<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Gadi Samet on Medium]]></title>
        <description><![CDATA[Stories by Gadi Samet on Medium]]></description>
        <link>https://medium.com/@Gadi_87522?source=rss-9f6abe481078------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*nfRYc2O6jkjTUtT-6ynHtg.jpeg</url>
            <title>Stories by Gadi Samet on Medium</title>
            <link>https://medium.com/@Gadi_87522?source=rss-9f6abe481078------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Wed, 18 Nov 2026 06:45:11 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@Gadi_87522/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Ranking First Is No Longer Enough: 7 Rules for Being Seen in AI Answers]]></title>
            <link>https://medium.com/@Gadi_87522/ranking-first-is-no-longer-enough-7-rules-for-being-seen-in-ai-answers-d795da331a86?source=rss-9f6abe481078------2</link>
            <guid isPermaLink="false">https://medium.com/p/d795da331a86</guid>
            <dc:creator><![CDATA[Gadi Samet]]></dc:creator>
            <pubDate>Fri, 30 Oct 2026 17:29:23 GMT</pubDate>
            <atom:updated>2026-10-30T17:29:23.781Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Gjd66hrCBUk7bwW5BCM0cw.png" /></figure><p>For almost 20 years I worked on one goal: getting brands to the top of Google. This year I wrote <a href="https://zenodo.org/records/23475575">an academic paper</a> on what changed, and the short version is that the top spot is now only half the job. AI tools like ChatGPT, Gemini and Google’s AI Overviews often answer the question themselves, so a brand is only seen if the answer picks it, cites it and describes it correctly. Here are the seven rules from the paper, in plain words.</p><h3>A top ranking no longer guarantees a place in the answer</h3><p>Two numbers made this real for me. First, Ahrefs found that when an AI Overview appears, the number one result <a href="https://ahrefs.com/blog/ai-overviews-reduce-clicks-update/">gets about 58% fewer clicks</a>, based on Search Console data for 300,000 keywords as of December 2025. Second, a <a href="https://arxiv.org/abs/2512.09483">study of source selection across engines</a> found that ChatGPT cites domains that overlap with Google’s results only 11.8% of the time. For Gemini it is 12.0%, for Copilot 17.8%, and Perplexity is the outlier at 40.8%.</p><p>So a page can rank well and still be missing from the answer. Classic SEO still matters, because it is how you get found at all. What changed is what happens next, which is why I treat SEO, answer engine optimization (AEO) and generative engine optimization (GEO, making sure AI answers choose, cite and use your content) as one chain, not three competing jobs.</p><h3>Rules 1 and 2: Be one clear brand, and back every claim</h3><p>AI models work by recognizing a clear entity, meaning a specific brand, person or product, not just a page. Keep your name, founding facts, leadership, products and descriptions identical across your website, structured data, directories and, where it fits, Wikipedia and Wikidata. This matters because Wikipedia can account for <a href="https://arxiv.org/abs/2512.09483">a third to over 40% of an engine’s citations</a>, and a recent <a href="https://ahrefs.com/blog/ai-overviews-on-branded-searches/">Ahrefs study</a> found Wikipedia cited in 71.9% of the 57 AI Overviews with source lists for 100 big brand names.</p><p>Then make every important claim easy to trust. My recommendation is simple: next to each key fact, put a number, a named expert or a link to a primary source. It helps AI systems, and it also helps a human reader decide whether to believe you.</p><h3>Rule 3: Let other people say it too</h3><p>Ranking well on your own site is not enough, because answer engines look for confirmation elsewhere. In one controlled study, language models <a href="https://arxiv.org/abs/2601.03746">tended to favor institutionally corroborated information</a>, such as news outlets and government sources, over social media personas or individuals. That is why I see digital PR and earned media as part of the infrastructure for AI visibility, not a brand awareness extra.</p><h3>Rule 4: Check that the door is open</h3><p>If your robots.txt blocks AI crawlers, or a technical error serves the wrong version of a page, nothing else on this list can work. Google says there is <a href="https://developers.google.com/search/docs/fundamentals/ai-optimization-guide">no separate technical requirement</a> for AI Overviews or AI Mode beyond normal SEO. The same basics apply to crawlers from other AI companies: they need access too.</p><h3>Rules 5 and 6: Test every engine, and measure more than rankings</h3><p>Because each engine picks sources so differently, I would test your brand in ChatGPT, Gemini, Perplexity and Google, not assume that one result applies everywhere.</p><p>Measurement needs an upgrade as well. Pew Research found that users <a href="https://www.pewresearch.org/short-reads/2025/07/22/google-users-are-less-likely-to-click-on-links-when-an-ai-summary-appears-in-the-results/">clicked a traditional result in 8% of visits</a> when an AI summary appeared, compared with 15% when it did not. A dashboard can look steady while your presence inside answers quietly shrinks. I would add three numbers next to rankings and sessions: how often you are cited, how often you are included in the answer, and how often your brand is mentioned, by engine.