<?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"><channel><title><![CDATA[Growth-onomics: AI Search & Performance Marketing]]></title><description><![CDATA[Growth-onomics: AI Search & Performance Marketing]]></description><link>https://growth-onomics.hashnode.dev</link><generator>RSS for Node</generator><lastBuildDate>Sun, 22 Nov 2026 16:12:13 GMT</lastBuildDate><atom:link href="https://growth-onomics.hashnode.dev/rss.xml" rel="self" type="application/rss+xml"/><language><![CDATA[en]]></language><ttl>60</ttl><item><title><![CDATA[Technical SEO checklist for Cyprus websites: Greek, English and hreflang done right]]></title><description><![CDATA[Miltos George, Growth-onomics Cyprus 20 November 2026 Most Cyprus websites serve at least two audiences: Greek speakers on the island and English speakers…]]></description><link>https://growth-onomics.hashnode.dev/technical-seo-checklist-for-cyprus-websites-greek-english-and-hreflang-done-right</link><guid isPermaLink="true">https://growth-onomics.hashnode.dev/technical-seo-checklist-for-cyprus-websites-greek-english-and-hreflang-done-right</guid><category><![CDATA[SEO]]></category><category><![CDATA[Web Development]]></category><category><![CDATA[technical seo]]></category><category><![CDATA[cyprus]]></category><dc:creator><![CDATA[Growth-onomics]]></dc:creator><pubDate>Fri, 20 Nov 2026 08:10:01 GMT</pubDate><content:encoded><![CDATA[<p>Miltos George, Growth-onomics
Cyprus
20 November 2026</p>
<p>Most Cyprus websites serve at least two audiences: Greek speakers on the island and English speakers who live, work or buy here. Many also want to reach Greece. That mix is where technical SEO tends to break. Language versions point at the wrong pages, Greek slugs turn into long strings of percent-encoding, and canonicals quietly undo the hreflang setup.</p>
<p>This is a developer-focused checklist for getting the foundations right. It assumes you control the templates, the server or CDN, and the build pipeline.</p>
<h2>1. Pick a URL structure before you write a line of hreflang</h2>
<p>You have three common options for a bilingual Cyprus site:</p>
<table>
<thead>
<tr>
<th>Structure</th>
<th>Example</th>
<th>Pros</th>
<th>Cons</th>
</tr>
</thead>
<tbody><tr>
<td>Subfolders</td>
<td><code>example.com.cy/el/</code>, <code>example.com.cy/en/</code></td>
<td>One domain, shared authority, simple hosting</td>
<td>Needs clean routing per language</td>
</tr>
<tr>
<td>Subdomains</td>
<td><code>el.example.com</code>, <code>en.example.com</code></td>
<td>Easy to host separately</td>
<td>Treated more like separate sites, more to maintain</td>
</tr>
<tr>
<td>Separate ccTLDs</td>
<td><code>example.com.cy</code> and <code>example.gr</code></td>
<td>Strong country signal</td>
<td>Separate domains to build up and manage</td>
</tr>
</tbody></table>
<p>For most local businesses, subfolders on one domain are the simplest to maintain. A <code>.com.cy</code> domain sends a clear Cyprus signal; a <code>.com</code> with subfolders works fine too if the hreflang and content are right. Whatever you pick, keep it consistent: do not mix <code>/en/</code> on some pages with <code>?lang=en</code> on others.</p>
<h2>2. Hreflang for el-CY, en and el-GR</h2>
<p>Hreflang values are an ISO 639-1 language code, optionally followed by an ISO 3166-1 alpha-2 region code. So <code>el-CY</code> is Greek for Cyprus, <code>el-GR</code> is Greek for Greece, and <code>en</code> is English for everyone. A region code on its own (for example <code>cy</code>) is not valid.</p>
<p>Every version of a page should list all versions, including itself, using fully qualified URLs:</p>
<pre><code class="language-html">&lt;head&gt;
  &lt;link rel="alternate" hreflang="el-CY" href="https://www.example.com.cy/el/ypiresies/" /&gt;
  &lt;link rel="alternate" hreflang="el-GR" href="https://www.example.com.cy/gr/ypiresies/" /&gt;
  &lt;link rel="alternate" hreflang="en" href="https://www.example.com.cy/en/services/" /&gt;
  &lt;link rel="alternate" hreflang="x-default" href="https://www.example.com.cy/en/services/" /&gt;
&lt;/head&gt;
</code></pre>
<p>A few rules that save debugging time:</p>
<ul>
<li><strong>Return links are required.</strong> If page A lists page B but B does not list A, Google ignores the pair.</li>
<li><strong>Only create an <code>el-GR</code> version if the content is really different</strong>, for example different prices, delivery terms or contact details. If it is identical, a single <code>el</code> version is usually enough.</li>
