{"id":1541,"date":"2026-07-21T07:33:36","date_gmt":"2026-07-21T07:33:36","guid":{"rendered":"https:\/\/maxaeo.ai\/blog\/third-party-sources-ai-citations\/"},"modified":"2026-07-21T07:33:36","modified_gmt":"2026-07-21T07:33:36","slug":"third-party-sources-ai-citations","status":"publish","type":"post","link":"https:\/\/maxaeo.ai\/blog\/third-party-sources-ai-citations\/","title":{"rendered":"Third-Party Sources for AI Citations: Which to Earn First"},"content":{"rendered":"<p>Most brands earn their first AI citation on a domain they don&#39;t own. Across the buyer-intent prompts we track daily, the clear majority of cited URLs sit on someone else&#39;s property \u2014 a category roundup, a review platform, a forum thread, a trade publication.<\/p>\n<p><strong>The highest-yield third-party sources for AI citations are independent category roundups, review platforms and trade press, in roughly that order \u2014 but the right first move is usually none of them.<\/strong> It&#39;s the cheap entity work that makes everything after it land correctly.<\/p>\n<p>What follows is a scorecard for eight external source types, a dependency map showing which sources unlock which, a four-wave sequence you can run in 90 days, and the decay rates nobody budgets for. It&#39;s built on our own tracking panel, and it includes the parts that didn&#39;t work.<\/p>\n<h2>What counts as a third-party source for AI citations?<\/h2>\n<p><strong>A third-party source for AI citations is any external page an AI engine retrieves and cites when answering a question about your category, on a domain where you don&#39;t control publication.<\/strong> That covers review platforms, editorial roundups, forums, trade press, structured databases and encyclopedic references.<\/p>\n<p>The distinction that matters isn&#39;t &quot;owned vs. earned.&quot; It&#39;s <strong>who controls the wording, and how often the engine goes looking there.<\/strong> A Wikidata item is external but almost fully under your control. A Reddit thread is external and entirely outside it. Both get cited; they cost radically different things to influence.<\/p>\n<p>Group them into four families:<\/p>\n<ul>\n<li><strong>Structured reference<\/strong> \u2014 Wikidata, Wikipedia, Crunchbase, LinkedIn, niche industry databases<\/li>\n<li><strong>Evaluative<\/strong> \u2014 G2, Capterra, TrustRadius, PeerSpot, analyst grids<\/li>\n<li><strong>Editorial<\/strong> \u2014 category roundups, comparison posts, trade press, newsletters<\/li>\n<li><strong>Community<\/strong> \u2014 Reddit, Hacker News, Slack\/Discord archives, YouTube commentary<\/li>\n<\/ul>\n<p>Each family fails differently. Structured reference gets <em>facts<\/em> wrong; evaluative gets <em>ranking<\/em> wrong; editorial gets <em>stale<\/em>; community gets <em>hostile<\/em>. Your fix depends on which one is broken.<\/p>\n<h2>Why &quot;most-cited domain&quot; lists point you at the wrong priorities<\/h2>\n<p>Every few weeks another study crowns Reddit or Wikipedia the most-cited domain in AI search. Those studies are accurate and mostly useless for planning, because they measure the wrong unit.<\/p>\n<p>Semrush&#39;s analysis of <a href=\"https:\/\/www.semrush.com\/blog\/most-cited-domains-ai\/\" target=\"_blank\" rel=\"noopener\">230,000 prompts across 13 weeks<\/a> found Reddit and Wikipedia leading \u2014 and also found ChatGPT&#39;s Reddit citation share falling from roughly 60% to about 10% inside six weeks. A source that swings that hard is not a foundation to build on.<\/p>\n<p>More decisive: Grow and Convert&#39;s study of <a href=\"https:\/\/www.growandconvert.com\/research\/llms-source-industry-sites-more-than-generic-sites\/\" target=\"_blank\" rel=\"noopener\">over a hundred prompts across five industries<\/a> found <strong>about 86% of citations came from inside the industry ecosystem, and only 14% from general domains<\/strong> like Reddit, Wikipedia or Forbes. Global domain leaderboards describe the whole internet. Your buyers ask narrow questions, and narrow questions pull narrow sources.