</p><h3>Rule 7: Keep it fresh</h3><p>In the paper I argue that stale statistics make a page less likely to be cited. Refreshing key pages with updated numbers and visible update dates should be a regular editorial job, not an occasional cleanup.</p><h3>Where I would start</h3><p>I suggest four phases. In weeks 1 to 4, audit crawl access, your entity signals and what AI engines currently say about you. In weeks 4 to 12, fix your entity facts, update Wikidata and markup, and add sources to your key pages. From month 3 to 6, add digital PR and run prompt tests across engines. After that, tracking and a quarterly content refresh become part of normal reporting.</p><p>Here is a small challenge for this week. Ask ChatGPT, Gemini and Perplexity three questions about your brand, write down which pages they cite, and check whether those pages tell the right story. That quick exercise usually shows you where to begin.</p><p>The full paper, with all the sources, is here: <a href="https://zenodo.org/records/23475575">From Search Engines to AI Answers</a>.</p><h3>Reference</h3><p>Samet, G. (2026). <em>From Search Engines to AI Answers: The New Rules of SEO, GEO, and Digital Marketing</em>. Zenodo. <a href="https://doi.org/10.5281/zenodo.23475575">https://doi.org/10.5281/zenodo.23475575</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=d795da331a86" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Which Files Should You Add to Help AI Systems Read Your Website?]]></title>
            <link>https://medium.com/@Gadi_87522/which-files-should-you-add-to-help-ai-systems-read-your-website-cc006b76893c?source=rss-9f6abe481078------2</link>
            <guid isPermaLink="false">https://medium.com/p/cc006b76893c</guid>
            <category><![CDATA[generative-engine-seo]]></category>
            <category><![CDATA[seo]]></category>
            <category><![CDATA[onpage-seo]]></category>
            <category><![CDATA[google-ai-overview]]></category>
            <dc:creator><![CDATA[Gadi Samet]]></dc:creator>
            <pubDate>Tue, 29 Sep 2026 05:35:17 GMT</pubDate>
            <atom:updated>2026-09-29T05:35:17.985Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*pCyxC0yjrxhR5pZSLYFwFg.png" /></figure><h4>A practical guide to robots.txt, sitemap.xml, llms.txt, llms-full.txt, ai.txt and ai-index.json, including what is standard, what is experimental, and what you can safely copy.</h4><p>Your website can be technically open to AI crawlers and still be difficult for an AI system to understand.</p><p>That tension has created a growing collection of files with similar sounding names. Some control crawler access. Some help systems discover URLs. Some offer a cleaner representation of a site. Others are early proposals with little documented adoption. Treating them as interchangeable creates false confidence.</p><p>The useful question is not, “Which AI file should every website have?” It is, “Which problem am I trying to solve, and is there a documented mechanism for it?”</p><p>Access, discovery and understanding are three different jobs. No single file does all three.</p><h3>The foundation is still the open web, not a special AI layer</h3><p>Before adding a new file, make sure important pages return a normal 200 response, contain meaningful HTML, use crawlable links, declare canonical URLs correctly and appear in a current XML sitemap. AI search products often depend on familiar crawling and retrieval infrastructure.</p><p>robots.txt controls whether cooperative crawlers may request paths. sitemap.xml supplies a discoverable list of canonical URLs. Structured data describes entities and page meaning inside the page. I covered that last layer in<a href="https://medium.com/@Gadi_87522/the-real-job-of-schema-markup-and-why-you-still-need-it-bb8ba2bf5483"> The Real Job of Schema Markup, and Why You Still Need It</a>.</p><p>An llms.txt file is different. It is a curated map written for agents and language models. It does not replace crawling, indexing, structured data or a sitemap.</p><h3>Use this status table before you add anything</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/704/0*V-PmlPItDZ5PBCKX.png" /></figure><h3>robots.txt decides access, including for named AI crawlers</h3><p>The Robots Exclusion Protocol is the most mature file in this group. It must live at the origin root, such as <a href="https://example.com/robots.txt.">https://example.com/robots.txt.</a> Rules apply to a user agent and URL paths on that same protocol, host and port. It is a public policy file, not an authentication system, and a malicious bot can ignore it.</p><p>Google documents User-agent, Allow, Disallow and Sitemap. Lines it does not recognize are ignored. That is why adding invented directives such as LLMS: or AI-Index: to robots.txt is not reliable discovery.