<li><strong>Use <code>x-default</code></strong> for the page users should see when no listed language or region matches, often your English page or a language selector.</li>
<li><strong>Use one method per site.</strong> HTML <code>&lt;link&gt;</code> tags, HTTP headers or the XML sitemap all work, but mixing them makes errors harder to find.</li>
</ul>
<h2>3. Greek characters in URLs</h2>
<p>Greek slugs are allowed, but they appear percent-encoded when copied and shared. <code>/υπηρεσίες/</code> becomes <code>/%CF%85%CF%80%CE%B7...</code> in many tools, emails and analytics reports.</p>
<p>Two sensible approaches:</p>
<ol>
<li><strong>Transliterate</strong> Greek slugs into Latin characters (<code>/el/ypiresies/</code>). Short, readable everywhere and easy to type.</li>
<li><strong>Keep native Greek slugs</strong> and make sure your server, CMS, CDN and redirects handle UTF-8 consistently.</li>
</ol>
<p>Most teams choose transliteration for maintainability. Whatever you choose, decide once, document the transliteration rules and never generate two URLs for the same page. Keep titles, headings and body text in proper Greek; only the slug changes.</p>
<h2>4. Canonicals that agree with hreflang</h2>
<p>Each language version should be self-canonical:</p>
<pre><code class="language-html">&lt;!-- On the Greek (Cyprus) page --&gt;
&lt;link rel="canonical" href="https://www.example.com.cy/el/ypiresies/" /&gt;
</code></pre>
<p>Do not canonicalise the Greek page to the English one. That tells Google the Greek page is a duplicate, and the hreflang set falls apart. Also check that canonical URLs match the hreflang URLs exactly: same protocol, same <code>www</code> choice, same trailing slash.</p>
<h2>5. XML sitemaps</h2>
<p>Hreflang can live in the sitemap instead of the <code>&lt;head&gt;</code>, which is handy for large sites:</p>
<pre><code class="language-xml">&lt;urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9"
        xmlns:xhtml="http://www.w3.org/1999/xhtml"&gt;
  &lt;url&gt;
    &lt;loc&gt;https://www.example.com.cy/el/ypiresies/&lt;/loc&gt;
    &lt;xhtml:link rel="alternate" hreflang="el-CY" href="https://www.example.com.cy/el/ypiresies/" /&gt;
    &lt;xhtml:link rel="alternate" hreflang="en" href="https://www.example.com.cy/en/services/" /&gt;
    &lt;xhtml:link rel="alternate" hreflang="x-default" href="https://www.example.com.cy/en/services/" /&gt;
  &lt;/url&gt;
&lt;/urlset&gt;
</code></pre>
<p>List only canonical, indexable URLs that return <code>200</code>. Submit the sitemap in Search Console and reference it in <code>robots.txt</code>.</p>
<h2>6. Core Web Vitals basics</h2>
<p>Google's Core Web Vitals are Largest Contentful Paint (loading), Interaction to Next Paint (responsiveness) and Cumulative Layout Shift (visual stability). Check them with field data in Search Console and PageSpeed Insights, not only lab tests on a fast office connection.</p>
<p>Common fixes on local business sites:</p>
<ul>
<li>Serve images in modern formats with explicit <code>width</code> and <code>height</code>.</li>
<li>Load Greek web fonts carefully: subset them and use <code>font-display: swap</code>.</li>
<li>Delay chat widgets, map embeds and third-party scripts until they are needed.</li>
<li>Avoid cookie banners that push content around when they appear.</li>
</ul>
<h2>7. LocalBusiness schema in JSON-LD</h2>
<p>Structured data helps search engines connect your site with your business details. All values below are <strong>placeholders for illustration only</strong>; replace them with your real, verified details.</p>
<pre><code class="language-html">&lt;script type="application/ld+json"&gt;
{
  "@context": "https://schema.org",
  "@type": "LocalBusiness",
  "name": "EXAMPLE Business Name",
  "url": "https://www.example.com.cy/",
  "telephone": "+357-00-000000",
  "address": {
    "@type": "PostalAddress",
    "streetAddress": "EXAMPLE Street 1",
    "addressLocality": "Limassol",
    "postalCode": "0000",
    "addressCountry": "CY"
  },
  "openingHours": "Mo-Fr 09:00-17:00",
  "sameAs": ["https://www.example.com/your-profile"]
}
&lt;/script&gt;
</code></pre>
<p>Use a more specific type where one fits, and validate the markup with Google's Rich Results Test.</p>
<h2>8. Google Business Profile consistency</h2>
<p>Your business name, address and phone number should match across the website, schema and Google Business Profile. In Cyprus, addresses are often written differently in Greek and English, so choose one format per language and use it everywhere. Link the profile to the matching language version of your site.</p>