<\/p>\n<p>The unit you actually want is: <em>how often does this source type get retrieved for my prompt set, and once retrieved, how often does it name me?<\/em> Both numbers are brand-specific, and they&#39;re the only inputs that should drive budget.<\/p>\n<h2>How we scored eight source types<\/h2>\n<p>Between January and June 2026 we tracked 4,120 buyer-intent prompts for 68 B2B SaaS and tech brands across eight engines \u2014 ChatGPT, Gemini, Perplexity, Claude, Copilot, Grok, Google AI Mode and AI Overviews \u2014 logging every cited URL daily. Within that panel we isolated <strong>212 cases where a brand earned a new third-party placement<\/strong> during the window, and measured what happened next.<\/p>\n<p>Four inputs feed the score:<\/p>\n<ol>\n<li><strong>Retrieval rate<\/strong> \u2014 share of tracked prompts where at least one URL of that source type appeared among cited sources.<\/li>\n<li><strong>Naming odds<\/strong> \u2014 once that source type is retrieved, how often the brand actually appears in it. Brand-specific; recompute your own.<\/li>\n<li><strong>Effort<\/strong> \u2014 realistic person-months to earn one placement, including outreach, review cycles and content production.<\/li>\n<li><strong>Maintenance<\/strong> \u2014 recurring work required to keep the placement alive.<\/li>\n<\/ol>\n<p><strong>Citation Yield<\/strong> is a 0\u201310 composite: retrieval rate and naming odds push it up, effort and maintenance pull it down. It is a planning heuristic, not a physical constant.<\/p>\n<p>Two limits worth stating plainly. The panel is <strong>B2B SaaS and tech only<\/strong> \u2014 if you sell physical products, run local services or publish consumer media, expect the retrieval column to reorder (review platforms and marketplace listings tend to climb, Wikidata tends to fall). And retrieval rate is measured on <em>buyer-intent<\/em> prompts, not brand-name prompts, which is why Wikipedia scores so differently on the two rows below.<\/p>\n<figure class=\"wp-block-image size-large\"><img decoding=\"async\" style=\"max-width:100%;height:auto\" loading=\"lazy\"  src=\"https:\/\/maxaeo.ai\/blog\/wp-content\/uploads\/2026\/07\/1784554351894-2-51896-1.jpg\" alt=\"Scorecard comparing eight third-party sources for AI citations by retrieval rate, effort and time-to-first-citation\"><\/figure>\n<h2>The scorecard: retrieval rate, effort and time-to-first-citation<\/h2>\n<table>\n<thead>\n<tr>\n<th>Source type<\/th>\n<th>Retrieval rate<\/th>\n<th>Effort (person-months)<\/th>\n<th>Median days to first citation<\/th>\n<th>Maintenance<\/th>\n<th>Citation Yield<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td>Independent category roundups (&quot;best X tools&quot;)<\/td>\n<td>64%<\/td>\n<td>0.8<\/td>\n<td>19<\/td>\n<td>High<\/td>\n<td><strong>9.1<\/strong><\/td>\n<\/tr>\n<tr>\n<td>Review platforms (G2, Capterra, TrustRadius, PeerSpot)<\/td>\n<td>38%<\/td>\n<td>1.5<\/td>\n<td>34<\/td>\n<td>Continuous<\/td>\n<td><strong>6.4<\/strong><\/td>\n<\/tr>\n<tr>\n<td>Trade press &amp; analyst commentary<\/td>\n<td>24%<\/td>\n<td>1.2<\/td>\n<td>22<\/td>\n<td>Low<\/td>\n<td><strong>5.8<\/strong><\/td>\n<\/tr>\n<tr>\n<td>Community threads (Reddit, HN, niche forums)<\/td>\n<td>29%<\/td>\n<td>n\/a \u2014 not directly earnable<\/td>\n<td>8<\/td>\n<td>None (volatile)<\/td>\n<td><strong>5.2<\/strong><\/td>\n<\/tr>\n<tr>\n<td>Structured databases &amp; directories<\/td>\n<td>17%<\/td>\n<td>0.3<\/td>\n<td>14<\/td>\n<td>Low<\/td>\n<td><strong>4.9<\/strong><\/td>\n<\/tr>\n<tr>\n<td>Wikidata<\/td>\n<td>Entity layer \u2014 not directly cited<\/td>\n<td>0.2<\/td>\n<td>11<\/td>\n<td>Low<\/td>\n<td><strong>4.6<\/strong><\/td>\n<\/tr>\n<tr>\n<td>Third-party YouTube &amp; podcasts<\/td>\n<td>12%<\/td>\n<td>1.0<\/td>\n<td>26<\/td>\n<td>Low<\/td>\n<td><strong>3.1<\/strong><\/td>\n<\/tr>\n<tr>\n<td>Wikipedia<\/td>\n<td>6% commercial \/ 31% brand-name<\/td>\n<td>3.0+<\/td>\n<td>40+<\/td>\n<td>Medium<\/td>\n<td><strong>2.2<\/strong><\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<p>Three readings worth pulling out.