</p><h3>A copy ready robots.txt example</h3><pre># Allow general crawling of public content</pre><pre>User-agent: *</pre><pre>Allow: /</pre><pre>Disallow: /admin/</pre><pre>Disallow: /account/</pre><pre>Disallow: /checkout/</pre><pre># Allow ChatGPT search discovery</pre><pre>User-agent: OAI-SearchBot</pre><pre>Allow: /</pre><pre># Make a separate choice about potential training</pre><pre>User-agent: GPTBot</pre><pre>Disallow: /</pre><pre># Declare the XML sitemap</pre><pre>Sitemap: <a href="https://www.example.com/sitemap.xml">https://www.example.com/sitemap.xml</a></pre><p>Replace the domain and private paths. OpenAI documents OAI-SearchBot for surfacing pages in ChatGPT search and GPTBot as the control for potential training. This makes it possible to allow search discovery while declining the training crawler. If you want both uses, change the final Disallow: / to Allow: /.</p><p>Do not copy a crawler name merely because it appears in a blog post. Confirm the current user agent and control behavior in the operator’s official documentation first.</p><h3>llms.txt is a curated guide, not an indexing command</h3><p>The<a href="https://llmstxt.org/"> llms.txt proposal</a> asks site owners to publish Markdown at /llms.txt. Its current format requires only one element, an H1 containing the project or site name. A useful implementation then adds a short blockquote, optional context and H2 sections containing Markdown links with descriptions.</p><p>The value is editorial. Instead of making an agent infer which of thousands of URLs matter, the publisher identifies the best sources and explains what each contains. The proposal now points to adoption by documentation platforms and to examples published by major AI developer documentation sites. That is real evidence of use, but it is not evidence that Google Search, ChatGPT or every major model automatically discovers, consumes or rewards the file.</p><h3>A copy ready llms.txt example</h3><pre># Example Company&gt; Example Company provides analytics software for independent retailers.</pre><pre>&gt; This file points to the most accurate public sources about the company,</pre><pre>&gt; its product, pricing and policies.</pre><pre>Use the product and policy pages below as primary sources. Prices are in USD.</pre><pre>The changelog is the source of truth for feature availability by date.</pre><pre>## Company</pre><pre>- [About](https://www.example.com/about.md): Company background, leadership and locations</pre><pre>- [Contact](https://www.example.com/contact.md): Official contact channels</pre><pre>## Product</pre><pre>- [Product overview](https://www.example.com/product.md): Core capabilities and intended users</pre><pre>- [Pricing](https://www.example.com/pricing.md): Current plans, limits and billing terms</pre><pre>- [Documentation](https://www.example.com/docs/index.md): Setup and product reference</pre><pre>## Policies</pre><pre>- [Privacy](https://www.example.com/privacy.md): Data collection and processing practices</pre><pre>- [Terms](https://www.example.com/terms.md): Current service terms</pre><pre>## Optional</pre><pre>- [Blog](https://www.example.com/blog/): Articles and commentary</pre><pre>- [Changelog](https://www.example.com/changelog.md): Dated product updates</pre><p>Use absolute canonical URLs. Prefer clean Markdown versions when you genuinely maintain them. Do not include confidential, gated or unpublished material. The file is public, and its claims should match the pages it cites.</p><h3>llms-full.txt trades navigation for a single large context</h3><p>llms-full.txt is commonly used as a companion containing the full text of selected documentation. Unlike llms.txt, which helps an agent choose what to fetch, this file puts the chosen material in one response.</p><p>That can be convenient for a small documentation set. It becomes counterproductive when the file is huge, repetitive or stale. Context limits still exist, download size affects performance, and a combined file needs regeneration whenever a source changes. For most sites, llms.txt is the safer first step.</p><h3>A copy ready llms-full.txt example</h3><pre>&gt; Generated from selected public pages on 2026-09-13.</pre><pre>&gt; Canonical website: <a href="https://www.example.com/">https://www.example.com/</a></pre><pre>## About</pre><pre>Source: <a href="https://www.example.com/about">https://www.example.com/about</a></pre><pre>Last updated: 2026-09-01</pre><pre>Example Company provides analytics software for independent retailers.</pre><pre>[Insert the complete approved text of the page here.]</pre><pre>## Product</pre><pre>Source: <a href="https://www.example.com/product">https://www.example.com/product</a></pre><pre>Last updated: 2026-09-10</pre><pre>The product connects sales, inventory and campaign data.</pre><pre>[Insert the complete approved text of the page here.]</pre><pre>## Pricing</pre><pre>Source: <a href="https://www.example.com/pricing">https://www.example.com/pricing</a></pre><pre>Last updated: 2026-09-12</pre><pre>[Insert current plans, limits and billing terms here.]</pre><p>Generate this file from canonical source content rather than maintaining a second manual copy. Include source URLs and update dates. Exclude navigation, cookie notices, repeated footers and unrelated archives.</p><h3>ai.txt and ai-index.json need a narrower claim</h3><p>There are projects that use names such as ai.txt, ai.json and ai-index.json. The problem is not that JSON is unsuitable. The problem is interoperability. A filename has value only when a known consumer documents how it discovers the file, what fields it understands and what behavior follows.