Amgad Salem

Book an audit

Author: Amgad Salem

  • Reviving Dead Articles: The Content Revamp Framework I Teach

    Most sites do not have a content problem. They have a content graveyard — two hundred published articles of which maybe fifteen bring traffic, and a plan to fix it by publishing more.

    Reviving what already exists beats publishing new almost every time. The URLs are indexed, they have age, some have links, and the marginal cost of a rewrite is a fraction of commissioning a new piece. This is the framework I use for it and teach in the SEO Content Revamp workshop.

    Step 1: Classify before you touch anything

    Pull every URL with 12 months of impressions, clicks, average position, and its inbound internal link count. Then sort every article into exactly one of five buckets:

    Striking distance — ranks 5–20, real impressions, low clicks. This is your entire quick-win budget. Do these first.

    Impressions without position — high impressions, position 20+. The topic is right, the execution is not. Full rewrite candidates.

    Decayed — used to perform, now does not. Usually one of: the SERP intent shifted, the content went stale, or a competitor published something comprehensively better. Diagnose which before rewriting, because the fix differs.

    Cannibalizing — two or more URLs competing for the same query cluster. These get merged, not improved. Merging is the highest-ROI action on this entire list and the one people avoid because it feels like deleting work.

    Dead weight — no impressions, no links, no strategic purpose. Consolidate into a stronger page with a 301, or leave it alone. Do not “refresh” it; you are polishing something nobody requested.

    If you skip this step and just start rewriting the oldest posts, you will spend three weeks improving pages that were never going to rank.

    Step 2: Re-derive the intent from the live SERP

    Before rewriting a single line, open the target query and read what is actually ranking today. Not what ranked when the article was written — today.

    Three questions: What format do the top results take (guide, listicle, tool, comparison, forum thread)? What sub-questions do the People Also Ask boxes and AI Overview surface? And critically — has the intent shifted commercially? A query that returned informational guides two years ago and now returns product pages is telling you the article can no longer win, and no amount of rewriting changes that.

    Step 3: Rewrite for extraction, not for length

    The instinct is to add words. Resist it. What actually moves these pages:

    • Open with the answer. The first 40–60 words after the H1 should answer the query completely enough to stand alone. Everything else is elaboration for readers who need it.
    • Restructure headings as real questions, in the phrasing users type, ordered from most to least common.
    • Cut every paragraph that exists to hit a word count. Filler dilutes topical focus and gives extraction models more noise to sift. Delete it outright — do not replace it.
    • Add what is genuinely missing: the sub-question competitors answer and you skip, the current number, the specific example, the objection nobody addresses.
    • Update every stale fact and date, and say plainly when the piece was last reviewed.

    The rewritten version is frequently shorter than the original. That is a good sign, not a compromise.

    Step 4: Fix the internal links in both directions

    This is the step that gets skipped and it is often where the ranking actually comes from.

    Inbound: find every existing page on the site that mentions the topic and link it to the revamped URL with descriptive anchor text. Three to five contextual inbound internal links from relevant pages will move a striking-distance article more reliably than another thousand words.

    Outbound: link the revamped article to your money pages and to sibling articles in the cluster. An article with no outbound internal links is a dead end for both crawlers and readers.

    Keep a simple table of anchor text against destination URL as you go. It stops you from pointing eleven links with identical anchors at the same page, and it makes the work reviewable by someone else.

    Step 5: Republish deliberately

    Same URL — always, unless the slug is genuinely wrong and you are willing to 301. Update the modified date, not the publish date, unless the piece has been substantially rewritten. Resubmit in Search Console, and log the date so you can attribute movement later.

    Then wait. Six to eight weeks before judging anything. The single most common mistake after a revamp is panicking at week two and changing it again, which resets the clock and destroys your ability to tell what worked.

    What to expect

    On striking-distance pages with a competent rewrite and real internal linking, I typically see movement within four to six weeks on a decent share of the batch. Decayed content is slower and less certain. Merges are the most reliable of all — consolidating four thin competing pages into one strong one is close to a guaranteed improvement.

    None of this is exotic. It is just done in the right order, on the right pages, chosen before the writing starts.

    Amgad Salem is SEO Head at Eye Advertising and teaches SEO and GEO at Learn N’ Digital in Cairo. His portfolio and case studies are here, and he posts regularly on LinkedIn.

  • SEO First, GEO Second: Why I Refuse to Reverse the Order

    Every week someone asks me to skip to the interesting part. They have read that Google’s organic click-through rate is falling, they have seen their brand fail to appear in ChatGPT, and they want a GEO engagement — no audit, no technical remediation, straight to getting cited by the models.

    I turn those down, and this is the argument I make first.

