<?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 OMNIUS on Medium]]></title>
        <description><![CDATA[Stories by OMNIUS on Medium]]></description>
        <link>https://medium.com/@omniusSo?source=rss-571819055169------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*axGnRi5gem0ip8fRh9V6Iw.png</url>
            <title>Stories by OMNIUS on Medium</title>
            <link>https://medium.com/@omniusSo?source=rss-571819055169------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Tue, 17 Nov 2026 02:48:02 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@omniusSo/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[How LLMs actually decide what to cite - and what to ignore]]></title>
            <link>https://medium.com/@omniusSo/how-llms-actually-decide-what-to-cite-and-what-to-ignore-14a1cfc71dc7?source=rss-571819055169------2</link>
            <guid isPermaLink="false">https://medium.com/p/14a1cfc71dc7</guid>
            <category><![CDATA[aeo]]></category>
            <category><![CDATA[llm]]></category>
            <category><![CDATA[geo]]></category>
            <dc:creator><![CDATA[OMNIUS]]></dc:creator>
            <pubDate>Tue, 11 Aug 2026 11:00:41 GMT</pubDate>
            <atom:updated>2026-08-11T11:00:41.427Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*7wJuImb-S4EHeahb9f3-dA.jpeg" /></figure><p>Updated: August 2026</p><p>Citation is the SEO metric that now matters most for organic visibility. When ChatGPT, Perplexity, Google AI Mode, or Gemini generate an answer about a category, they select a small set of sources to reference. The brands that aren’t in that set are invisible to the buyer at the moment of decision.</p><p>The mechanics are knowable. The retrieval process is documented. The signals that models weight are trackable. The failure modes are consistent across platforms. This article covers what LLMs actually do when they select citations, what causes content to be skipped, and the operational choices that move a page from ignored to cited.</p><h3>How LLMs source citations in the first place</h3><p>LLMs do not pull answers from a static index. When a user asks a live question, the model triggers a retrieval process (called RAG, or Retrieval-Augmented Generation) that fetches sources from the open web in real time. The citation decision is made in that retrieval step, not at the training-data level.</p><p>Retrieval-Augmented Generation combines the reasoning capability of an LLM with a live search of external sources. The model rewrites the user’s query into one or more search-friendly variations, sends those queries to a retrieval system, ranks the returned candidates, and selects a small subset to reference when generating the answer.</p><p>This changes the nature of the visibility problem. In traditional SEO, the question is whether the page ranks against a keyword. In LLM citation, the question is whether the page appears in the retrieval set that the model considers before generating the answer. A page can rank first on Google and still not be in the RAG retrieval set. The correlation between Google top-10 rankings and ChatGPT citations sits at approximately 12% overlap (Omnius AI citation tracking research, 2026).</p><p>The retrieval set typically contains between 5 and 15 candidate sources, of which the model may reference 1 to 5 in the final answer. Which sources enter that set, and which are cited from it, depends on the signals covered in the next section.</p><figure><img alt="Citation tracking across ChatGPT, Perplexity, and Google AI Overviews, from Atomic AGI." src="https://cdn-images-1.medium.com/max/1024/1*dkcas1t7mJ1FSSc5cZBzZA.png" /><figcaption><em>Citation tracking across ChatGPT, Perplexity, and Google AI Overviews, from Atomic AGI.</em></figcaption></figure><h3>What signals LLMs weight when selecting sources</h3><p>LLM citation selection weights five signals consistently across ChatGPT, Perplexity, Google AI Mode, and Gemini: authority on the specific topic, structural clarity for extraction, freshness relative to query intent, source diversity across the answer, and entity clarity for verification.</p><p>Authority is the composite signal LLMs use to assess whether a source is credible on a specific topic. It draws from traditional SEO signals like backlinks and domain age, but it also incorporates off-domain signals: mentions in third-party publications, presence on review platforms, coverage in the sources the model was trained on. The pattern is consistent across audits: authority is domain-specific, not domain-general. A page from a specialist source often outranks a page from a broader site when the query matches the specialist’s remit.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*d4byt1wA6NpjZl4d2hcwGw.png" /><figcaption>ChatGPT citation share by top 10 sources — Wikipedia dominates at 47.9%, followed by Reddit at 11.3%.</figcaption></figure><p>Structural clarity determines whether the model can extract a usable answer from the page. Pages structured with direct answers, clear subheadings, FAQ blocks, and comparison tables are extracted more reliably than pages that require the model to synthesize meaning across paragraphs. Reference-grade content with proper schema and chunked structure receives 3 to 5x more AI citations than comparable pages without it (OtterlyAI, 2026). Tables specifically are cited 2.5x more often than equivalent paragraph content (Princeton GEO Study).</p><p>Freshness varies by query intent. Time-sensitive queries (news, product updates, current pricing) weight recency heavily. Evergreen queries (definitions, frameworks, how-to content) weight recency less. LLMs infer the freshness requirement from the query itself, which means “updated on” dates and version notes matter more for some content types than others.</p><p>Source diversity is a defensive signal. LLMs avoid concentrating citations on a single source to reduce the risk of amplifying a wrong answer. This is why a page can be authoritative and structurally clear and still not be cited if the model has already selected another source from the same domain.</p><p>Entity clarity is the ability of the model to verify who is behind the claim. Pages with Person schema, Organization schema, and clear author credentials are easier to verify than pages that surface claims anonymously. Verification is what turns a candidate source into a cited one.</p><h3>Why some content gets ignored despite ranking on Google</h3><p>Content is ignored by LLM citation when it fails one of three tests: the model cannot access it, cannot parse it, or cannot verify it. Ranking on Google addresses none of these directly.</p><p>The first failure is access. AI crawlers operate on different rules than Googlebot. Many sites block AI crawlers explicitly through robots.txt, or implicitly through JavaScript rendering, CDN rules, or authentication walls. 73% of brand sites have technical barriers that block AI crawler access entirely (OtterlyAI, 2026). A page that can’t be crawled by GPTBot, ClaudeBot, or PerplexityBot cannot be considered for citation regardless of how well it ranks in Google.</p><p>The second failure is parsing. Content that reads as authoritative to a human but requires the model to synthesize meaning across paragraphs is skipped for content that presents the answer directly. This is particularly common in regulated industries. Fintech and healthcare content often hedges every claim with regulatory qualifiers, which is the same language answer engines read as low-confidence and skip.</p><p>The third failure is verification. Pages that make claims without linking to verifiable authorities, without author attribution, without organization schema, and without third-party validation are treated as low-trust sources. LLMs weight unverified claims lower than the same claim on a page with proper trust architecture, even if both pages rank identically in Google.</p><h3>What structural choices move a page from ignored to cited</h3><p>The operational shifts fall into four categories: fix the access layer, restructure content for extraction, add entity verification, and build off-domain citation surfaces. Each addresses one of the failure modes above and moves the page into the retrieval set that LLMs actually consider.</p><p>Access is the first fix because it gates everything else. Review robots.txt to confirm AI crawlers (GPTBot, ClaudeBot, PerplexityBot, GoogleOther) are allowed. Audit JavaScript rendering to confirm content is server-rendered or pre-rendered rather than client-only. Remove authentication walls or CDN rules that block legitimate crawlers. A page that becomes crawlable often enters the retrieval set within weeks.</p><p>Structural extraction requires restructuring content around the queries LLMs actually receive. Every section should open with a direct 40 to 60 word answer to the query it targets, before any context. TL;DR blocks at the top of every long-form page make the definitive claim easy to extract. Tables, FAQ blocks, and comparison structures are extracted more reliably than the same information delivered as prose.</p><p>Entity verification requires encoding trust signals in schema. Organization schema, Person schema with role and credentials, reference links to regulatory bodies and industry sources with verifiable identifiers. The compliance badges and press logos that reassure human buyers on a homepage don’t appear in the content blocks LLMs extract. Moving these references into the page body where the model reads is what makes them count.