</p><p>As of this article’s publication, there is no comparable IETF standard or broadly documented support across Google, OpenAI and Anthropic for a root ai.txt or ai-index.json file. Some emerging specifications define their own JSON envelopes or verification mechanisms, but those should be described by project name and version, not presented as universal AI SEO requirements.</p><p>Therefore, do not add either file simply to “rank in AI.” Add one only when:</p><ul><li>A product you use explicitly documents the filename and schema</li><li>You operate both the publisher and consuming application</li><li>You are deliberately testing a named draft and can monitor adoption</li></ul><p>If you do publish a custom JSON index, document it as your API, version its schema and link to it from developer documentation. Do not assume unknown crawlers will discover it.</p><h3>Make the files discoverable without inventing robots directives</h3><ol><li><strong>Publish at stable HTTPS URLs.</strong> Use /robots.txt, /sitemap.xml, /llms.txt and, if justified, /llms-full.txt. Return 200 with an appropriate text or XML content type.</li><li><strong>Keep the XML sitemap in robots.txt.</strong> This is documented and widely supported. Submit the sitemap through Google Search Console and other webmaster tools you actively use.</li><li><strong>Link to llms.txt through normal web mechanisms.</strong> Add a small link in developer documentation or the site footer when appropriate. Version 2 of the llms.txt proposal also recommends an HTML link element or HTTP Link header using rel=”describedby”.</li><li><strong>Offer Markdown alternatives when useful.</strong> The proposal recommends rel=”alternate” type=”text/markdown” for a clean Markdown representation of a page.</li><li><strong>Test the whole path.</strong> Fetch every file without cookies, validate robots rules and sitemap XML, follow every llms link, and confirm that generated content matches its canonical page.</li></ol><p>&lt;link rel=”describedby” href=”/llms.txt”&gt;</p><p>&lt;link rel=”alternate” type=”text/markdown” href=”/product.md”&gt;</p><p>These link relations follow the llms.txt proposal. They are not a ranking signal promised by search engines.</p><h3>What should you actually implement?</h3><p>For a typical content or company site, the priority order is straightforward. First, maintain crawlable HTML, correct canonical tags, structured data where relevant, a clean robots.txt and an accurate XML sitemap. Next, add llms.txt if you can curate a genuinely useful map of authoritative content. Add llms-full.txt only if automation can keep it compact and current. Treat ai.txt and ai-index.json as named experiments, not a checklist item.</p><p>None of these files can force inclusion, citation or ranking. Their honest purpose is smaller: remove avoidable friction for systems that choose to crawl and read your public content.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=cc006b76893c" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Google Play Got Its Own AI Overview, and It’s Reading Your Website Too]]></title>
            <link>https://medium.com/@Gadi_87522/google-play-got-its-own-ai-overview-and-its-reading-your-website-too-e16a58c38025?source=rss-9f6abe481078------2</link>
            <guid isPermaLink="false">https://medium.com/p/e16a58c38025</guid>
            <category><![CDATA[seo]]></category>
            <category><![CDATA[google-play-store]]></category>
            <category><![CDATA[seo-tips]]></category>
            <category><![CDATA[app-store-optimization]]></category>
            <dc:creator><![CDATA[Gadi Samet]]></dc:creator>
            <pubDate>Mon, 14 Sep 2026 20:56:01 GMT</pubDate>
            <atom:updated>2026-09-29T05:24:02.662Z</atom:updated>
            <content:encoded><![CDATA[<h3>Google Play Just Got Its Own AI Overview, and It’s Reading Your Website Too</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*tDWkDI19xA9TNNxKgOltNg.png" /></figure><p>Google launched something inside Google Play that looks a lot like the AI Overviews you already see on Google Search, and almost nobody in the app world is talking about it yet. It’s called Ask Play, and where it’s actually live, it changes how people find apps. This post covers what it is, where it works and where it doesn’t, why the results change every day, and whether there’s even a way to measure any of it yet.</p><h4>What Ask Play Actually Is</h4><p><a href="https://www.androidheadlines.com/2026/05/google-play-store-io-2026-ai-features-ask-play.html">Google announced Ask Play</a> at I/O 2026, in May. It’s a Gemini powered conversational search overlay built into Google Play. Instead of typing exact keywords, someone can describe what they need in plain language, ask a follow up question, and get a personalized recommendation with a short explanation of why a specific app fits.