<h2>9. Hosting and CDN notes for Cyprus</h2>
<p>Your visitors are mostly on the island, in Greece and across Europe. A host with European data centres plus a CDN usually gives good response times. Check that:</p>
<ul>
<li>The CDN does not cache one language version and serve it to everyone.</li>
<li>You do not auto-redirect users by IP or browser language. Offer a language switcher instead, so Googlebot can crawl every version.</li>
<li>HTTPS, HTTP/2 or HTTP/3 and compression are enabled.</li>
</ul>
<h2>The checklist</h2>
<table>
<thead>
<tr>
<th>Area</th>
<th>Check</th>
</tr>
</thead>
<tbody><tr>
<td>URL structure</td>
<td>One pattern for all languages, documented</td>
</tr>
<tr>
<td>Hreflang</td>
<td>Valid codes, return links, <code>x-default</code> set</td>
</tr>
<tr>
<td>Slugs</td>
<td>Transliteration rule chosen and applied</td>
</tr>
<tr>
<td>Canonicals</td>
<td>Self-referencing, matching hreflang URLs</td>
</tr>
<tr>
<td>Sitemaps</td>
<td>Only canonical <code>200</code> URLs, submitted</td>
</tr>
<tr>
<td>Core Web Vitals</td>
<td>Field data reviewed, main issues fixed</td>
</tr>
<tr>
<td>Schema</td>
<td>LocalBusiness JSON-LD validated with real data</td>
</tr>
<tr>
<td>Business Profile</td>
<td>Name, address, phone consistent</td>
</tr>
<tr>
<td>Hosting/CDN</td>
<td>No language cache mix-ups, no forced redirects</td>
</tr>
</tbody></table>
<p>If you want a second opinion on your setup, our <a href="https://growth-onomics.com/technical-seo/">technical SEO</a> team works on exactly these issues.</p>
<h2>FAQ</h2>
<p><strong>Should I use <code>el</code> or <code>el-CY</code> for my Greek pages?</strong>
If you only have one Greek version aimed at Cyprus, <code>el-CY</code> is accurate. Adding a plain <code>el</code> version as well helps Greek speakers elsewhere find a match.</p>
<p><strong>Is <code>.com.cy</code> better than <code>.com</code> for ranking in Cyprus?</strong>
A <code>.com.cy</code> domain gives a clear country signal, but a <code>.com</code> with good content, correct hreflang and local signals can rank well too.</p>
<p><strong>Can I automatically redirect visitors based on their location?</strong>
It is better not to force it. Suggest a language with a banner and let users choose, so crawlers can reach every version.</p>
<p><strong>Do Greek URLs hurt SEO?</strong>
Not by themselves. The main issues are readability and consistency once the URLs are encoded, which is why many sites transliterate.</p>
<p><strong>How do I check that hreflang works?</strong>
Inspect the rendered HTML, crawl the site with an SEO crawler that reports hreflang errors, and review Search Console for indexing problems.</p>
<p>Miltos George, Growth-onomics, Cyprus</p>
]]></content:encoded></item><item><title><![CDATA[AI search visibility checklist for fintech and trading sites]]></title><description><![CDATA[Free checklist for fintech and trading sites: AI crawler access, robots.txt for GPTBot and PerplexityBot, schema, entities, llms.txt, AI mentions.]]></description><link>https://growth-onomics.hashnode.dev/ai-search-visibility-checklist-fintech-trading-sites</link><guid isPermaLink="true">https://growth-onomics.hashnode.dev/ai-search-visibility-checklist-fintech-trading-sites</guid><category><![CDATA[fintech]]></category><category><![CDATA[marketing]]></category><category><![CDATA[SEO]]></category><category><![CDATA[AI]]></category><dc:creator><![CDATA[Growth-onomics]]></dc:creator><pubDate>Thu, 19 Nov 2026 17:41:50 GMT</pubDate><content:encoded><![CDATA[<p><strong>Miltos George, Growth-onomics</strong><br />Cyprus<br />19 November 2026</p>
<p>Fintech and trading sites are often easy for a person to use and hard for an assistant to quote. The licence sits in a PDF, fees sit in an image, <code>robots.txt</code> blocks a search crawler by accident, or public profiles describe the firm differently.</p>
<p>This free checklist covers crawl access, <code>robots.txt</code>, schema, entity consistency, <code>llms.txt</code>, compliance-safe claims, and a way to measure whether ChatGPT, Perplexity or Google’s AI experiences mention you. It will not make a model recommend you on demand, and is not legal advice.</p>
<h2>1. Confirm that AI search crawlers can read the pages that matter</h2>
