by tracking prompt performance and citations across LLM platforms. It also offers insights into how to improve content so it gets picked up more often by AI systems.

How It Works: Scrunch monitors how often your brand and content show up in AI responses, then provides insights and actionable recommendations to improve AI search performance.

Features:

• Prompt performance monitoring
• Multi-engine citation tracking
• Content optimization suggestions
• Insights by model, region, and topic

Pricing: Plans begin around $300/month and scale with the number of prompts tracked.

4. AI Product Rankings — Free & Immediate Visibility Snapshot

AI Product Rankings is a free tool that lets you quickly see how your website or product appears within LLM outputs and generative search responses without needing an account. It’s ideal for small brands or early research before investing in a paid platform.

How It Works: Users enter a brand or topic and receive a report showing mentions and citations across supported AI sources.

Features:

• Completely free
• Instant visibility snapshot
• Useful for quick brand presence checks

Pricing: Free to use (no account or billing required).

5. Hall — Budget-Friendly AI Mention & Prompt Tool

Hall is a simpler, affordable AI visibility tool that helps startups and small teams track mentions and generate prompt ideas. While lighter on enterprise-grade analytics, it’s good for getting started with prompt visibility and sentiment tracking.

How It Works: Hall tracks mentions across LLM outputs and gives insights into prompt performance with clear dashboards, plus some support for prompt ideation.

Features:

• Prompt & mention tracking
• Simple dashboards
• Free plan available

Pricing: Offers a free tier and affordable upgrades compared to bigger enterprise tools.

Final Thoughts

SEO continues to evolve, and the era of AI-powered search is simply the next major challenge. GEO has emerged as the industry’s response, focusing on ensuring that your website’s knowledge works for you even when users do not click links.

By applying the principles described — structuring content around questions, making text citation-ready, strengthening expertise, and addressing technical details — you significantly increase the chances that your website will be cited and mentioned in AI-generated answers. This ensures continued value and visibility in a rapidly changing digital landscape.

Case Study: GEO Promotion for a London Bookbinder’s Website

Maria Ruzaikina is a professional bookbinder working in London, in the Pimlico area. Her business focuses on hand bookbinding, restoration of old books, gold tooling, and custom projects for private clients and designers.

At the start of the project, the website:

• had no basic SEO optimization and ranked poorly in SERPs;
• did not rank for geo-queries such as “bookbinding London,” “bookbinder Pimlico,” or “custom bookbinding near me”;
• did not appear in AI search answers (Google AI Overview, Perplexity);
• had brand recognition limited to a small circle of returning clients and word-of-mouth referrals.

The first step involved classic SEO, followed by GEO optimization focused on AI answers and local search, with strong emphasis on expertise and personal branding.

SEO and GEO Optimization Strategy

1. A new website was developed and launched on a new domain containing a geo-targeted keyword: pimlico-bookbinding.co.uk

2. The site was populated with AI-oriented content. Instead of standard SEO texts, GEO-focused articles were created to answer real user questions.

3. E-E-A-T enhancement through personal branding. Demonstrating a real expert is critical for GEO. The following were added:

• a detailed “About” page with Maria’s story;
• descriptions of her experience, craftsmanship approach, and materials;
• photos of the process and workshop (as proof of a real business);
• article authorship attribution, with the client herself acting as the expert.

This significantly increased trust from AI algorithms, which increasingly prioritize personalized expertise.

4. A clear content structure aligned with AI reading logic:

• citation-ready article formats;
• H2/H3 subheadings with direct phrasing;
• FAQ blocks on service pages and articles;
• structured data markup;
• visible authorship.

5. Brand mention strategy beyond traditional backlinks:

• inclusion in professional directories and bookbinding catalogs in the UK;
• listings in local directories of creative studios and workshops in London;
• publications and interviews on niche platforms, including: https://bookjelly.com/conversation-with-master-bookbinder-maria-ruzaikina/

Results After a Few Months

Within several months of implementing the GEO approach, the website began appearing in AI-generated answers from Google, ChatGPT, DeepSeek, and Perplexity for queries such as:

• bookbinder in London
• bookbinding London
• book restoration London
• fine binding London