</p><p>Here’s the part that really does resemble an AI Overview. <a href="https://www.apptweak.com/en/aso-blog/google-i-o-2026-aso-updates">For certain conversational searches</a>, the results screen now shows, in order: paid ads first, then an AI generated “Researched with Gemini” listicle, then an Ask Play listicle, then suggested follow up questions, and only after all of that, the traditional organic rankings you’d normally see first. There’s also Guided Search, which has been rolling out since September 2025, splitting a broad search into narrower ones and showing AI generated “Game Highlights” cards pulled straight from listing metadata and reviews.</p><h4>Where It Works, and Where It Doesn’t</h4><p>This is the part worth being precise about, because the honest answer is messier than the announcement made it sound.</p><p><a href="https://android-developers.googleblog.com/2026/05/io-2026-whats-new-in-google-play.html">Google’s own developer blog</a> names the US as the clearest launch market, mentioning Play Shorts rolling out there to “select developers.” It doesn’t give a real country list for Ask Play itself. <a href="https://www.yellowhead.com/blog/ask-play-gemini-google-io-2026-aso-strategy/">EEA markets, with France and Germany named specifically</a>, are significantly behind with no confirmed timeline, and Google’s blog doesn’t explain why.</p><p>There’s no official country list I could find anywhere, which tells me the actual rollout right now is narrower than “everywhere except Europe,” probably close to US only, with no announced date for when that changes. If you’re not seeing it, you’re probably not missing anything broken. You’re just not in the rollout yet.</p><p>On iOS, there’s nothing to see at all. <a href="https://www.apple.com/newsroom/2026/06/apple-unveils-next-generation-of-apple-intelligence-siri-ai-and-more/">Apple’s WWDC 2026 announcement</a> on June 8 overhauled Siri and Apple Intelligence and rebuilt search for Spotlight and Mail, with zero mention of App Store search or discovery. The App Store’s own AI features are still limited to <a href="https://www.mobileaction.co/blog/natural-language-search-and-effects-on-the-app-store/">Natural Language Search</a>, quietly introduced in iOS 18.1 back in October 2024, and <a href="https://appleinsider.com/articles/25/04/24/apple-explains-how-app-store-reviews-are-summarized-with-ai">AI generated review summaries</a>, which arrived with iOS 18.4 in March 2025. Neither one is a conversational layer, and nothing like Ask Play has been announced for iOS.</p><p>One more thing worth flagging. Apple’s new Siri AI is also delayed in the EU for iOS, iPadOS, and watchOS, though it works fine on Mac and Vision Pro, and it’s blocked in China pending regulatory approval. Two competitors delaying their AI features in the same region at the same time looks less like a coincidence and more like a pattern. I can’t point to an official statement from either company confirming why, so take that as context, not a proven reason.</p><h4>The Results Change Daily. Here’s What That Means</h4><p>Ask Play’s results run through constant server side testing, so the same search can return different results on different days for different people. Chasing an exact position here the way you’d chase a keyword rank is chasing something that won’t hold still.</p><p>What ASO people are actually recommending instead: track semantic clusters, meaning groups of related intents, rather than one exact keyword’s position, and connect whatever ranking data you do have to real install numbers in the Developer Console rather than the position number alone. Treat a good keyword rank as something that gets you considered, not something that guarantees anyone sees you.</p><h4>How to Actually Optimize for It Right Now</h4><p>There’s one concrete, doable thing here. Ask Play draws from your Play Store description, and it also pulls from your website. That means your website and your store listing need to say the same thing, in the same language, about what your app actually does. If you’ve read anything I’ve written before about why your app’s website matters for search beyond the app store itself, this is the same argument with new evidence behind it.</p><p>Beyond that, write your description so the important information comes first, since the AI is actively parsing that text, and think in terms of the specific problems and questions your app answers rather than a list of keywords.</p><h4>Can You Even Monitor This Yet?</h4><p>Honestly, not really. I looked for a tool or method that tracks whether your app shows up in Ask Play results, and I couldn’t find one. Even the ASO commentary that covers this in the most depth admits the same thing: the same query can return different results for different people on different days, and nobody has published a real way to track that consistently yet.</p><p>That’s not a gap in my research. It’s the actual state of things right now, a few months after launch. The practical workaround, for now, is manual: sample your own likely queries every so often, and watch for anything unusual in the gap between impressions and installs in your own Developer Console data. Real tracking tools will likely catch up, the same way they eventually did for AI Overviews in regular search.</p><h4>The Real Takeaway</h4><p>None of this is a reason to panic, especially since most of us, myself included, aren’t even seeing this feature yet. It’s a reason to get ahead of it while you still can. Align your website and your store listing now, before the rollout reaches your market and before the tools to measure it even exist. By the time both of those catch up, the app teams that already did the boring alignment work will be the ones AI Overview shows up already recommending.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=e16a58c38025" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[ASO Gets You Found in the App Store. SEO Gets You Found Everywhere Else.]]></title>