</p><figure><img alt="Schema markup implementation example — Organization and Person schema encoded for machine extraction." src="https://cdn-images-1.medium.com/max/1024/1*5eB7mLYxsLmbLkaxXz9RJw.png" /><figcaption>Schema markup implementation example — Organization and Person schema encoded for machine extraction.</figcaption></figure><p>Off-domain citation surfaces are the highest-leverage move for brands starting from a low citation base. The Atomic AGI Q1 2026 report shows the top citation sources AI engines pull from are Reddit (14.3%), LinkedIn (14.05%), and YouTube (6.38%). A brand present on those surfaces enters retrieval sets it would never enter through its own domain alone.</p><p>At one of our fintech engagements, this combination produced a 164% increase in AI search traffic over six months alongside a 60% increase in weekly signups. Both curves moved on the same timeline because both were driven by the same structural changes.</p><h3>Content LLMs ignore vs content LLMs cite</h3><figure><img alt="Content LLMs ignore vs content LLMs cite" src="https://cdn-images-1.medium.com/max/892/1*p1lEPoRpuaQXNvXc3j3Weg.png" /></figure><p>LLM citation is not a mystery once the mechanics are understood. The retrieval step decides which sources enter the model’s consideration. The ranking step decides which of those sources get referenced in the final answer. Both steps run on signals that are trackable, and content can be structured for them or against them.</p><p>For any brand that ranks on Google but doesn’t appear in AI-generated answers, the fix is operational, not editorial. The content isn’t wrong. It’s encoded for a retrieval process it was never designed for.</p><h3>References</h3><ul><li>Omnius — RAG in SEO: What It Is, How It Works &amp; Key Benefits — <a href="https://www.omnius.so/blog/what-is-retrieval-augmented-generation">https://www.omnius.so/blog/what-is-retrieval-augmented-generation</a></li><li>Omnius — How to Get Cited by AI: 5 Strategies to Optimize for LLMs — <a href="https://www.omnius.so/blog/how-to-get-cited-by-ai">https://www.omnius.so/blog/how-to-get-cited-by-ai</a></li><li>Omnius — LLMs.txt for SEO: Importance, Structure &amp; How to Build It — <a href="https://www.omnius.so/blog/llms-txt-file">https://www.omnius.so/blog/llms-txt-file</a></li><li>Matija Golubovic — AI citation tracking: the complete guide for SEO and marketing teams, Atomic AGI, April 2026 — <a href="https://www.atomicagi.com/blog/ai-citation-tracking-the-complete-guide-for-seo-and-marketing-teams">https://www.atomicagi.com/blog/ai-citation-tracking-the-complete-guide-for-seo-and-marketing-teams</a></li><li>Atomic AGI — Q1 2026 AI Search Quarter Report — <a href="https://app.atomicagi.com/public/reports/q1-2026-ai-search-quarter-report">https://app.atomicagi.com/public/reports/q1-2026-ai-search-quarter-report</a></li><li>OtterlyAI — The AI Citation Economy: What 1+ Million Data Points Reveal About Visibility in 2026 — <a href="https://otterly.ai/blog/the-ai-citations-report-2026/">https://otterly.ai/blog/the-ai-citations-report-2026/</a></li><li>Princeton — GEO Study on generative engine optimization — <a href="https://arxiv.org/abs/2311.09735">https://arxiv.org/abs/2311.09735</a></li><li>Omnius — ANNA Money Fintech SEO Case Study — <a href="https://omnius.so/blog/anna-seo-fintech-case-study">https://omnius.so/blog/anna-seo-fintech-case-study</a></li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=14a1cfc71dc7" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The B2B SaaS funnel in 2026: separating real change from vendor narrative]]></title>
            <link>https://medium.com/@omniusSo/the-b2b-saas-funnel-in-2026-separating-real-change-from-vendor-narrative-1820afc0fb43?source=rss-571819055169------2</link>
            <guid isPermaLink="false">https://medium.com/p/1820afc0fb43</guid>
            <category><![CDATA[b2b-marketing]]></category>
            <category><![CDATA[b2b]]></category>
            <category><![CDATA[b2b-content-marketing]]></category>
            <category><![CDATA[saas]]></category>
            <category><![CDATA[b2b-saas]]></category>
            <dc:creator><![CDATA[OMNIUS]]></dc:creator>
            <pubDate>Fri, 10 Jul 2026 10:19:48 GMT</pubDate>
            <atom:updated>2026-07-10T10:19:48.430Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="The B2B SaaS Funnel cover image" src="https://cdn-images-1.medium.com/max/1024/1*v1UgZ_EBWW5UD3fQX-b41g.png" /></figure><p>Updated: July 2026</p><p>The headline has appeared in a dozen forms this year, under a dozen logos, most of them an intent-data vendor or an agency. The buyer journey is a pinball machine now. The MQL is obsolete. Attribution is a fantasy. Everything the previous playbook assumed is wrong, and the product being sold is the one that fixes it.</p><p>Most of it is one true change wearing five costumes.</p><p>One thing did break, cleanly and for good, and it deserves to be taken seriously. Almost everything stacked on top of it is either the same change described again or a vendor’s roadmap dressed up as a law of nature. This piece separates the two: what actually changed, what only looks like it changed, and what did not move at all, which turns out to be most of what matters.</p><h3>What actually changed: the information gate is gone</h3><p>For about a decade, B2B SaaS marketing worked because the vendor controlled the information. If a buyer wanted to understand a category, compare options, or see how a product handled a specific use case, the useful material sat behind the vendor’s gate. An email traded for a whitepaper. A demo booked to get a straight answer on pricing. The sales rep was not just a closer; the rep was the path to the facts.</p><p>That assumption is gone, and it is not coming back.</p><p>Buyers now assemble most of their understanding before vendors know they exist. 51% of B2B software buyers now start their research inside an AI chatbot more often than in Google, up from 29% in April 2025. That figure roughly doubled in eleven months (G2, March 2026 survey of 1,076 B2B decision-makers). Reddit has overtaken review sites for a lot of bottom-of-funnel searches precisely because it is not curated by anyone selling anything. By the time a buyer reaches a vendor’s site, the pattern that has held for years still applies: they are 60 to 70% through their decision before first contact, with 95% of winning vendors already sitting on the Day-One shortlist (6sense, 2025 Buyer Experience Report).</p><figure><img alt="Share of B2B software buyers who start research in an AI chatbot" src="https://cdn-images-1.medium.com/max/1024/1*rHDf2ZU4spwfQUm6JVflow.png" /></figure><p>This genuinely breaks things, and the breakage is worth naming plainly. The gated whitepaper is dead as a lead source because the information inside it is free and better elsewhere. Top-of-funnel explainer content decays faster because the AI answers the question without the click. Any funnel built to walk an uninformed buyer through stages in sequence is describing a buyer who no longer exists.</p><p>The Atomic AGI Q1 2026 report captures the scale of the compression: 48.6% of landing pages across 11,000+ tracked domains received zero AI search traffic in the first quarter of 2026. The information gate did not just open. Half of vendor content is no longer being routed to at all.</p><p>Something specific and large changed. Buyers stopped needing vendors to find things out. The question the rest of this piece asks is narrower and more useful than “did the funnel die.” It is: now that buyers inform themselves, what exactly did that break, and what did it leave completely untouched.</p><h3>What did not change: the buyer’s real job</h3><p>Losing the information gate was the real change. It is worth being clear about what it did not touch.</p><p>The person spending the money in B2B is spending someone else’s, and answering to people who can hurt their standing if it goes wrong. That single condition explains almost everything slow and expensive about how businesses buy. It has nothing to do with how buyers find vendors, and everything to do with what they need before they commit.</p><p>So the buyer’s real job was never to find the best product. It is to make a choice they can defend. If the purchase fails, they have to be able to point to the diligence they did, the people who agreed, the process they followed. A great product that cannot be defended loses to an adequate one that can. This is not irrationality. It is a rational response to the fact that being wrong in public costs more than being slightly suboptimal in private.</p><p>Once the purchase is understood as a defense being assembled, the whole B2B apparatus stops looking strange. Committees are not indecision: they are blame distributed across enough names that no single one carries the fall. Procurement is not obstruction: it is a paper trail built on purpose. Reference calls are not research: they are borrowed credibility, a way to say someone comparable already took this risk and survived. The average B2B buying committee now includes 6 to 10 stakeholders, and 79% of B2B purchases require CFO approval (Gartner; TrustRadius). The cycle is long because a defense takes time to build.</p><p>None of that is nostalgia, because none of it is going away. It is rooted in employment and organizational politics, not in search behavior. Technology changed how buyers gather information. It did not change what happens to them if the decision blows up.