    The uncomfortable mechanical fact

    Generative engines do not have a private, superior index of the web. Whatever their architecture, in production they are overwhelmingly reading pages that were discovered, fetched, parsed and ranked by conventional retrieval — their own crawlers, a search partner’s index, or a hybrid. The retrieval step is still retrieval.

    Which means the failure modes stack in a specific order:

    1. If your page cannot be crawled, nothing downstream happens.
    2. If it renders its content only in client-side JavaScript, many retrieval pipelines see an empty shell.
    3. If it is crawled but ranks nowhere for the topic, it is rarely in the candidate set the model gets to read.
    4. If it is in the candidate set but structurally unreadable, it does not get quoted.
    5. Only at that fifth layer do the interesting GEO questions begin.

    People want to work on layer five. Almost every underperforming site I audit is failing at layers one through three.

    What “SEO first” concretely means in my process

    It is not a stalling tactic and it is not a bigger invoice. It is a fixed list, and it is usually short:

    Crawl and index integrity. Can Googlebot and the AI crawlers reach the pages that matter? Is the robots.txt accidentally excluding them — including GPTBot, ClaudeBot, PerplexityBot and Google-Extended, which I now check explicitly, because I have found more than one client blocking every AI crawler while paying someone to improve their AI visibility.

    Rendering. Does the content exist in the served HTML, or only after hydration? This single issue silently disqualifies more pages from AI citation than any content problem.

    Cannibalization. If four of your pages half-answer the same question, retrieval has four mediocre candidates instead of one strong one. Classical search handles this badly. Generative retrieval handles it worse, because there is no second result — there is one answer, assembled, and your four pages compete to be a fragment of it.

    Entity clarity. Does the site state plainly who it is, what it does, and where? Organization and Person schema, a real About page, consistent naming. Models are entity-resolution machines, and ambiguity about who you are is fatal at the citation step even when the content is good.

    Baseline quality and trust signals. Author attribution, dates, sourcing, and for anything YMYL, credentials that a reasonable evaluator could verify.

    That list is unglamorous. It is also where the traffic is.

    Then, and only then, GEO

    Once the foundation holds, the optimization genuinely changes, and this is the part worth being excited about.

    Answer-first structure. Lead each section with a direct, self-contained, extractable answer of one to three sentences, then expand. Models lift passages, not pages. A passage that only makes sense after reading the two paragraphs above it will not be lifted.

    Semantic completeness over keyword density. The question is no longer whether you used the phrase. It is whether you covered the entities and sub-questions that co-occur with the topic across the corpus. Gaps in coverage are why a well-ranking page still loses the citation to a thinner competitor that happened to address the specific sub-question asked.

    Question-shaped headings. Real user phrasing, not marketing phrasing. “How much does X cost in Egypt?” retrieves; “Pricing” does not.

    Explicit, quotable claims with sources. Models preferentially cite text that carries its own evidence. A sentence containing a number, a date and an attribution is worth more than a paragraph of confident assertion.

    Consistent facts across every surface. Your site, LinkedIn, Crunchbase, directory listings, Wikipedia if you are there. When sources disagree about your founding year or your service list, the model’s confidence drops and it reaches for a competitor it can state cleanly.

    What I tell clients about measurement

    GEO measurement is genuinely immature, and I would rather say so than sell certainty. Rank tracking does not apply — there is no position three. What I actually track:

    • Prompt-set monitoring: a fixed list of 30–60 real buying-intent prompts, run on a schedule across ChatGPT, Gemini, Perplexity and Claude, logging whether the brand appears and whether it is cited or merely mentioned.
    • Referral traffic segmented by AI source in analytics, which is undercounted but directionally useful.
    • AI crawler hits in server logs — the leading indicator that anything is working at all.
    • Branded search volume, which tends to move before anything else does.

    None of these is a clean KPI. Anyone showing you a single confident “GEO score” has invented it.

    The honest summary

    GEO is real, it matters, and the shift in click behaviour is not hype. But it is a layer on top of search infrastructure, not a replacement for it, and the agencies selling it as a standalone product are mostly selling the same technical SEO audit with a new cover page — or worse, skipping the audit.

    Fix the foundation. Then optimize for the answer. In that order, every time.

    Amgad Salem leads SEO at Eye Advertising and teaches the AI-Assisted SEO & GEO programme at Learn N’ Digital. See his SEO and GEO case studies, or follow his work on LinkedIn.

  • From Tour Guide to SEO Head: What 14 Years of Answering Questions Taught Me About Search

    Before I ever opened Search Console, I spent four years walking groups of strangers through Karnak and Saqqara for Memphis Tours. It is not the background people expect from someone who now runs SEO for a Cairo agency and teaches Generative Engine Optimization. But it is the reason I am reasonably good at this job, and I want to explain why, because the lesson generalizes past my own CV.