            <link>https://medium.com/@Gadi_87522/aso-gets-you-found-in-the-app-store-seo-gets-you-found-everywhere-else-dadec5080e4a?source=rss-9f6abe481078------2</link>
            <guid isPermaLink="false">https://medium.com/p/dadec5080e4a</guid>
            <category><![CDATA[app-store-optimization]]></category>
            <category><![CDATA[aso]]></category>
            <category><![CDATA[seo]]></category>
            <dc:creator><![CDATA[Gadi Samet]]></dc:creator>
            <pubDate>Wed, 09 Sep 2026 05:26:36 GMT</pubDate>
            <atom:updated>2026-09-29T05:25:22.706Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*8O9AY3McpQTOzML68XAxIA.png" /></figure><p>Most app teams pour their entire discovery budget into one thing: getting found in the search bar inside the App Store or Google Play. That’s called App Store Optimization, or ASO, and it works. But it’s only fighting for a slice of the people who might actually want your app. There’s a real gap between what ASO can reach and what your app’s own website can reach, and this post walks through that gap in plain terms, and why it’s worth closing.</p><h4>What ASO Can Actually Reach</h4><p>Search inside the app stores is a real channel, and it’s the dominant one once someone is already inside the store.<a href="https://www.digitalapplied.com/blog/app-store-optimization-aso-statistics-2026-data"> AppTweak’s 2026 benchmarks</a> put it at 65% of installs on iOS and 58% on Google Play, with browsing and recommendation tabs picking up most of the rest. So when ASO is working, it’s genuinely doing a lot of the job.</p><p>Here’s the catch.<a href="https://appdrift.co/guides/app-store-connect/keyword-field"> Apple only lets three fields count</a> toward that search: your app’s title (30 characters), its subtitle (30 characters), and a hidden keyword field capped at 100 characters, commas included. That’s it. Your full description, your screenshots, everything that actually explains what your app does, none of it is searchable inside the App Store. Google Play gives you a bit more room since it also indexes your full description, but the title and short description that carry the most ranking weight are still tight. And once you’re squeezed into that tiny word count, you’re also up against reviews, download momentum, and engagement, things you only partly control.</p><h4>The Users Who Never Open the App Store</h4><p>Not everyone looking for an app like yours starts inside a store. Plenty of people start on Google, and they usually type something more specific: “best budgeting app for freelancers,” “is there an app that does X,” “alternative to [competitor].” That shows up in the numbers too. Traffic arriving from outside the store makes up 12% of iOS installs and 14% of Google Play installs, per<a href="https://www.digitalapplied.com/blog/app-store-optimization-aso-statistics-2026-data"> AppTweak’s numbers</a>. That’s a real chunk of people who did their research somewhere else and showed up already convinced.</p><p>I want to be upfront about something here. I looked for a clean stat comparing how many people search for apps on Google versus inside a store, and I couldn’t find one, so I’m not going to pretend one exists. What I did find is a rough estimate from<a href="https://semnexus.com/app-store-conversion-rate-by-traffic-source-2026"> SEM Nexus</a>, an agency benchmark rather than a formal study, suggesting people who arrive from the web install at roughly the same rate, 25% to 55%, as people who search inside the store itself, 30% to 65%. Treat that as a directional hint, not a hard number. But directionally, it says something simple: people coming from the web aren’t less interested. They just never show up in your ASO numbers, because they never searched inside the app store to begin with.</p><h4>Your Website Has No Character Limit</h4><p>This is really the whole case for SEO, and it’s a simple one. A hundred characters can’t hold all the different ways real people describe what they need. Your website can. A landing page, a comparison post, a “best apps for X” guide, an FAQ, each one can target a different real question, in full sentences, with a real explanation, something no keyword field will ever let you do. That’s not a nice extra on top of ASO. It’s coverage ASO simply cannot give you, no matter how well you use those 100 characters.</p><h4>This Also Feeds How AI Recommends Apps</h4><p>There’s a newer reason this matters too, and it connects to something I’ve<a href="https://substack.com/home/post/p-214179409"> written about before.