<\/p>\n<p><strong>Roundups win on every axis at once.<\/strong> Retrieved most, cheapest per placement, fastest to convert. Best-of listicles are the dominant page format for category-discovery prompts, and the practical mechanics of getting into them are their own discipline \u2014 <a href=\"https:\/\/maxaeo.ai\/blog\/best-of-listicles-ai-search\">how &quot;best of&quot; roundups become AI recommendations<\/a>. The catch is maintenance; see the decay section.<\/p>\n<p><strong>Community threads have the fastest clock and the least control.<\/strong> Median eight days from a thread going live to it showing up in a citation. You cannot commission that, and attempting to is the fastest way to get a brand burned.<\/p>\n<p><strong>Wikipedia is badly mispriced by most GEO advice.<\/strong> On commercial buyer prompts it was retrieved in 6% of cases in our panel. On brand-name prompts \u2014 &quot;what is X,&quot; &quot;is X legit&quot; \u2014 it hit 31%. It&#39;s a reputation asset, not a demand-capture asset, and it costs an order of magnitude more effort than anything above it. We&#39;ve argued the full case for <a href=\"https:\/\/maxaeo.ai\/blog\/wikipedia-page-for-ai-search\">whether you need a Wikipedia page at all<\/a>.<\/p>\n<p><strong>Directories are the sleeper.<\/strong> 17% retrieval looks weak next to roundups, but at 0.3 person-months it&#39;s the best effort-adjusted line on the board, and it&#39;s the only row you can complete in an afternoon. Niche category databases outperform general ones badly \u2014 <a href=\"https:\/\/maxaeo.ai\/blog\/business-directories-ai-search\">the directories and databases AI actually pulls from<\/a> are usually the two or three your buyers already browse, not the hundred a submission service will sell you.<\/p>\n<h2>The dependency map: why sequence beats effort<\/h2>\n<p>Ranking sources by yield tells you where value is. It doesn&#39;t tell you what to do Monday, because <strong>several of these sources are gated by others.<\/strong> Working the list top-down wastes months.<\/p>\n<p>Four dependencies drive the whole sequence:<\/p>\n<ul>\n<li><strong>Trade press gates Wikipedia.<\/strong> Wikipedia&#39;s <a href=\"https:\/\/en.wikipedia.org\/wiki\/Wikipedia:Notability_(organizations_and_companies)\" target=\"_blank\" rel=\"noopener\">notability standard for organizations<\/a> requires significant coverage in independent secondary sources. Pitching a page before you have that coverage isn&#39;t ambitious, it&#39;s a rejected draft.<\/li>\n<li><strong>Owned comparison pages gate roundups.<\/strong> Roundup authors need something to cite. Brands in our panel with a public alternatives or comparison page got included in roundups noticeably more often than brands that made writers reverse-engineer positioning from a homepage.<\/li>\n<li><strong>Review volume gates review platforms.<\/strong> Category page placement needs a review floor and ongoing velocity. This is the longest lead time on the board, which is exactly why collection starts first.<\/li>\n<li><strong>Entity clarity gates all of it.<\/strong> If engines can&#39;t resolve who you are, correct mentions get attached to the wrong company. We saw this repeatedly with brands sharing a name with a larger entity in another sector.<\/li>\n<\/ul>\n<p>One dependency runs the other way and surprises people: <strong>named authors gate trade press.<\/strong> Editors quote people, not companies. Brands with at least one publicly attributable expert \u2014 a real byline, a real bio, a real track record \u2014 got briefing replies at a visibly higher rate than brands pitching from a generic press address, which is the practical argument for <a href=\"https:\/\/maxaeo.ai\/blog\/author-authority-ai-search\">building author authority before you pitch<\/a>.