</p><p>Discovery moved. The defense did not.</p><p>This gives a better account of modern buying than the one that circulates most often. The vendor-side story is that buyers self-educate now, so they arrive late and rep-averse. True, but it explains the surface, not the reason. Buyers do their research privately because they are assembling their defense before they expose themselves to a salesperson. They avoid reps early because early sales contact adds pressure and no defensibility. And on the largest deals they invite reps back in late — which the self-education story cannot explain at all — because at high stakes a named human on the vendor side becomes part of the defense, someone to point to when the decision is challenged.</p><p>That is the filter for the rest of this piece. Every claim about the 2026 funnel can be sorted with one question. Does it change how buyers discover, or does it claim to change how they defend the choice? The first is usually real. The second is usually a vendor hoping the two get confused.</p><h3>Running the filter: three claims worth testing</h3><h3>GEO gets brands shortlisted. It does not get them bought.</h3><p>The loudest advice this year is to optimize for AI answers. Get cited in ChatGPT, Perplexity, and Gemini, because that is where research starts now. This is true, and worth doing. The overreach is the next sentence, the one that says GEO replaces the funnel.</p><p>It does not, because a citation is not evidence. When a buyer takes a vendor name into a procurement review, “an AI recommended it” carries no weight against a security questionnaire, a reference call, or a payback model. Those artifacts have not changed. The buyer still has to stand in front of people who can say no and make the case in terms they can repeat. GEO moves a vendor from unknown to considered. Everything after considered runs on the same proof it always did.</p><p>This is visible in the citation surface itself. Across 11,000+ domains tracked by Atomic AGI in Q1 2026, the top citation sources AI engines pull from are Reddit (14.3%), LinkedIn (14.05%), and YouTube (6.38%). These are earned surfaces. Off-domain signals shaped over years of participation. A newly-optimised landing page does not enter the citation set overnight. The GEO lift is real, but it is a discovery lift, not a conversion lift.</p><figure><img alt="A real buyer assembling a shortlist before any vendor knows they exist — weighing HubSpot, Pipedrive, and Bigin on Reddit, not a vendor’s site. Source: r/CRM." src="https://cdn-images-1.medium.com/max/1024/1*vYsoIiGwWj147s7uAEuF7Q.jpeg" /></figure><p>The practical split is this. Treat GEO as discovery spend, measured by whether the brand enters the consideration set. Do not let it starve the assets that actually close, which are the boring, defensible ones. ANNA Money’s GEO rebuild produced a 164% increase in AI search traffic over six months alongside a 60% increase in weekly signups. The AI traffic figure is the citation result. The signup figure is the commercial result. Both required both layers of work: discovery infrastructure and defensible product content. (ANNA Money case study)</p><h3>Rep-free buying is real until the deal gets expensive, then it reverses.</h3><p>Every article cites the same figures. Buyers reach out most of the way through their research. A majority say they would prefer to buy without talking to a salesperson at all. The conclusion drawn from this is that sales is receding across the board.</p><p>The invariant predicts a different pattern. Buyers avoid salespeople to avoid pressure and wasted time on low-stakes decisions. But the same buyer facing a large contract, a hard-to-reverse switch, or a committee of eleven does the opposite. They pull humans back in, late and deliberately, because a shared decision spreads the blame if it goes wrong. Nobody wants to be the single name on a failed six-figure purchase.</p><p>That is why rep-free buying holds for a self-serve tool and collapses for enterprise. The behavior does not scale smoothly. It inverts somewhere along the deal-size band. Any article selling one funnel for all of B2B SaaS is quietly writing for the segment where its advice happens to be true, usually mid-market and product-led, and mis-selling it as universal. The most useful discipline before applying any 2026 funnel advice: find out which segment it was written for. If the author will not say, that is the answer.</p><h3>Most of what gets blamed on AI is capital discipline returning.</h3><p>The framing is that AI rewrote the rules. Buyers self-educate, attribution broke, the old playbook died. AI did some of this. It did not do most of it.</p><p>The median SaaS company now spends around $1.30 to acquire $1 of new recurring revenue, up sharply from a few years ago (Aleph × Benchmarkit 2025 data across 342 SaaS companies). Payback periods stretched from 14 months in 2023 to 18 months in 2024, before improving slightly to 16 months in 2025. Sales and marketing efficiency dropped by roughly half in 2024–2025 compared to 2023 benchmarks (Lighter Capital 2025 B2B SaaS Startup Benchmarks). None of that came from ChatGPT. It came from the end of cheap capital, which ended the era when growth could be bought and the math worried about later.</p><p>The scrutiny marketers now face — the demand to prove pipeline, the death of the vanity metric — is capital discipline returning after a long absence. It is wearing an AI costume because that is the more flattering story. AI made the discovery layer look completely different at the same moment that unit economics started getting audited seriously. Two things happened. They are being described as one thing.</p><p>This matters for what to do about it. If the pipeline problem is a discovery problem, GEO and community help. If it is a payback problem, no amount of AI-anything fixes it, because the constraint is unit economics, and unit economics do not care which channel found the customer. Diagnose which problem the company actually has before buying the tool that promises to solve both.</p><h3>What actually changed vs what did not</h3><figure><img alt="What actually changed vs what did not table comparison" src="https://cdn-images-1.medium.com/max/1024/1*iboCh75zPlPCaTIoe1jGCg.png" /></figure><h3>Who benefits from the funeral</h3><p>It is worth noticing who keeps repeating that the funnel is dead.</p><p>The “MQL is obsolete” case is made most often by companies selling intent data to replace it. The “attribution is broken” case comes from attribution platforms. The “old playbook is finished” case comes from agencies who would like to sell the new one. This does not make the claims wrong. It makes them interested, and interested parties describe change in the shape that sells their product.</p><p>The tell is the size of the claim. A vendor with a signal-scoring tool has a reason to say the entire qualification model collapsed, not that one part of it aged badly. A total revolution needs a total replacement, and the replacement is for sale. The more complete the death, the more the buyer needs what the vendor has.</p><p>Read the genre with that in mind and the numbers get more interesting too. The figure that half of buyers start in an AI chatbot comes from a company that sells AI-search visibility. The pinball-machine buyer journey is a fine argument for a platform that promises to light up the dark parts. None of this is fabrication. It is emphasis, chosen by people who benefit from where the emphasis lands. The data can be used without accepting the conclusion attached to it.</p><h3>What to actually do</h3><p>Do not rebuild the gate. It is gone, and the effort to wall information back up behind forms and demos is the one genuinely obsolete move. Buyers inform themselves now. Meet them already informed.</p><p>Do not buy the revolution either. The buyer still has to defend the choice to people who can say no, and that has not changed and will not, because it is built into how organizations spend money, not into how they search. Much of what is being sold as a new funnel is the old discipline, applied to a discovery process that moved out of vendor control.</p><p>The work is not to find a new model. It is to tell the one real change from the five costumes it is wearing, and to spend accordingly. Invest in discovery infrastructure — GEO citations, community presence, earned surface area — because that is where the buyer’s understanding is being assembled. Invest in defensibility infrastructure — clear pricing, credible case studies with real numbers, reference architecture, security documentation — because that is where the purchase gets defended. Cut spend on anything that assumes buyers arrive uninformed.</p><p>The B2B SaaS funnel narrative for 2026 has one real change and several vendor narratives layered on top of it. Discovery moved. Defense did not. Every claim that says both moved is either wrong or selling something. The work is separating them.