Search is no longer just Google. Your potential customer in Egypt or Saudi Arabia is now asking ChatGPT, 'What's the best website design company in Egypt?' or 'What accounting software fits a small shop?', and getting a single answer that names specific brands. This new discipline is called GEO (Generative Engine Optimization), and this is Namra Tech's practical implementation guide.
The first thing to understand: these engines mostly read raw HTML. Bots like GPTBot, OAI-SearchBot, PerplexityBot and ClaudeBot fetch a page and don't reliably execute JavaScript. If your site is an app that builds all its content in the browser, what reaches these engines is essentially a blank page. The fix: a static prerender or server-side rendering for every important page, so the title, paragraphs and links appear directly in the page source.
Step two — let the bots in: open your robots.txt file and make sure you're not blocking GPTBot, OAI-SearchBot, PerplexityBot, ClaudeBot, Google-Extended or CCBot. Many sites block these bots with good intentions ('protecting content'), then wonder why they're never mentioned in AI answers. If you want customers coming from these engines, allowing them is a non-negotiable first step.
Step three — the llms.txt file: this is a Markdown file at the root of your site (e.g. /llms.txt) that describes your business in machine-readable language: who you are, what services you offer, and a list of links to your important pages with a short description for each. Add an expanded version (/llms-full.txt) listing every article and page. It's not yet a formally adopted standard, but it's cheap to implement and gives a clear, organized index of your identity and content.
Step four — direct-answer blocks: AI engines quote the passages that answer a question in two or three sentences. Put an explicit answer block at the top of every important page: 'What is X? X is ... it includes ... and starts from ...'. Write it as full, natural sentences rather than truncated bullet points, since quoting happens at the paragraph level.
Step five — structured data (Schema): use the correct type for each page — Organization for the company, Person for the founder, Service for service pages, SoftwareApplication for software products (not Product, which requires a price and rating and otherwise triggers Search Console errors), FAQPage for FAQs, and Article for blog posts with publisher and logo. Schema is the clearest way to tell a machine 'who I am and what I offer' without relying on interpretation.
Step six — entity consistency: if you own multiple domains and products, tie them all to a single entity: use sameAs fields in your Schema, cross-link in each site's footer, and use a consistent description of your company name and founder everywhere. Inconsistent naming or descriptions across your domains makes machines treat them as separate, unknown entities, so each one loses the strength of the others.
Step seven — verifiable numbers and facts: language models weight concrete content more heavily. State your founding year, your review count and actual rating, the markets you serve, your price range, and typical delivery timelines. A page that says '5.0/5 rating from 24 reviews on Google, operating since 2015' gets quoted far more often than one that says 'long experience and happy clients.' And never invent a number — invented figures get exposed by conflicting sources and damage your credibility.
Step eight — verify it yourself instead of guessing: request the page exactly as the bot would. A one-line example: curl -A "GPTBot" https://www.namratech.com/seo-company, then check whether the actual text appears in the output. Repeat the test with OAI-SearchBot and PerplexityBot across your top 10 pages. If you see an empty shell, the problem is rendering, not content.
Step nine — measuring impact: there's no Search Console for AI engines yet, so track proxies instead: referral traffic from domains like chatgpt.com and perplexity.ai in your analytics, a rise in direct visits accompanied by customers saying 'I found you through ChatGPT,' and periodically asking the engines themselves about your industry to see who they mention.
The biggest mistake: treating GEO as a replacement for SEO. The truth is that AI engines rely heavily on the search index and on sources Google considers trustworthy. A solid technical foundation, good Arabic content, and a verified business profile are exactly the raw material these engines quote from. GEO is a layer on top of SEO, not a substitute for it — see our guide to choosing an SEO company and its cost.
An actionable checklist for this week: open robots.txt and allow AI bots, create llms.txt and llms-full.txt, add a direct-answer block to the top of your 5 most important pages, correct your Schema types, and update your business profile data. If you want full, documented implementation, contact Namra Tech — we applied all of the above to our own site and our product sites before selling it to anyone else.