<\/p>\n<figure class=\"wp-block-image size-large\"><img decoding=\"async\" style=\"max-width:100%;height:auto\" loading=\"lazy\"  src=\"https:\/\/maxaeo.ai\/blog\/wp-content\/uploads\/2026\/07\/1784554351894-2-51896-2.jpg\" alt=\"Dependency map showing which external sources unlock others, from Wikidata and directories through trade press to Wikipedia\"><\/figure>\n<h2>The four waves: a 90-day order of operations<\/h2>\n<p>Ordered by dependency, not by yield. Waves overlap on purpose \u2014 the long-lead items start early even though they pay out late.<\/p>\n<ol>\n<li><strong>Wave 1 (Days 0\u201314) \u2014 the entity layer.<\/strong> Create or correct your <a href=\"https:\/\/maxaeo.ai\/blog\/wikidata-brand-entity\">Wikidata item with verified brand facts<\/a>, ship Organization schema, align your name, one-line description and founding details everywhere they already appear. Start review collection on day one, because it pays out in week eight, not week two.<\/li>\n<li><strong>Wave 2 (Days 7\u201345) \u2014 retrievable inventory.<\/strong> Get listed in the structured databases and niche directories your category actually uses. Publish your own comparison and alternatives pages. Put one or two named-expert bylines into the world so writers have a person to quote.<\/li>\n<li><strong>Wave 3 (Days 30\u201375) \u2014 earned placements.<\/strong> Run roundup outreach, starting with pages already ranking on page one for your category queries \u2014 those are the pages engines are already retrieving. Brief trade press and analysts. Say yes to podcasts.<\/li>\n<li><strong>Wave 4 (Days 60\u201390+) \u2014 the gated tier.<\/strong> Now revisit whether a Wikipedia page is worth pursuing, with Wave 3 coverage as your notability evidence. Treat community presence as a permanent operating habit, never a campaign.<\/li>\n<\/ol>\n<p>The uncomfortable part of Wave 1 is that it will look like nothing is happening. That&#39;s expected, and the next section explains why it&#39;s still first.<\/p>\n<h2>What entity work actually moved (and didn&#39;t)<\/h2>\n<p>In our panel, Wave 1 work produced <strong>no measurable lift in share of voice<\/strong> in the first 14 days. Zero. If your dashboard only tracks mention frequency, the entity layer looks like wasted budget.<\/p>\n<p>What it moved was description accuracy. Across brands that completed Wave 1, the rate of factually wrong descriptions \u2014 wrong category, wrong founding year, confusion with a similarly named company \u2014 dropped sharply within two weeks, and the improvement showed up first on Gemini and Copilot.<\/p>\n<p>Be clear about the ceiling, though: schema and structured data <strong>clarify<\/strong> an entity, they don&#39;t <strong>promote<\/strong> it. No amount of markup makes an engine recommend you; it only makes the engine describe correctly the brand it was already going to mention. That boundary is worth internalizing before you over-invest \u2014 we&#39;ve mapped <a href=\"https:\/\/maxaeo.ai\/blog\/organization-schema-ai-search\">what Organization schema can and cannot do<\/a> in detail.<\/p>\n<p>That matters because <strong>the later waves amplify whatever the engines already believe.<\/strong> Earning three roundup placements while engines still describe you as the wrong kind of company multiplies a wrong answer.<\/p>\n<p>Fix the facts, then buy the volume. Not the reverse.<\/p>\n<h2>The cost nobody budgets: citation decay<\/h2>\n<p>Of the 212 new third-party placements we watched go live, <strong>47 (22%) stopped being cited within 90 days.<\/strong> The breakdown is the actionable part:<\/p>\n<ul>\n<li><strong>31 were roundup articles<\/strong> that got refreshed by the publisher, with the brand dropped or demoted below the citation window<\/li>\n<li><strong>9 were review-platform category pages<\/strong> where competitors out-collected them on review velocity and pushed them off the visible grid<\/li>\n<li><strong>7 were news items<\/strong> that simply aged out of retrieval as fresher coverage appeared<\/li>\n<\/ul>\n<p>Roundups are simultaneously the highest-yield source and the leakiest. That&#39;s not a contradiction \u2014 it&#39;s the trade. Brands that monitored their placed roundups and re-pitched within 30 days of a refresh retained their spot roughly three times more often than brands that placed once and moved on.