</p><h3>References</h3><ul><li>G2 — Half of B2B Software Buyers Now Start Their Research With AI Chatbots (March 2026 survey, 1,076 B2B decision-makers). <a href="https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html">https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html</a></li><li>6sense — 2025 Buyer Experience Report (nearly 4,000 B2B buyers). <a href="https://6sense.com/science-of-b2b/buyer-experience-report-2025/">https://6sense.com/science-of-b2b/buyer-experience-report-2025/</a></li><li>Atomic AGI — Q1 2026 AI Search Quarter Report. <a href="https://app.atomicagi.com/public/reports/q1-2026-ai-search-quarter-report">https://app.atomicagi.com/public/reports/q1-2026-ai-search-quarter-report</a></li><li>Aleph × Benchmarkit — 2026 SaaS &amp; AI Performance Benchmarks (342 SaaS companies). <a href="https://www.getaleph.com/answers/cac-payback-period-saas-2026">https://www.getaleph.com/answers/cac-payback-period-saas-2026</a></li><li>Lighter Capital — 2025 B2B SaaS Startup Benchmarks. <a href="https://azbigmedia.com/business/why-b2b-saas-companies-are-losing-the-paid-acquisition-game-in-2026/">https://azbigmedia.com/business/why-b2b-saas-companies-are-losing-the-paid-acquisition-game-in-2026/</a></li><li>Gartner — Gartner Sales Survey Finds 74% of B2B Buyer Teams Demonstrate “Unhealthy” Conflict During the Decision Process. <a href="https://www.gartner.com/en/newsroom/press-releases/2025-05-07-gartner-sales-survey-finds-74-percent-of-b2b-buyer-teams-demonstrate-unhealthy-conflict-during-the-decision-process">https://www.gartner.com/en/newsroom/press-releases/2025-05-07-gartner-sales-survey-finds-74-percent-of-b2b-buyer-teams-demonstrate-unhealthy-conflict-during-the-decision-process</a></li><li>Omnius — ANNA Money Fintech SEO Case Study. <a href="https://omnius.so/blog/anna-seo-fintech-case-study">https://omnius.so/blog/anna-seo-fintech-case-study</a></li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=1820afc0fb43" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Content marketing for fintech: five principles that build trust in 2026]]></title>
            <link>https://medium.com/@omniusSo/content-marketing-for-fintech-five-principles-that-build-trust-in-2026-49008c1cd8ce?source=rss-571819055169------2</link>
            <guid isPermaLink="false">https://medium.com/p/49008c1cd8ce</guid>
            <category><![CDATA[content-marketing]]></category>
            <category><![CDATA[fintech]]></category>
            <category><![CDATA[fintech-content-marketing]]></category>
            <dc:creator><![CDATA[OMNIUS]]></dc:creator>
            <pubDate>Thu, 02 Jul 2026 12:01:03 GMT</pubDate>
            <atom:updated>2026-07-02T12:01:03.788Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="Fintech Content Marketing — Five principles that build trust in 2026 cover image" src="https://cdn-images-1.medium.com/max/1024/1*mdo3luR-VbMe6abAaXLJCg.png" /></figure><p>Updated: June 2026</p><p>Fintech content earns trust differently than SaaS content. The product isn’t software. It’s the belief that money will be safe tomorrow. Content marketing playbooks built for SaaS, e-commerce, or B2B services don’t account for this distinction, and the result is fintech content that ranks well but doesn’t convert, or converts but doesn’t compound brand trust over time.</p><p>The five principles below come from fintech client engagements across regulated and budget-constrained markets. Each one names a specific operational discipline, the mechanic that makes it work, and the failure mode that signals it isn’t in place.</p><h3>1. Write for the buyer’s stakeholders, not the buyer</h3><p>In B2B fintech, the person reading the content is rarely the person who approves the purchase. Champions need to defend the decision to CFOs, compliance officers, and boards. Content that arms champions with answers for those conversations outperforms content that pitches the buyer directly.</p><p>The scale of this matters in fintech specifically. The average B2B buying committee now includes 6 to 10 stakeholders (Gartner), and 79% of B2B purchases require CFO approval (TrustRadius). Content that arms the champion to navigate that committee outperforms content that pitches the buyer alone.</p><p>The most effective fintech content takes this seriously at the structural level. “Five reasons to switch to [product]” is content for the buyer. “What your CFO will ask before approving a payment platform migration, and how to answer” is content for the buyer’s boss. Champions share the second piece because it makes their internal advocacy easier. They rarely share the first because forwarding marketing collateral signals a lack of confidence in the analysis.</p><p>ANNA Money’s content rebuild applied this principle by restructuring around the buyer’s actual stakeholders. Rather than addressing “small business owners” as a single audience, the content was rebuilt around the industries those owners operated in — landlords, contractors, ecommerce founders, freelancers — each cluster naming the specific tax, registration, and compliance questions those stakeholders would face. The result over six months: a 60% increase in weekly signups, with most of the growth coming from buyers arriving on pages that named their exact regulatory context. (ANNA Money case study)</p><p>The test: would a champion forward this page to their CFO without feeling embarrassed? If the answer is no, the page is doing the wrong job.</p><h3>2. Show the math, not the outcome</h3><p>In categories where trust is the product, claims without verifiable inputs carry no weight. The numbers that earn trust are the ones a reader can reproduce — methodology, assumptions, edge cases — published before the headline, not buried in a footnote afterwards.</p><p>A claim like “saved customers 40% on fees” is free to make and therefore worth nothing to a sophisticated buyer evaluating financial software. The same claim becomes credible when the methodology sits next to it: which customers, over what window, calculated against what baseline, with what assumptions about transaction volume and currency mix.</p><p>Apply this to every content asset. A case study without numbers is a testimonial in a suit. A benchmark report without raw data is content that gets cited by nobody. A “return on investment” calculator without the underlying formula is marketing dressed as analysis.</p><p>The practical version: publish the methodology before the headline. Open with what was measured, how, and where the methodology breaks down — then state the result. Buyers who matter respond to the willingness to be checked. They distrust the absence of that willingness.</p><p>This applies to product claims as much as content. Pricing pages that show the calculation. Comparison tables that link to verifiable sources. Case studies that name the engagement window, the starting baseline, and the assumptions behind the result. The willingness to be specific is what compounds.</p><p>The Edelman Trust Barometer 2026 captures the structural opportunity for fintech specifically. While 73% of people say CEOs are obligated to facilitate trust-building, only 44% believe they’re doing it well — a 29-point gap. Trust in the Financial Services sector globally now sits at 63%, the only sector to record double-digit trust growth since 2021. Content that shows the math closes the CEO trust gap. Content that doesn’t widens it.</p><h3>3. Compliance is content infrastructure, not a constraint</h3><p>The constraints regulatory and legal teams impose on fintech content are often treated as a tax on the marketing function. In a category where trust is the product, those constraints function as a signal. Content that names jurisdictions, qualifies claims, and links to actual regulations does brand work that no amount of marketing copy can replicate.</p><p>The standard fintech marketing workflow treats legal review as a final-stage friction point — copy gets written, then sent for clearance, then revised under protest. This sequence misses where compliance creates content value.</p><p>Bringing compliance and legal into the brief stage, rather than the final review stage, surfaces two things content teams rarely capture on their own: the questions every buyer is already asking quietly, and the qualifying language that signals genuine institutional credibility rather than aspirational positioning.</p><p>The most underrated content asset in many fintech sites is a “How we’re regulated” page. The structure: every license, every regulator, every audit, with active links to the original sources. Long, dense, and unmistakably specific. These pages rarely make a content calendar because they read as unsexy. They appear in every champion’s internal pitch deck because they answer the only question that matters in a CFO conversation: is this defensible to procurement and risk?</p><p>This principle also drives AI citation in regulated content. LLMs need explicit references to verifiable authorities to confidently cite a fintech page. Schema markup, regulator references, and clear attribution all move from compliance hygiene to citation infrastructure.</p><h3>4. Pick defensible positions, not slogans</h3><p>Fintech POV content compounds when the position is narrow, specific, and backed by receipts. A slogan (“we believe in transparency”) builds no track record. A defensible specific claim (“cross-border providers commonly hide a 3% FX markup in their ‘free transfer’ messaging”) builds one — and creates a body of work that compounds across multiple publications.</p><p>The criteria for a position worth taking publicly:</p><ul><li>A specific target rather than an abstract one</li><li>A claim that can be checked</li><li>Receipts the team is actually willing to produce when challenged</li></ul><p>A fintech that publishes a benchmark on transaction fees, names which providers it tested, shows the methodology, and updates the benchmark quarterly is building an asset that compounds. A fintech that publishes “we believe in fair fees” without supporting data is publishing a tagline.