</a><a href="https://ahrefs.com/blog/ai-overview-citations-top-10"> Ahrefs recently found</a> that only 37% to 38% of Google AI Overview citations now come from pages ranking in the top 10 organic results, down sharply from about 76% just a few months earlier. Google’s shift toward “query fan out,” where one search quietly turns into several related searches behind the scenes, is a big part of why. So ranking well in Google is a weaker guarantee of showing up in an AI answer than it used to be. It’s still a real advantage, just a smaller one.</p><p>It also matters because of the kind of query someone types. AI Overviews don’t show up evenly across searches.<a href="https://neilpatel.com/marketing-stats/ai-overview-trigger-rates-by-query-length/"> Industry tracking from Adthena</a> found that longer, more specific searches trigger AI Overviews far more often than short ones in the categories they measured. Separately,<a href="https://www.omnibound.ai/blog/google-ai-overviews-statistics"> tracking that leans informational</a> found AI Overviews show up on 80% or more of searches in informational-heavy areas like healthcare and education, while short, purely transactional searches, the ones Google makes the most ad money from, get deliberately suppressed, down to around 4%. That data isn’t specific to app searches, so take it as a pattern rather than a rule, but it lines up with common sense: “best budgeting app for freelancers” or “alternative to [competitor]” is exactly that longer, comparison shaped question, the kind AI Overviews tend to answer. A bare app category search rarely gets one. The specific, decision stage question almost always does.</p><p>If your app has no real website content answering that kind of question, you’re not in that conversation anywhere, not in Google’s organic results, and not in what an AI assistant tells someone asking for a recommendation.</p><h4>The Real Takeaway</h4><p>ASO and SEO aren’t competing for the same budget. ASO wins you the people already searching inside a store, working within a hundred characters and some ranking factors you only half control. SEO wins you everyone else: the people who start on Google, the long list of specific questions your app can answer, and increasingly, the AI tools summarizing an answer before a person even opens a store. If your app’s whole discovery plan stops at the store listing, that’s not a finished plan. It’s half of one.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=dadec5080e4a" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Real Job of Schema Markup (And Why You Still Need It)]]></title>
            <link>https://medium.com/@Gadi_87522/the-real-job-of-schema-markup-and-why-you-still-need-it-bb8ba2bf5483?source=rss-9f6abe481078------2</link>
            <guid isPermaLink="false">https://medium.com/p/bb8ba2bf5483</guid>
            <category><![CDATA[on-page-seo]]></category>
            <category><![CDATA[seo-services]]></category>
            <category><![CDATA[seo-tips]]></category>
            <category><![CDATA[seo]]></category>
            <dc:creator><![CDATA[Gadi Samet]]></dc:creator>
            <pubDate>Sat, 05 Sep 2026 06:48:13 GMT</pubDate>
            <atom:updated>2026-09-29T05:25:25.954Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*PFERV02gDAARqWsIA2IU8w.png" /></figure><p>If you’ve talked to an SEO or GEO consultant in the last year, you’ve probably heard some version of “add schema markup” as step one for AI visibility. It comes up constantly, and almost always with the same quiet promise attached: mark your pages up correctly, and ChatGPT, Perplexity, and Google’s AI Overviews will start citing you.</p><p>I like that idea a lot. I just don’t think it’s true, and I went and checked.</p><p>The evidence is thinner than the advice suggests, and honestly, once you actually look at it, the real story is more interesting than the hype was. So let’s start from the beginning.</p><h3>Schema Markup, Quickly</h3><p>Structured data is a bit of code you add to a page, usually as a block called JSON-LD, that spells out what the page is about in a way machines can parse cleanly: this is an Organization, its name is X, its founder is Y. It sits quietly in the background, separate from anything a visitor actually sees.</p><p>Google recommends JSON-LD as the preferred format, and they’re refreshingly honest about what it does and doesn’t do: it makes a page <em>eligible</em> for certain search features and helps their systems understand context. That’s it. It’s an invitation, not a guarantee.</p><h3>The Five Types That Actually Matter for a Brand</h3><p>Most companies don’t need to touch more than five schema types. Here’s what each one is actually doing for you:</p><p><strong>Organization</strong> carries your entity’s core facts ,name, logo, founding date, address. The property that matters most here is sameAs, which links your site’s version of “who we are” to your canonical profiles elsewhere: Wikipedia, LinkedIn, Crunchbase. It’s the thread that ties your own claims to independently verifiable ones, and it’s the exact mechanism behind something I wrote about a while back ,if you missed it,<a href="https://gadisamet.substack.com/p/your-brand-has-a-wikipedia-problem"> <em>Your Brand Has a Wikipedia Problem You Don’t Know About</em></a> covers why that matters for how AI models talk about you.