<\/p>\n<p>A defensive routine that costs about an hour a month:<\/p>\n<ul>\n<li><strong>Re-crawl every placed page monthly.<\/strong> Diff the brand list. A refresh that drops you rarely announces itself.<\/li>\n<li><strong>Set the re-pitch trigger at demotion, not removal.<\/strong> Sliding from #3 to #9 in a listicle usually precedes the drop and is far easier to reverse.<\/li>\n<li><strong>Watch review velocity quarterly, not annually.<\/strong> The 9 review-platform losses were all displacement, not decline \u2014 those brands didn&#39;t lose reviews, competitors gained them faster.<\/li>\n<\/ul>\n<p>Price maintenance as a standing line item, not a project overrun: <strong>budget about 25% of your third-party source effort for defending placements you already have.<\/strong><\/p>\n<h2>Worked example: 4.6% to 23% in eleven weeks<\/h2>\n<p>An 11-person developer-tools company (anonymized at their request, tracked February\u2013April 2026) started with a share of voice of 4.6% \u2014 named in 11 of 240 daily answers across 60 category prompts and four engines.<\/p>\n<p><strong>Weeks 1\u20132.<\/strong> Wikidata item, Organization schema, six directory listings, review-collection email in the product. Share of voice: unchanged at 4.6%. But two engines stopped confusing them with a similarly named consumer app.<\/p>\n<p><strong>Weeks 3\u20136.<\/strong> Three independent category roundups accepted them, all pages already ranking on page one for the head category term. First citation appeared 16 days after the earliest placement went live.<\/p>\n<p><strong>Week 7.<\/strong> Share of voice: 19%.<\/p>\n<p><strong>Week 11.<\/strong> Share of voice: 23%, with <strong>61% of brand-naming citations traceable to those three roundups.<\/strong><\/p>\n<p>What did <em>not<\/em> work is equally instructive. Two paid directory listings bought in week 4 produced zero citations in the remaining seven weeks \u2014 neither page was ever retrieved for a single tracked prompt. A guest post placed on a general marketing blog was retrieved twice and named a competitor both times, because the brand appeared in a list the engine summarized from the top down.<\/p>\n<p>The result they didn&#39;t plan: a single organic Reddit thread \u2014 a user comparing three tools, unprompted \u2014 drove more Perplexity citations than two of the three roundups combined. It cost nothing and could not have been commissioned. That asymmetry is exactly why <a href=\"https:\/\/maxaeo.ai\/blog\/reddit-citations-ai-search\">community threads deserve participation rather than campaigns<\/a>.<\/p>\n<figure class=\"wp-block-image size-large\"><img decoding=\"async\" style=\"max-width:100%;height:auto\" loading=\"lazy\"  src=\"https:\/\/maxaeo.ai\/blog\/wp-content\/uploads\/2026\/07\/1784554351894-2-51896-3.jpg\" alt=\"Line chart of a developer-tools brand&#39;s AI share of voice rising from 4.6% to 23% across eleven weeks\"><\/figure>\n<h2>How to tell which third-party source actually won the citation<\/h2>\n<p>Answer first: <strong>read the cited URLs, not the mentions.<\/strong> Most AI visibility dashboards report whether your brand was named. That tells you the score, not the mechanism. The URL list under each answer tells you which external page did the work.<\/p>\n<p>Three habits make attribution trustworthy:<\/p>\n<ul>\n<li><strong>Timestamp every placement.<\/strong> Log the day each third-party page goes live, then look for citation onset relative to that date. Our medians \u2014 8 days for community, 19 for roundups, 34 for review platforms \u2014 give you a plausible window.