</p><p>The selection criteria for a fight worth picking: pick fights where the track record supports the claim. A customer service position requires customer service data. A speed claim requires published benchmarks. The trust dividend on being right specifically is meaningful. The trust cost of being caught wrong specifically is terminal — fintech buyers have long memories, and a wrong specific claim ends conversations that took quarters to start.</p><h3>5. Write fewer pieces, source them harder</h3><p>Trust-building content runs on a different rhythm than volume-led calendars. The teams producing trust-compounding content publish less and source more. One substantive quarterly report with original data outperforms thirteen blog posts with rehashed industry takes — and is significantly more efficient at every stage of the funnel.</p><p>Distribution does the work that publishing volume tries to do. A single substantive asset gets pitched to journalists, sent to analysts, referenced by customers in their internal advocacy, and cited by AI engines for months after publication. Volume-led content rarely earns any of these.</p><p>This shift was visible in a budget-constrained fintech engagement that produced a 73.9% increase in organic clicks in three months without a content team or paid acquisition budget. The mechanic wasn’t publishing more — it was selecting the keywords the site could realistically rank for inside the timeline, building content against those queries first, and refusing to spend effort on content the site couldn’t compete for in the window. The constraint forced the selection discipline that volume-led calendars rarely develop. (Fintech SEO case study)</p><p>The test: is the content team generating one citable, original-data piece per quarter? If not, the content programme is operating as a treadmill, not as a trust-building system.</p><h3>Volume-led vs trust-led fintech content</h3><figure><img alt="Volume-led vs trust-led fintech content table preview" src="https://cdn-images-1.medium.com/max/1024/1*bhwgdb2hAnbNvwyv-0v3Nw.png" /></figure><p>Trust in fintech content isn’t a tone, a brand voice, or a content format. It’s a pattern of operational discipline across the entire content surface: writing for the stakeholders who block the deal, publishing the methodology before the headline, treating compliance as content infrastructure, picking specific fights, and refusing to publish content the team can’t defend.</p><p>The fintech brands compounding trust through content in 2026 made these decisions in 2022. The ones starting now will compound through 2028, if the discipline holds.</p><h4>References</h4><ol><li>Omnius — ANNA Money Fintech SEO Case Study <a href="https://omnius.so/blog/anna-seo-fintech-case-study">https://omnius.so/blog/anna-seo-fintech-case-study</a></li><li>Omnius — Fintech SEO Case Study: 73.9% Organic Traffic Growth <a href="https://omnius.so/blog/fintech-seo-case-study">https://omnius.so/blog/fintech-seo-case-study</a></li><li>Edelman — 2026 Edelman Trust Barometer: Key Insights for Financial Services <a href="https://www.edelmansmithfield.com/2026-Edelman-Trust-Barometer-Key-Insights-for-Financial-Services">https://www.edelmansmithfield.com/2026-Edelman-Trust-Barometer-Key-Insights-for-Financial-Services</a></li><li>Gartner — Gartner Sales Survey Finds 74% of B2B Buyer Teams Demonstrate “Unhealthy” Conflict During the Decision Process <a href="https://www.gartner.com/en/newsroom/press-releases/2025-05-07-gartner-sales-survey-finds-74-percent-of-b2b-buyer-teams-demonstrate-unhealthy-conflict-during-the-decision-process">https://www.gartner.com/en/newsroom/press-releases/2025-05-07-gartner-sales-survey-finds-74-percent-of-b2b-buyer-teams-demonstrate-unhealthy-conflict-during-the-decision-process</a></li></ol><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=49008c1cd8ce" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Why fintech brands aren’t cited in ChatGPT, Perplexity, or Google AI Mode]]></title>
            <link>https://medium.com/@omniusSo/why-fintech-brands-arent-cited-in-chatgpt-perplexity-or-google-ai-mode-05d17fd8142e?source=rss-571819055169------2</link>
            <guid isPermaLink="false">https://medium.com/p/05d17fd8142e</guid>
            <category><![CDATA[seo]]></category>
            <category><![CDATA[geo]]></category>
            <category><![CDATA[fintech]]></category>
            <category><![CDATA[ai-search]]></category>
            <dc:creator><![CDATA[OMNIUS]]></dc:creator>
            <pubDate>Wed, 10 Jun 2026 14:26:30 GMT</pubDate>
            <atom:updated>2026-06-10T14:26:30.013Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*7S73nccBGgcqqdJMhhX9fg.png" /></figure><p><em>Updated: May 2026</em></p><p>Ranking on Google does not translate to citation in ChatGPT, Perplexity, or Google AI Mode. The gap is sharpest in fintech, where the structural requirements of regulated content collide with what answer engines actually retrieve. A fintech brand can hold first-page rankings across its core terms and still appear in zero AI-generated answers about its category, which is the surface where evaluation-stage buyers now form their shortlists.</p><p>Only 12% of ChatGPT’s cited sources overlap with Google’s top-10 results, and across 11,000+ domains tracked by Atomic AGI in Q1 2026, 48.6% of landing pages received zero AI traffic at all. The reasons are structural, not editorial.</p><h3>What “invisible in AI search” actually means for fintech</h3><p>Invisibility in AI search is a citation problem, not a ranking problem. A fintech brand is invisible when answer engines do not select its pages as sources when generating responses to category-defining queries, even when those pages rank well in traditional search. The brand has the index position. It does not have the citation position.</p><p>The two surfaces operate on different signals. Google ranks pages on relevance and authority signals tied to backlinks, technical health, and content quality. AI engines select citations based on entity clarity, answer-first structure, schema density, and off-site signals that confirm the brand is a credible source on the specific topic. Citation sources on AI platforms also churn 40 to 60% monthly, so even temporary visibility doesn’t translate into sustained citation share without continuous content and trust signal work.</p><p>For fintech buyers researching a business account, a lending product, or a regtech tool, the practical consequence is straightforward. 68.6% of AI search queries are educational and 23.1% are commercial (Atomic AGI Q1 2026 report). The category research happens inside AI. The shortlist is built before the buyer ever lands on a fintech homepage. A brand that is invisible at that stage is invisible during the only decision-making window that matters.</p><h3>Why fintech is structurally harder than SaaS</h3><p>Fintech faces three structural conditions that SaaS does not. Regulated content carries hedging language that answer engines read as low-confidence and skip. Trust requirements demand explicit credentials and proof that fintech pages often bury rather than surface. Cross-market operations produce localised content that is translated rather than rebuilt for the search behaviour of each market.</p><p>In SaaS, a clear answer is usually a competitive advantage. In fintech, a clear answer can be a compliance risk. The cautious legal language that protects the brand on the regulated side is the same language that makes answer engines bypass the content on the search side. The page reads as authoritative to a human but as evasive to an LLM extracting a direct response.</p><p>The trust architecture problem compounds this. Fintech buyers want regulator references, executive credentials, audit certifications, and customer proof surfaced explicitly. Answer engines look for the same signals: author schema, organisation schema, reference citations to recognised bodies. The pages that buyers find reassuring on a homepage rarely contain the structured markup that AI engines need to verify the entity behind the claim.</p><p>Off-site signals matter even more in fintech than in SaaS. The Atomic AGI Q1 2026 report shows the top citation sources AI engines pull from are Reddit (14.3%), LinkedIn (14.05%), and YouTube (6.38%). Fintechs are structurally less present on all three than B2B SaaS: fewer organic Reddit threads, less employee LinkedIn activity on regulated topics, and rarely a YouTube strategy on financial products.</p><p>The third condition is multi-market behaviour. A UK business account product, a German lending product, and a US payments product face different search queries, different trust signals, and different competitive citation environments. Content translated across markets without restructuring carries the search assumptions of the source market into a context where they do not apply.</p><h3><strong>The three failure modes seen across fintech audits</strong></h3><h3><strong>Failure mode 1: Compliance language hedging</strong></h3><p>Compliance-driven copy hedges every concrete claim. Phrases like “may be eligible,” “could potentially provide,” and “subject to terms and conditions” appear in the same paragraph as the core product description. The legal protection is necessary. The structural cost is that answer engines cannot extract a confident response from the page.