<p>If a crawler cannot fetch a page as text, you are not in the source pool. Check the rendered HTML of your product, pricing, help and licence pages, not only the design after login or a cookie banner. Use Search Console’s URL Inspection tool for Google, and a raw fetch or log check for other bots. Watch for cookie walls, CDN challenges, facts that exist only in PDFs or images, and noindex or canonical tags that point a definitive page to a thinner one. Content must still be crawlable for ordinary search if you want it in AI features in Search. There is no special “AI schema” that replaces that.</p>
<p><strong>Pass:</strong> a crawler can retrieve the licence, fees, eligibility and product description as HTML text.</p>
<h2>2. Review robots.txt for GPTBot, PerplexityBot and Google-Extended</h2>
<p>Treat each user-agent as a separate decision. Training, indexing and a live user fetch are not the same thing.</p>
<table>
<thead>
<tr>
<th>User-agent</th>
<th>Typical use</th>
<th>If you block it</th>
</tr>
</thead>
<tbody><tr>
<td><code>GPTBot</code></td>
<td>OpenAI training</td>
<td>Less likely to be used to train OpenAI models. Not the same as ChatGPT search.</td>
</tr>
<tr>
<td><code>OAI-SearchBot</code></td>
<td>OpenAI search index for ChatGPT search</td>
<td>Harder for ChatGPT search to surface your pages.</td>
</tr>
<tr>
<td><code>ChatGPT-User</code></td>
<td>Live fetch when a person asks ChatGPT to open a page</td>
<td>A user-initiated visit may fail.</td>
</tr>
<tr>
<td><code>PerplexityBot</code></td>
<td>Perplexity search index</td>
<td>Harder to appear in Perplexity results.</td>
</tr>
<tr>
<td><code>Perplexity-User</code></td>
<td>Live fetch when a Perplexity user needs your page</td>
<td>The live citation path can fail.</td>
</tr>
<tr>
<td><code>Google-Extended</code></td>
<td>Certain Gemini / generative AI uses outside Search</td>
<td>Does <strong>not</strong> remove you from Google Search or AI features in Search.</td>
</tr>
<tr>
<td><code>Googlebot</code></td>
<td>Google Search, including AI features that use the Search index</td>
<td>Do not block this if you want to remain in Google Search.</td>
</tr>
</tbody></table>
<p>A common mistake is to paste a “block AI bots” snippet that also blocks <code>OAI-SearchBot</code> or <code>PerplexityBot</code>. Another is to assume that blocking <code>Google-Extended</code> opts you out of AI Overviews or AI Mode. It does not. Read the live <code>/robots.txt</code> file, keep staging and account URLs closed, and align any CDN or WAF rule with that file. An allow in <code>robots.txt</code> is useless if the firewall returns 403.</p>
<p><strong>Pass:</strong> legal and engineering can explain, for each bot, whether it is allowed and why.</p>
<h2>3. Use schema only where it repeats visible text</h2>
<p>Structured data helps a machine identify the organisation and the questions you answer. It should never invent a claim that is not on the page. Useful types are <code>Organization</code> and, where it truly fits, <code>FinancialService</code>; <code>FAQPage</code> only for questions visible on that URL; and offer markup only if price and availability are on the page and still accurate. Use the legal name, licence number, regulator and canonical URL. Do not mark testimonials or “best broker” copy as factual offer data.</p>
<p><strong>Pass:</strong> every schema property can be found, in the same words or figures, on the page itself.</p>
<h2>4. Make the entity consistent everywhere an assistant might look</h2>
<p>Assistants also read directories, official registers, app stores, LinkedIn and comparison sites. Keep a short source-of-truth record identical everywhere: legal and trading names; licence number, regulator and jurisdiction; registered office; what the product is and is not; countries served and not served; official website and support contact. If the firm is authorised in one jurisdiction and uses a partner in another, say so plainly. Inconsistent data is a common reason an assistant merges you with a similarly named brand, or declines to name you.</p>
<p><strong>Pass:</strong> a reviewer can copy those facts from three public sources and get the same answer.</p>
<h2>5. Treat llms.txt as an optional pointer, not a ranking lever</h2>
<p>An <code>llms.txt</code> file, usually at <code>/llms.txt</code>, is a proposed convention for pointing models at a few definitive pages, such as your licence page, fee tables and policies. It is optional. Major answer engines have not committed to using it when they choose sources. If you add one, keep the links public and already crawlable, and do not put stronger claims in the file than the site itself makes. It is not a substitute for <code>robots.txt</code> or a readable page.</p>