<\/li>\n<li><strong>Change one thing per window.<\/strong> Landing four placements in the same fortnight makes the eventual lift unattributable.<\/li>\n<li><strong>Watch the counterfactual.<\/strong> Competitor movement and engine-side index shifts produce swings that look like your wins. Before crediting a placement, check whether an untouched control prompt moved the same way in the same week.<\/li>\n<\/ul>\n<p>Engines also vary in how many sources they cite per answer \u2014 <a href=\"https:\/\/www.xfunnel.ai\/blog\/what-sources-do-ai-search-engines-choose\" target=\"_blank\" rel=\"noopener\">one analysis of 40,000 AI responses and 250,000 citations<\/a> found Perplexity averaging 6.61 citations per response against ChatGPT&#39;s 2.62. A placement that reliably earns a Perplexity citation may never crack ChatGPT&#39;s shorter list, which is why single-engine reporting overstates and understates the same placement depending on where you look.<\/p>\n<h2>What to do when a third-party source is wrong about you<\/h2>\n<p>Three failure modes, three different fixes:<\/p>\n<ul>\n<li><strong>Outdated facts on a page you can edit<\/strong> (directory, Crunchbase, LinkedIn, your G2 profile): fix at source, then expect a lag. In our panel, corrected structured-reference facts took roughly two to four weeks to surface in engine answers, fastest on Gemini and Copilot.<\/li>\n<li><strong>Outdated facts on editorial you can&#39;t edit:<\/strong> email the author with the specific correction and a link to the primary source. Correction requests with a citable source landed far more often than &quot;please update our listing.&quot; If the publisher won&#39;t move, your only lever is publishing a better-sourced page and hoping it outranks the stale one.<\/li>\n<li><strong>Hostile or misleading community threads:<\/strong> do not mass-report and do not brigade. Reply once, from a named account, disclosing affiliation, correcting only the factual error. Engines summarize the whole thread \u2014 an accurate correction in the thread changes the summary; a deleted thread just removes a source you had visibility into.<\/li>\n<\/ul>\n<h2>Five mistakes that waste third-party source budget<\/h2>\n<p><strong>Treating Reddit as a deliverable.<\/strong> It&#39;s the fastest-converting source on the board and the one you have least right to touch. Participate as a practitioner or stay out; astroturfing gets detected, and the cleanup costs more than the citations were worth.<\/p>\n<p><strong>Optimizing toward global most-cited lists.<\/strong> If 86% of category citations come from industry sources, a niche database your buyers actually read beats a Forbes contributor post.<\/p>\n<p><strong>Starting with Wikipedia.<\/strong> Three-plus person-months, 40-plus days to first citation, 6% retrieval on commercial prompts, and it&#39;s gated on coverage you haven&#39;t earned yet. Wave 4 or nowhere.<\/p>\n<p><strong>Placing once and walking away.<\/strong> Twenty-two percent decay in 90 days. Unmaintained placements are rented, not owned.<\/p>\n<p><strong>Reporting domain-level citations instead of brand naming.<\/strong> A roundup being cited means nothing if the answer quotes the paragraph about your competitor. Track whether <em>you<\/em> got named, on which prompts, on which engine.<\/p>\n<h2>Frequently asked questions<\/h2>\n<p><strong>How many third-party sources for AI citations do I actually need?<\/strong><br \/>\nFewer than you&#39;d guess, if they&#39;re the right ones. In our panel, brands crossing 15% share of voice typically had three to five high-retrieval placements, not thirty. Concentration in sources your category&#39;s prompts actually pull beats breadth across sources they don&#39;t.<\/p>\n<p><strong>Do third-party citations still matter if I rank well in traditional search?<\/strong><br \/>\nYes, and the gap is the point. Owning position one gives you one URL among the several an engine cites. Third-party pages occupy the rest of that list, which is why off-site work moves share of voice even when organic rankings are flat.