</p><p>The fix is not to remove the hedging. The fix is to separate the definitive claim from the regulatory qualifier. A direct answer at the start of each section carries the assertion. The qualifier sits in a clearly demarcated block below. Answer engines extract the direct answer. Regulators are satisfied by the qualifier. The same page serves both audiences.</p><h3>Failure mode 2: Missing trust architecture</h3><p>Trust signals exist on the page but are not encoded for machine extraction. The compliance badge sits in the footer as an image. The executive bio links to a CV rather than carrying Person schema. The audit certification appears in the press release archive rather than as a structured reference on the product page. According to OtterlyAI’s citation study, 73% of brand sites have technical barriers that block AI crawler access entirely. For fintechs specifically, this often comes from CDN rules and JavaScript rendering on regulated pages that exist for buyers but can’t be parsed by the AI engines those buyers are now querying.</p><p>The fix is encoding what already exists. Organisation schema, Person schema with role and credentials, reference links to regulatory bodies with verifiable identifiers. Reference-grade content with proper schema and chunked structure receives 3 to 5x more AI citations than comparable pages without it (OtterlyAI). Off-site trust signals (third-party mentions in industry publications, regulator listings, recognised review platforms) carry more weight in fintech than in any other vertical, and answer engines weight them accordingly.</p><h3>Failure mode 3: Surface translation across markets</h3><p>Cross-market content is translated rather than rebuilt. The page that ranks for “open a business account” in the UK is translated into German and expected to perform for “Geschäftskonto eröffnen.” The keyword is correct. The buyer intent, the trust references, and the competitive citation environment in the German market are different.</p><p>The fix is rebuilding the content for the market it operates in. Local trust signals, local regulator references, local terminology that the AI engines in that market have learned to associate with the category. ANNA Money’s UK content treats landlords, contractors, and ecommerce founders as distinct buyer segments with distinct content clusters. A translated version of that approach into another market would carry assumptions that do not apply there.</p><h3>Ranking-optimised vs. AI-citation-optimised fintech content</h3><figure><img alt="Ranking-optimised vs. AI-citation-optimised fintech content table" src="https://cdn-images-1.medium.com/max/1024/1*0T2eBypharVk4QxNIjDI3g.png" /></figure><h3>What the fix looks like in practice</h3><p>The fix runs across content architecture, schema, and off-site signal strategy at the same time. Ranking and citation begin to move on different timelines, and citation share is the metric that confirms the AI search visibility problem is being solved.</p><p>ANNA Money entered the engagement with strong brand recognition among UK entrepreneurs and an organic presence that did not match it. The rebuild restructured the content around the industries and professions ANNA’s customers actually belong to (landlords, contractors, ecommerce founders, freelancers), with a GEO layer applied across every cluster. Six months later: a 60% increase in weekly signups, a 164% increase in AI search traffic, and 314% growth in top-three keyword positions. The AI search traffic figure is the citation result. The signups figure is the commercial result. (ANNA Money case study)</p><p>AI-referred traffic converts at 14 to 40% depending on platform and vertical, against 2 to 3% for traditional organic search. The conversion premium is why citation share now functions as a leading indicator of pipeline, not just a visibility metric.</p><p>The pattern repeats across fintech engagements that run this structure. The citation result precedes the commercial result, and signups follow once AI search traffic has moved first.</p><p>For fintech brands that hold strong organic rankings but do not appear in AI-generated answers about their category, the problem is structural, not editorial. The compliance language, the trust signals, and the multi-market content can all stay. The way they are encoded for answer engines is what needs to change.</p><p>For teams building a citation tracking program from scratch, our complete AI citation tracking guide covers the implementation framework in detail.</p><p><strong>References</strong></p><ol><li>Atomic AGI — Q1 2026 AI Search Quarter Report<a href="https://app.atomicagi.com/public/reports/q1-2026-ai-search-quarter-report"> https://app.atomicagi.com/public/reports/q1-2026-ai-search-quarter-report</a></li><li>Matija Golubovic — AI citation tracking: the complete guide for SEO and marketing teams, Atomic AGI, April 2026<a href="https://www.atomicagi.com/blog/ai-citation-tracking-the-complete-guide-for-seo-and-marketing-teams"> https://www.atomicagi.com/blog/ai-citation-tracking-the-complete-guide-for-seo-and-marketing-teams</a></li><li>OtterlyAI — The AI Citation Economy: What 1+ Million Data Points Reveal About Visibility in 2026<a href="https://otterly.ai/blog/the-ai-citations-report-2026/"> https://otterly.ai/blog/the-ai-citations-report-2026/</a></li><li>Omnius — ANNA Money Fintech SEO Case Study<a href="https://omnius.so/blog/anna-seo-fintech-case-study"> https://omnius.so/blog/anna-seo-fintech-case-study</a></li></ol><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=05d17fd8142e" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The B2B SaaS content stack that wins in Google and AI search]]></title>
            <link>https://medium.com/@omniusSo/the-b2b-saas-content-stack-that-wins-in-google-and-ai-search-35941c103338?source=rss-571819055169------2</link>
            <guid isPermaLink="false">https://medium.com/p/35941c103338</guid>
            <category><![CDATA[seo]]></category>
            <category><![CDATA[saas]]></category>
            <category><![CDATA[saas-marketing]]></category>
            <category><![CDATA[content-marketing]]></category>
            <dc:creator><![CDATA[OMNIUS]]></dc:creator>
            <pubDate>Wed, 13 May 2026 07:22:16 GMT</pubDate>
            <atom:updated>2026-05-13T07:46:38.152Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="“The B2B SaaS content stack that wins in Google and AI search” Cover image" src="https://cdn-images-1.medium.com/max/1024/1*0mMPTlW51zel_9SzAK62Ng.png" /></figure><p><em>Updated: May 2026</em></p><p>Organic search drives 44.6% of B2B revenue on average (Whitehat SEO). Yet most B2B SaaS content teams measure success in traffic. The gap between those two facts is where most content strategies fail. Not through bad execution, but through the wrong architecture from the start.</p><h3>Why most B2B SaaS content strategies are built backwards</h3><p>Most B2B SaaS content strategies start with awareness and work down: educational content first, then consideration content, then product pages. It’s the sequence that makes intuitive sense and the one most content playbooks teach. It’s also the one that produces traffic without pipeline.</p><p>The structural error: awareness content attracts the widest audience and generates the most traffic. It also attracts the audience with the lowest purchase intent. Teams optimise for what they can measure easily (impressions, sessions, keyword rankings) and end up with dashboards that look healthy while sales teams report qualified lead shortfalls.</p><p>In B2B SaaS specifically, buying decisions involve multiple decision-makers across long sales cycles. The content that matters isn’t the content that gets the most views. It’s the content that shows up when a decision-maker is actively evaluating options, comparing alternatives, or trying to justify a purchase to a CFO. That content lives at the bottom of the funnel. Most SaaS sites build it last, if at all.</p><p>We’ve seen this consistently when clients come to us after previous agency engagements. Traffic numbers were strong. Conversion rates were not. The problem in almost every case: BOFU was thin or missing entirely. A pattern that repeats across audits: 70% of content on a site drives traffic that never converts. The three pages that actually do convert haven’t been updated in 18 months. (Omnius client audit data)</p><h3>The BOFU-first content strategy: what it is and what it changes for SaaS</h3><p>BOFU-first means the content strategy starts with the pages closest to conversion and builds outward. Product pages, comparison pages, alternative pages, use case pages, and pricing context come first. Blog content and educational resources are built on top of that foundation, not instead of it.</p><figure><img alt="Hotelsynch pricing" src="https://cdn-images-1.medium.com/max/1024/1*yuv2VDkE0GGQRQMyyjzbqg.png" /><figcaption>Example BOFU page: a SaaS pricing table with plan tiers, feature comparisons, and free trial CTAs — the conversion surface that most content strategies build last.</figcaption></figure><p>This inverts the standard content pyramid. Instead of building reach and hoping some of it converts, you build conversion surfaces first and then generate demand that flows into them.</p><p>The practical result: pages that attract buyers who are already comparing options sit at the centre of the site architecture. Every other content piece links to them. Internal linking reinforces their authority rather than dispersing it across awareness content with no commercial intent.