</p><p><strong>Person</strong> does the same thing for your people. jobTitle, worksFor, and a sameAs pointing to their own verified profiles is what lets a knowledge graph confidently say “Gadi Samet, COO of Buzz Dealer” instead of guessing which Gadi Samet you mean.</p><p><strong>WebPage</strong> declares what a given page is ,its headline, its author, its publisher, when it was last reviewed. It’s not glamorous. It’s the difference between a system trusting your date or byline and a system guessing at it.</p><p><strong>FAQPage</strong> pairs your actual questions with your actual answers in a structured block. It works well on pages that are genuinely FAQ-shaped. Where I see it misused most is when it gets bolted onto pages that aren’t FAQs at all, just to tick a box.</p><p><strong>Service</strong> spells out what you sell ,the service type, the provider, where it’s available. It’s the type I see left half-finished more than any other, which is a shame, because it’s the most direct, machine-readable answer to “what does this business actually do.”</p><h3>The Uncomfortable Data on Schema and AI Citations</h3><p>Here’s where it gets genuinely interesting, and where I’d love to tell you I was right to be skeptical from day one ,I wasn’t, not really, until I saw the numbers.</p><p><a href="https://ahrefs.com/blog/schema-ai-citations/">Ahrefs ran the test </a>most of us in this industry have only ever assumed the answer to. They tracked 1,885 pages that added JSON-LD between August 2025 and March 2026, compared against 4,000 control pages that didn’t, and measured citation activity 30 days before and after, across three platforms. The result: nothing moved in any meaningful way. ChatGPT citations shifted +2.2%. Google AI Mode moved +2.4%. Both are statistically indistinguishable from doing nothing at all. Google AI Overviews actually dropped -4.6% ,small, but per their statistical testing, a real effect rather than noise.</p><p>Now, here’s the twist that explains why the “add schema” advice took hold in the first place. In a separate analysis of 6 million URLs, Ahrefs found that pages cited by AI were almost three times more likely to be running JSON-LD than pages that weren’t. That’s a real pattern. It’s just not the pattern people think it is. Ahrefs is upfront that this is correlation, not causation: the kind of site that bothers implementing clean schema is usually also the kind of site already winning on the things that genuinely earn citations ,strong content, real structure, third-party coverage. Schema is a symptom of a well-run site, not the cause of its visibility.</p><p><a href="https://otterly.ai/blog/schema-markup-real-impact-ai-search/">An experiment from Otterly.a</a>i fills in the mechanical “why.” Testing five schema types over three months, they found that six out of seven AI search platforms couldn’t even retrieve schema data when asked to fetch it directly. And when a piece of information existed <em>only</em> inside FAQ schema, with nothing matching in the visible page text, not a single AI platform used it to answer a related question. Most of these systems are reading your page roughly the way a person would, by scanning the visible content, not parsing the JSON-LD tucked away in your code.</p><h3>Building It Correctly (Because It’s Still Worth Doing)</h3><p>None of this means skip schema. It means stop expecting it to do a job it was never actually doing. If you’re implementing it, a few things genuinely matter:</p><p>Use JSON-LD. It’s<a href="https://developers.google.com/search/docs/appearance/structured-data/sd-policies"> Google’s own recommended format</a>, and it’s the easiest to keep clean without touching your visible page code.</p><p>Only mark up what’s true and actually visible on the page. Google’s guidelines are explicit that hidden or inaccurate markup breaks policy, and it can cost you rich-result eligibility altogether, not just fail to help.</p><p>Fill in the properties that carry real information, not just the bare minimum a validator will accept ,sameAs, founder, and foundingDate for Organization; jobTitle and sameAs for Person.</p><p>Run it through Google’s Rich Results Test before you ship it, and go back and keep sameAs links current as profiles change. A stale one is worse than none.</p><p>And mentally file it under ORM and entity clarity, not GEO citations. That’s the job it’s actually good at.</p><h3>The Real Takeaway</h3><p>Schema markup’s real job is telling machines, clearly and consistently, who you are. That’s genuinely valuable ,it’s core infrastructure for reputation management, and it feeds the knowledge graphs that sit upstream of a lot of what AI systems end up saying about you. What it isn’t, based on everything I could actually find, is a way to get ChatGPT to start quoting you.</p><p>If your entire GEO plan begins and ends with adding schema, you’re pulling the wrong lever. The things that actually move AI citations is content these systems can read directly, and real third-party coverage ,are a separate, harder project. Worth doing both, honestly. Just not for the same reason, and not with the same expectations.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=bb8ba2bf5483" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>