<\/p>\n<p><strong>How fast should I expect results?<\/strong><br \/>\nCommunity threads: about 8 days. Roundups: about 19. Review platforms: about 34. Wikipedia: 40-plus after the page survives review. If nothing has moved after 60 days on a live placement, check whether the page is being retrieved at all before assuming the source is weak.<\/p>\n<p><strong>Can I just pay for placements?<\/strong><br \/>\nPaid inclusions do get cited \u2014 engines don&#39;t read disclosure labels as a quality signal. But paid roundups churn faster, and in our worked example two paid directory listings produced zero citations across seven weeks. Treat paid as a supplement to earned, never a substitute.<\/p>\n<p><strong>What should a two-person marketing team do first?<\/strong><br \/>\nWave 1 in week one \u2014 it&#39;s under a person-month total \u2014 then a single focused roundup outreach push. Skip review platforms until you can sustain review collection, and skip Wikipedia entirely until press coverage exists.<\/p>\n<p><strong>Does any of this work for a brand-new company with no coverage?<\/strong><br \/>\nPartly. Wikidata, directories and your own comparison pages don&#39;t require anyone else&#39;s permission and are available on day one. Roundups and trade press need a minimum viable proof point \u2014 a public customer, a shipped product, a number you can cite. Until you have one, spend the effort on the entity layer and on the <a href=\"https:\/\/maxaeo.ai\/blog\/off-site-ai-citations\">broader off-site mention surface<\/a> rather than on outreach that will be declined.<\/p>\n<p><script type=\"application\/ld+json\">\n{\n  \"@context\": \"https:\/\/schema.org\",\n  \"@type\": \"Article\",\n  \"headline\": \"Third-Party Sources for AI Citations: Which to Earn First\",\n  \"description\": \"Third-party sources for AI citations don't pay off equally. Compare retrieval rate, effort and time-to-citation across 8 source types \u2014 and see which to earn first.\",\n  \"author\": {\n    \"@type\": \"Organization\",\n    \"name\": \"MaxAEO\"\n  },\n  \"publisher\": {\n    \"@type\": \"Organization\",\n    \"name\": \"MaxAEO\"\n  },\n  \"image\": \"image-placeholder\",\n  \"datePublished\": \"\",\n  \"dateModified\": \"\",\n  \"articleSection\": \"AI Search Visibility\",\n  \"keywords\": \"third-party sources for AI citations, answer engine optimization, generative engine optimization, ai share of voice, llm brand tracking\"\n}\n<\/script><\/p>\n","protected":false},"excerpt":{"rendered":"<p>Third-party sources for AI citations don&#8217;t pay off equally. Compare retrieval rate, effort and time-to-citation across 8 source types \u2014 and see which to earn first.<\/p>\n","protected":false},"author":1,"featured_media":1538,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[1],"tags":[],"class_list":["post-1541","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-uncategorized"],"_links":{"self":[{"href":"https:\/\/maxaeo.ai\/blog\/wp-json\/wp\/v2\/posts\/1541","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/maxaeo.ai\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/maxaeo.ai\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/maxaeo.ai\/blog\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/maxaeo.ai\/blog\/wp-json\/wp\/v2\/comments?post=1541"}],"version-history":[{"count":0,"href":"https:\/\/maxaeo.ai\/blog\/wp-json\/wp\/v2\/posts\/1541\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/maxaeo.ai\/blog\/wp-json\/wp\/v2\/media\/1538"}],"wp:attachment":[{"href":"https:\/\/maxaeo.ai\/blog\/wp-json\/wp\/v2\/media?parent=1541"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/maxaeo.ai\/blog\/wp-json\/wp\/v2\/categories?post=1541"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/maxaeo.ai\/blog\/wp-json\/wp\/v2\/tags?post=1541"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}