</p><p>On Omnius’s own site, this architecture generates over 1,000 conversions per quarter (meetings booked and forms filled), tracked through Atomic AGI across every touchpoint. (Omnius BOFU case study) The starting point for every client engagement is the same: map the conversion pages first, then build the content that feeds them.</p><p>What changes downstream: content briefs are written with a conversion destination in mind, not just a keyword target. Every piece has a commercial purpose: either to rank for a BOFU query directly, or to build topical authority that supports a page that does.</p><p>One client reduced output from 40+ articles per quarter to around 25 after mapping their content against conversion intent. Organic traffic went up. Stopping publication of content with no commercial purpose freed budget and attention for the pages that actually moved pipeline.</p><h3>Cluster structure and internal linking: how Google and AI engines read the same signals</h3><p>A single authoritative article on a topic is not sufficient for either Google or AI engines. Both systems look for interconnected, logically structured content that demonstrates a site owns a subject.</p><p>Google uses content clusters and internal linking to understand topical depth and distribute authority across a site. AI systems use the same signals to determine whether a site is a credible source on a given topic. Citation probability increases when content is clustered and cross-linked rather than isolated.</p><p>The mechanics are the same. Content clusters built for SEO authority also build AI citation surface area. A well-structured cluster around “B2B SaaS onboarding” (pillar page, supporting articles on specific use cases, internal links flowing back to conversion pages) serves both purposes simultaneously.</p><p>For a SaaS client starting from zero organic presence in a competitive AI niche, this architecture produced 2.73M organic clicks in 13 months and 600+ keywords ranking in the top 3 positions. (Omnius SaaS growth case study) The cluster structure was the primary driver: product-related keyword clusters supported by alternative and comparison pages, all internally linked to conversion destinations.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*jLDXNhssFWG_1e-BvJQNoA.png" /></figure><p>The internal linking architecture matters as much as the content itself. Pages that are well-linked internally get crawled more frequently, indexed more reliably, and cited more often by AI engines, which have significantly less compute budget than Googlebot and rely more heavily on clear site structure to navigate content.</p><h3>Schema and structure: the layer most teams skip</h3><p>Schema markup is the most consistently skipped layer in B2B SaaS content strategies. It’s technical enough that content teams leave it to developers, and functional enough that developers deprioritise it against product work. The result: sites with strong content and weak structured data signals.</p><p>For AI citation specifically, schema is not optional infrastructure. FAQPage schema correlates with 3.2x higher citation rates in AI Overviews (Princeton GEO Study). Article and FAQPage JSON-LD combined are associated with 36% visibility gains in AI search (Princeton GEO Study). A BOFU page with strong content and no schema is less likely to be cited than a weaker page with schema implemented correctly.</p><p>The schema stack for B2B SaaS content, in priority order:</p><ul><li>Article schema + FAQPage JSON-LD — highest citation impact, implement on all blog and cluster content</li><li>Author schema with Person markup — E-E-A-T signal for AI knowledge graph cross-referencing</li><li>Breadcrumb schema — crawlability and structural clarity</li><li>FAQ schema on BOFU pages — comparison and alternative pages with FAQ sections should have this implemented by default</li></ul><p>Beyond schema, structure within the content itself determines AI extraction quality. Answer-first formatting (a direct 40–60 word answer at the start of every major section) is what AI engines extract before reading the detail. TL;DR blocks at article tops are extracted as standalone citation units. Tables are cited 2.5x more often than paragraphs (Princeton GEO Study).</p><p>Content that ranks in Google but lacks these structural elements will see lower AI citation rates than its organic position would suggest. The two are related but not equivalent.</p><h3>How to audit your existing B2B SaaS content against BOFU intent</h3><p>Before building new content, audit what already exists. Most B2B SaaS sites have more content than they need and less BOFU coverage than they think.</p><h3>Step 1: Export and map by intent</h3><p>Export all indexed URLs. Categorise each page by intent: awareness, consideration, or conversion. Be strict. A blog post titled “What is [your category]” is awareness regardless of how long it is. A comparison page is BOFU regardless of its traffic numbers.</p><h3>Step 2: Identify commercial influence</h3><p>Cross-reference your content list against three signals:</p><ul><li>Which pages appear in sales call recordings or are shared by prospects before a demo</li><li>Which pages have conversion events attached in analytics</li><li>Which pages rank for queries containing “best,” “vs,” “alternative,” “pricing,” or “review”</li></ul><p>Pages that score on any of these are commercially influential. Pages that score on none are candidates for consolidation or removal.</p><h3>Step 3: Find the BOFU gaps</h3><p>Map your commercially influential pages against your product’s full use case, industry, and competitor surface. Where are the gaps? A SaaS tool serving three industries but with BOFU content for only one has an obvious gap. A product with five competitors but comparison pages for two has missed three conversion opportunities.</p><h3>Step 4: Prioritise by search demand and pipeline proximity</h3><p>Not all BOFU gaps are equal. Prioritise by two factors: estimated search demand for the gap keyword, and proximity to conversion in your existing funnel. A comparison page for your highest-volume competitor beats a use case page for a niche industry segment every time.</p><p>This audit process typically takes one to two weeks. The output is a prioritised BOFU content roadmap — not a list of topics, but a ranked queue of pages that directly address conversion-stage buyer intent.</p><h3>What a winning B2B SaaS content strategy looks like in practice</h3><p>The architecture described above is not theoretical. It’s what we build for every B2B SaaS client engagement, and it’s what we run on our own site.</p><p>The sequence in practice:</p><ul><li>Audit existing content against conversion intent — identify BOFU gaps and pages with traffic but no commercial influence</li><li>Build or rebuild BOFU pages: comparisons, alternatives, use cases, pricing context</li><li>Map content clusters around product-related topics — pillar page plus 4–8 supporting articles per cluster</li><li>Implement the schema stack across all content</li><li>Apply answer-first structure and FAQ sections to every article</li><li>Build internal linking architecture that routes authority toward BOFU pages</li><li>Monitor AI citation rates alongside organic rankings — they move on different timelines</li></ul><p>Results across client accounts where this architecture has been fully implemented: pipeline attribution from organic content, not just traffic. AI-referred visitors converting at 4.4x the rate of traditional search visitors (Semrush, 2025). Organic growth that compounds rather than plateauing after initial ranking gains.</p><p>The content strategy that wins in Google and AI search simultaneously isn’t two strategies. It’s one architecture applied consistently.</p><p>For B2B SaaS teams evaluating whether their current content strategy is built for pipeline or just traffic, the audit starts at the bottom of the funnel. If BOFU is thin, everything built on top of it is working harder than it needs to.</p><p><strong>References</strong></p><ol><li>Whitehat SEO — B2B SEO Strategy: <a href="https://whitehat-seo.co.uk/blog/b2b-seo-strategy">https://whitehat-seo.co.uk/blog/b2b-seo-strategy</a></li><li>Princeton GEO Study — Generative Engine Optimisation: <a href="https://arxiv.org/abs/2311.09735">https://arxiv.org/abs/2311.09735</a></li><li>Omnius BOFU SEO Case Study: <a href="https://omnius.so/blog/omnius-bofu-seo-strategy">https://omnius.so/blog/omnius-bofu-seo-strategy</a></li><li>Omnius SaaS Growth Case Study: <a href="https://omnius.so/blog/how-we-grew-saas-tool-in-13-months">https://omnius.so/blog/how-we-grew-saas-tool-in-13-months</a></li><li>Semrush — We Studied the Impact of AI Search on SEO Traffic, June 2025: <a href="https://www.semrush.com/blog/ai-search-seo-traffic-study/">https://www.semrush.com/blog/ai-search-seo-traffic-study/</a></li><li>Originally published on LinkedIn by Omnius. <a href="https://www.linkedin.com/pulse/b2b-saas-content-stack-wins-google-ai-search-omniuspartners-qmtoe/">The B2B SaaS content stack that wins in Google and AI search</a></li></ol><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=35941c103338" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[GEO vs SEO: what it means for SaaS and fintech in 2026]]></title>