    The tour guide’s problem is the search engine’s problem

    A tour group of thirty people has roughly nine real questions. How old is it. Who built it. Why is that part broken. Where is the bathroom. Is the price on the sign the real price. What should I not miss. Is this safe. What does that symbol mean. When do we eat.

    They arrive in an unpredictable order, from different people, phrased differently every time, and about eighty percent of the value you deliver is answering them before anyone has to ask twice. Do that and the group relaxes and listens. Fail and you spend the afternoon repeating yourself while nobody hears the actual history.

    That is a search results page. It is also, now, an AI answer. The queries are finite, they cluster, they arrive phrased a hundred different ways, and the site that wins is the one that answered cleanly before the user had to reformulate.

    When I moved into digital marketing in 2012, I did not experience keyword research as a new discipline. I experienced it as the tour briefing, written down.

    What the first years actually looked like

    I want to be honest about the shape of the transition, because “career switch” stories usually get compressed into a single brave decision, and mine was not that.

    From 2012 I was doing both. Tourism paid, digital marketing was the thing I was learning at night and on the side, and the Egyptian tourism market in those years made the decision progressively easier. By 2015 I was taking freelance SEO work seriously — that freelance line on my profile has run continuously since November 2015 and still does. In 2021 I took a full-time SEO and Google Ads role at Engosoft Middle East in Saudi Arabia, which was the first time SEO was the whole job rather than most of it. In March 2022 I joined Eye Advertising in Cairo, where I now lead SEO. Along the way there was a stint at Zaher.AI as a query engineer and then SEO head, which is where a lot of my thinking about how retrieval systems actually behave got sharpened.

    Roughly: three years overlapping, six years building, and only then a title with SEO in it. If you are two years into a switch and it feels slow, that is what it looks like from the inside.

    Three things that transferred, and one that didn’t

    Transferred: reading the room before you talk. A guide who launches into the Eighteenth Dynasty before working out that the group is exhausted and hungry has lost them. Most bad SEO content has the same failure — it is technically correct and arrives at the wrong moment in the user’s need. Search intent is not a keyword attribute. It is a guess about what state the person is in.

    Transferred: the answer goes first. In a group, you give the answer and then the context, because half the group stops listening after the answer and that is fine — they got what they needed. Written down, this is answer-first structure, and it is now the single highest-leverage formatting decision for both featured snippets and LLM citation. I was doing it verbally for years before I knew it had a name.

    Transferred: you are accountable to a real outcome. A tour either went well or it did not, and you found out that day. It made me allergic to metrics that cannot be traced to something a client actually cares about. I have written more than one report explaining that a ranking improvement did not matter.

    Did not transfer: the technical layer. Nothing about tourism prepared me for log files, crawl budget, JavaScript rendering, hreflang, or schema. That part I had to learn the ordinary way — slowly, on real sites, breaking things. Anyone selling you a shortcut through it is selling you something.

    Where the numbers came from

    Across those years I have created or optimized content for over 120 sites, implemented foundational SEO on more than 200, and run extended programmes on more than 40. The average organic traffic increase across client work has been around 590%, and a handful of sites went from effectively zero to over 400,000 monthly organic visits within a year.

    I quote those numbers with a caveat I would want quoted back at me: percentage lifts from a low base are the easiest numbers in this industry to make look impressive. A site going from 400 to 2,800 monthly visits is a 600% increase and may be commercially irrelevant. The projects I am actually proud of are the ones where the traffic curve and the revenue curve moved together, and those are the ones documented in my case studies.

    What I am doing now

    Two things, in a deliberate order.

    The first is still classical SEO — technical health, information architecture, content that deserves to rank. The second is GEO: optimizing that same content for how large language models retrieve, summarize and cite. I teach both at Learn N’ Digital, where I run the AI-Assisted SEO & GEO programme.

    The order is not decorative. A site an ordinary crawler cannot reach, parse and trust will not be quoted by an answer engine either. Every retrieval system in production is still standing on the same foundations, and I have watched a lot of people try to skip step one because step two is more interesting to talk about.

    For anyone considering the same switch

    The industry is unusually open to people arriving from elsewhere, and the reason is that SEO is not really one discipline. It is editorial judgement, technical debugging, statistics, and persuasion, sharing an office. Whatever you did before, some part of it is a component. Teachers understand information hierarchy. Journalists understand the lede. Salespeople understand objection handling. Tour guides understand that the question is going to be asked again in twenty minutes by someone who was standing right there.

    What you cannot skip is the boring technical foundation and the years of doing it on real sites where being wrong costs someone money.

    Amgad Salem is SEO Head at Eye Advertising in Cairo and SEO & GEO Instructor at Learn N’ Digital. You can review his portfolio and client case studies, or connect with him on LinkedIn.