<p><strong>Pass:</strong> if the file exists, every link works and the wording matches the live site.</p>
<h2>6. Write claims that remain true when a model lifts one sentence</h2>
<p>A cited sentence will be read without the footnote beneath it. Avoid “best”, “number one”, “guaranteed”, “risk-free”, or undated performance claims. Avoid implying that a licence in one country covers every country, or promising approval and withdrawal speeds you cannot stand behind. Prefer sentences a compliance officer would still accept in isolation, and date any fee or limit table. Have compliance review the URLs you most want cited, not only the ads.</p>
<p><strong>Pass:</strong> a sentence lifted from the top of each key page is still accurate.</p>
<h2>7. Measure mentions, citations and mistakes on a fixed question set</h2>
<p>Build a stable list of 20 to 40 prompts from sales calls, support tickets and search queries, covering licensing, eligibility, fees and comparisons. Run that list, in the same order, through ChatGPT, Perplexity and the Google AI experiences that matter to your buyers. Score each run: named, cited, mentioned incorrectly, or absent. Repeat monthly. Watch referrals from <a href="http://chatgpt.com">chatgpt.com</a> and <a href="http://perplexity.ai">perplexity.ai</a>, and confirm in logs that allowed search crawlers are fetching the important URLs. When a description is wrong, fix the source and re-test the same prompt.</p>
<p>Growth-onomics documents how this kind of <a href="https://growth-onomics.com/answer-engine-optimization-services/">answer engine optimization work</a> is scoped, including a visibility baseline, access checks, entity consistency and citation tracking. Run the checklist first.</p>
<h2>Copy-paste checklist</h2>
<ul>
<li><input disabled type="checkbox" /> Product, pricing, help and licence pages return 200 and are readable as HTML text</li>
<li><input disabled type="checkbox" /> No cookie wall or challenge page hides those facts</li>
<li><input disabled type="checkbox" /> <code>robots.txt</code> allows or blocks <code>GPTBot</code>, <code>OAI-SearchBot</code>, <code>ChatGPT-User</code>, <code>PerplexityBot</code>, <code>Perplexity-User</code>, <code>Google-Extended</code> and <code>Googlebot</code> on purpose</li>
<li><input disabled type="checkbox" /> CDN / WAF rules match that <code>robots.txt</code> policy</li>
<li><input disabled type="checkbox" /> Schema matches visible text and uses only types the page supports</li>
<li><input disabled type="checkbox" /> Legal name, licence, regulator and availability match on-site and on major public profiles</li>
<li><input disabled type="checkbox" /> Fees, limits and eligibility sit in a dated HTML table, not only in a PDF</li>
<li><input disabled type="checkbox" /> Opening sentences can be quoted without creating a stronger claim than intended</li>
<li><input disabled type="checkbox" /> <code>llms.txt</code>, if used, points only to live, public, consistent pages</li>
<li><input disabled type="checkbox" /> A fixed prompt set is scored monthly for mentions, citations and errors</li>
<li><input disabled type="checkbox" /> Referrals and crawler logs are reviewed alongside the prompt scores</li>
</ul>
<h2>FAQ</h2>
<h3>Should we allow every AI bot?</h3>
<p>No. Decide by use. Many firms allow search and user-fetch bots so they can be cited, and decide separately on training bots. Avoid generic block-AI snippets.</p>
<h3>Will schema get us into ChatGPT or Perplexity?</h3>
<p>Not on its own. Schema helps machines parse facts you have already published. Weak content or a blocked crawler will still keep you out.</p>
<h3>Is llms.txt required?</h3>
<p>No. It is a low-cost extra. Crawl access, clear answers and consistent entity facts matter more.</p>
<h3>How do we handle multiple legal entities or brands?</h3>
<p>Give each entity its own name, licence and product scope. Do not let a group brand borrow a subsidiary’s licence.</p>
<h3>What is a good first monthly metric?</h3>
<p>The share of your fixed prompt set in which you are named or cited, plus a count of factual errors. Referral traffic is useful, but understates off-site shortlisting.</p>
<hr />
<p>Miltos George, Growth-onomics<br />Cyprus</p>
]]></content:encoded></item></channel></rss>