            <link>https://medium.com/@omniusSo/geo-vs-seo-what-it-means-for-saas-and-fintech-in-2026-fff12a9ec85a?source=rss-571819055169------2</link>
            <guid isPermaLink="false">https://medium.com/p/fff12a9ec85a</guid>
            <category><![CDATA[seo]]></category>
            <category><![CDATA[search-engine-optimizaion]]></category>
            <category><![CDATA[geo]]></category>
            <category><![CDATA[organic-growth]]></category>
            <dc:creator><![CDATA[OMNIUS]]></dc:creator>
            <pubDate>Wed, 29 Apr 2026 15:55:23 GMT</pubDate>
            <atom:updated>2026-04-29T15:55:23.029Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="GEO vs SEO cover image" src="https://cdn-images-1.medium.com/max/1024/1*b5MkS3fGlXAHw0MaysU1xA.png" /></figure><p><em>Updated: April 2026</em></p><p>There’s a debate running through the SEO industry about whether GEO is a real discipline or just a new name for the same work. That debate doesn’t matter much to growth teams with pipeline targets. What matters: 29% of B2B buyers now start research via AI tools more often than Google, AI-referred visitors convert at 2–4x the rate of organic search visitors (Forrester, 2026), and the correlation between Google rankings and ChatGPT mentions is nearly zero — approximately 0.034. A brand can rank first on Google and not appear in a single AI-generated answer about its category.</p><p>Both surfaces matter. Neither replaces the other.</p><h3>What GEO actually is (and what it isn’t)</h3><p>GEO is not a replacement for SEO, and it is not a cosmetic rebranding of it. SEO determines whether your content can be found by search engines. GEO determines whether AI systems — ChatGPT, Claude, Perplexity, Google AI Overviews, Gemini — select it when generating answers. The same content can pass SEO checks and still fail GEO selection entirely.</p><p>In controlled experiments, GEO tactics improved AI visibility by up to 40% (<a href="https://arxiv.org/abs/2311.09735">Princeton GEO Study</a>). The researchers were clear about one constraint: those gains only applied to content that was already discoverable and indexed. GEO needs SEO to work. Not instead of it. On top of it.</p><p>The confusion about what GEO “is” comes partly from the fact that the structural elements it requires (schema markup, answer-first formatting, entity clarity, FAQ sections) overlap with good SEO practice. They do. But the reason they matter is different. In SEO, schema helps crawlers classify content. In GEO, schema helps AI models extract citable units. The implementation is similar. The purpose and the measurement are not.</p><h3>Why SEO alone no longer covers the full discovery surface</h3><p>Google processes roughly 14 billion searches per day. ChatGPT processes approximately 2.5 billion prompts per day (OpenAI figures, <a href="https://techcrunch.com/2025/07/21/chatgpt-users-send-2-5-billion-prompts-a-day/">reported by TechCrunch</a>). Those numbers look like SEO wins by volume. They describe two different user behaviors, not one.</p><figure><img alt="ChatGPT vs Google comparison chart" src="https://cdn-images-1.medium.com/max/1024/1*P6O2FHNPdWf44Ax-iUXHpw.jpeg" /></figure><p>95% of ChatGPT users also use Google. Behavior is hybrid, not either/or. Buyers move between both surfaces depending on the stage and type of question. Early-stage research, comparisons, and “explain this to me” queries increasingly happen in AI. High-intent, navigational, and transactional queries still happen in Google. For SaaS and fintech brands, the mid-funnel — where buyers compare options and form preferences — is now happening in a place standard SEO measurement can’t see.</p><p>Only 62% of brands that appear on Google’s first page appear in ChatGPT answers. That gap doesn’t close with more SEO work. It requires a different layer.</p><p>Recent crawler data confirms this. In February 2026, 59% of sites received at least one AI crawler visit across a dataset of hundreds of thousands of sites (<a href="https://www.searchenginejournal.com/68-million-ai-crawler-visits-show-what-drives-ai-search-visibility/572386/">SEJ</a>, April 2026). Sites connected to external data systems — structured review integrations, verified business data, third-party entity signals — were crawled at rates as high as 97%. The difference between a site that gets crawled by AI systems and one that doesn’t is increasingly about structured, verifiable information architecture, not content volume.</p><p>The surface is also more fragmented than most brands assume. Data from Omnius’s own tracking across 11,000+ domains shows Claude has become the second-largest AI traffic source in B2B, with 21%+ share and +76.2% quarter-on-quarter growth (<a href="https://www.linkedin.com/posts/matdegaulle_data-study-it-looks-like-%F0%9D%97%96%F0%9D%97%B9%F0%9D%97%AE%F0%9D%98%82%F0%9D%97%B1%F0%9D%97%B2-is-ugcPost-7452008733938966530-xTs1/?utm_source=share&amp;utm_medium=member_desktop&amp;rcm=ACoAAB7hPqgBZFqTr_6KdAvFbZ9Fwo7BYqhZ8ew">Mat Golubović</a>, April 2026). More significantly, Claude assigns 10–20 percentage points more visibility to individual domains than ChatGPT does on the same queries, with some gaps exceeding 40 points. Domains sitting at 0% visibility in ChatGPT regularly reach 5–15%+ in Claude. Gemini and Perplexity track closer to ChatGPT’s behavior.</p><figure><img alt="Claude AI search performance in AtomicAGI SEO software" src="https://cdn-images-1.medium.com/max/1024/1*yYx0hRqLzjWCS2OFytbqoQ.png" /></figure><p>The practical implication: optimising for one AI engine does not mean you’re visible across the others.</p><h3>What GEO requires in practice: the structural layer</h3><p>GEO is not a content project. It is a structural layer that runs across content, technical setup, and off-site presence.</p><p>The elements that drive AI citation probability, based on the Princeton GEO Study and Semrush’s analysis of 89,000 cited LinkedIn URLs:</p><figure><img alt="Table showing content elements and their impact on AI citation rates, such as FAQ schema, tables, and recent updates." src="https://cdn-images-1.medium.com/max/1024/1*zWwHmmEPHBLiVl19ocVUiQ.png" /></figure><p>Content without these elements gets crawled and passed over. Content with them gets cited.</p><p>What GEO also requires — and this is where most brands have the largest gap — is a more deliberate off-site presence strategy. Traditional off-page SEO optimises for link equity. GEO requires broader distributed recognition: unlinked mentions in podcasts, publications, Reddit threads, review platforms, and communities. Bing, which powers most non-Google AI tools, weighs what others say about your brand more heavily than what your own site claims. AI systems pull from these sources directly, not just through backlink signals.</p><h3>Why SaaS and fintech brands need both</h3><p>SaaS and fintech buyers are among the most research-intensive in B2B. Procurement often involves multiple stakeholders, a long evaluation window, and significant comparison activity. That comparison activity is shifting to AI.</p><p>Applying the same GEO layer we build for clients to our own site, Omnius generated 36 SQLs from AI-referred traffic in six months — a 414% increase over the previous period (<a href="https://www.omnius.so/blog/omnius-ai-search-case-study">Omnius GEO Case Study,</a> 2026). The conversion rate differential was consistent with what Forrester reports across the market: AI-referred visitors arrive already educated and comparing, which shortens the sales cycle before a human ever gets involved.</p><figure><img alt="AI search and total conversions performance in the AtomicAGI SEO software" src="https://cdn-images-1.medium.com/max/1024/1*mA_lHx9qSXKRvY8aje3wbQ.jpeg" /></figure><p>Analysis of what’s winning in Google right now reinforces why SEO still matters in parallel. Across 400+ winning and losing websites, the highest-correlated predictor of traffic growth was offering a product or service as a core offering, present in 70% of winning sites vs 34% of losing ones (Zyppy / Cyrus Shepard, April 2026). Tight topical focus and proprietary assets followed. These are SEO fundamentals. They don’t go away when GEO enters the picture. They become prerequisites for GEO to work.</p><p>The practical split for SaaS and fintech growth teams:</p><p><strong>SEO covers:</strong> technical health, crawlability, indexation, keyword-targeted content, BOFU and comparison pages, backlink profile</p><p><strong>GEO covers:</strong> schema stack, answer-first content structure, FAQ sections, entity clarity, third-party mention strategy, AI citation monitoring</p><p>Neither is sufficient alone. A site with no SEO foundation won’t be indexed, and GEO has nothing to work with. A site with strong SEO but no GEO layer will rank in Google and be absent from AI answers — including the ones where buyers are forming their shortlists.</p><p>For SaaS and fintech brands still treating GEO as a future consideration, the attribution data is already there. The question is whether the discovery surface where buyers are forming opinions about your category includes you or your competitors.</p><p>We’ve published a detailed breakdown of the technical and content changes involved at [<a href="https://www.omnius.so/blog/seo-vs-geo">omnius.so/blog/seo-vs-geo</a>].</p><p><em>Originally published on LinkedIn by Omnius. GEO vs SEO: what it means for SaaS and fintech in 2026</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=fff12a9ec85a" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>