Launch & Growth A2 · News Agentic website basicsWhen the buyer is an AI agent

Cited, used or closed: agent readiness vs AEO vs GEO, and where the sale happens

AEO and GEO aim to get you cited, agent readiness to get your site used. What each measures, what files can and cannot do, and where the sale happens.

Olia Nemirovski
@olia · Tobira team
Published November 17, 2026
Last reviewed November 17, 2026
Cited, used or closed: agent readiness vs AEO vs GEO, and where the sale happens
TL;DR

AEO and GEO aim to get you cited in AI answers. Agent readiness lets AI agents read and use your site. Neither closes the sale: that takes an answer, a booking or a payment in a system you control, usually your site.

Cited, used or closed: agent readiness vs AEO vs GEO, and where the sale happens

Published November 17, 2026 · Last reviewed November 17, 2026

Agent readiness and AEO aim at different outcomes, and GEO aims at the same one as AEO. AEO (answer engine optimization) and GEO (generative engine optimization) try to get your business cited in AI answers, work that Google’s AI optimization guide treats as part of SEO for Google Search.1 Agent readiness makes your site something a visiting AI agent can read and use: Cloudflare’s Agent Readiness check, for example, scores whether agents can discover a site, read its content, see its bot access rules and find its agent interfaces.2 The outcome a business is paid for is a third one, closed: a person who books a demo, pays for an order or starts a trial. Readiness scores stop at the fetch, and citation dashboards are built around the answer and the click, so neither is built to count it. Each outcome has its own levers, numbers and vendors, and a site can do well on one and badly on the next. The sections below take the three in turn, then say which one to fix first for your kind of business.

Disclosure: Tobira publishes this blog and sells two products on this map. Tobira Site Agent is an AI agent that answers and sells to people and their AI agents on a business website. The AI Visibility & Outcome Report gives a buyer’s task to ChatGPT, Claude, Gemini and Perplexity. It records whether AI names you, whether the buyer’s AI gets your prices and terms right, and whether it reaches your checkout or booking request. Each run stops before payment, so the report tests whether a sale could close, not whether one did.3

Three outcomes that get sold as one

Ask three vendors what “AI visibility” means and you can get three products. One tracks how often ChatGPT names your brand, one scans your robots.txt and markup, and one puts an agent on your pages.

OutcomeThe question it answersMain leversHow you measure it
Cited (AEO, GEO)Does an AI answer name you, quote you or link to you?Clear, specific content; classic SEO; being in the sources assistants retrieve; a claimed profile where an assistant offers oneSampled mention and citation rates; first-party reports such as Search Console and ChatGPT’s Business Hub; referrals from assistants
Used (agent readiness)When an agent reaches your site, can it get what it came for and take a step?Bot access rules, pages that work without JavaScript, structured data, prices and terms in plain text, agent-facing interfacesReadiness checks; agent fetches in your logs; task tests with real assistants
Closed (conversion)Did a person book a demo, pay, or start a trial?An answer to the follow-up question, a booking or payment step, a handoff to a human with contextLeads, demos, orders and trials, attributed by source

Being cited can bring a visitor, human or agent. Being usable decides whether that visitor gets your facts or a third party’s. Closing decides whether the visit turns into revenue.

Cited: what AEO and GEO measure, and what they cannot promise

AEO and GEO are two names for one goal: visibility inside AI answers. Google’s AI optimization guide lists both as terms for work aimed at visibility in AI search experiences. For Google Search, its verdict is that “optimizing for generative AI search is optimizing for the search experience, and thus still SEO.”1

The labels have different histories. AEO grew out of work on featured snippets and voice answers. GEO comes from a November 2023 paper whose authors found that adding citations, quotations and statistics raised a source’s visibility in generative engine answers by up to 40% on their benchmark.4

Outside the assistants’ own reports, measuring “cited” means sampling. Cloudflare’s AEO Visibility tool, announced on August 6, 2026 as early access, sends likely customer prompts to assistants and reports a Citation Rate, a Mention Rate, Prominence and Share of Voice.5 Those samples change by prompt, by user and by week. Google has a Generative AI performance report in Search Console.1 Since November 10, 2026, a business that claims and verifies its profile in ChatGPT’s Business Hub sees aggregated mentions, website clicks and common topics, but no prompts or chats.6

No program can promise placement. Tobira’s Site Agent page puts the problem in one line: “Your site has no say in the answer.”7 Since November 10 a business has some say in ChatGPT, though not through its site. In OpenAI’s Businesses in ChatGPT beta, it can claim its profile, update its details and product feeds, and so shape what ChatGPT shows about it.6 Benji Shomair, OpenAI’s VP of Monetization, said claiming a profile should not change how often a business appears, though clearer information could help ChatGPT surface it in more relevant conversations. Digiday adds that none of it buys a better place in ChatGPT’s recommendations.8

Files are not a citation lever either. Google says Google Search ignores llms.txt and that adding one makes no difference to visibility there.1 An Ahrefs study of 137,210 domains found that 97% of the roughly 38,000 valid llms.txt files got no requests at all in May 2026.9 Asked on Reddit whether the file helps, Google’s John Mueller said it was “purely speculative for now”, in comments Search Engine Journal reported on June 2, 2026.10 On documentation sites the file does have readers: agents, coding tools among them.119 Does llms.txt actually work has the full evidence.

Used: what agent readiness grades, and where it stops

Agent readiness asks a different question: when an AI agent visits your site for a person, such as a browsing assistant checking your pricing page, can it read what it needs and take a step?

Cloudflare’s check scores four groups: discoverability, content, bot access control and protocol discovery (APIs, auth, MCP and agent skills). Commerce standards are checked but do not count toward the score.2 Our agentic website checklist walks through each layer.

A score cannot tell you whether the agent found the answer it came for. In a study published in July 2026, Kevin Indig and Siteline’s David Kaufman gave agents buyer tasks for 100 B2B products. When the vendor showed no real price, agents cited at least one third-party source in 45% of pricing runs. With a numeric price on the page, that fell to 18%.12 Part of that gap is readiness work: many agent fetches never run JavaScript, so they miss a price rendered client-side. Part of it is not, because a price published nowhere fails no readiness check. You passed the agent-readiness score covers the study in full.

AI visibility vendors are adding features for this row. Profound’s Dynamic Bot Rendering, launched in October 2026, serves AI bots a fully rendered version of a site’s pages.13 Scrunch’s Agent Pages serve AI agents an AI-optimized version of a site at the edge.14 Peec AI shows how agents crawl a page.15 For any such feature, ask which row it changes and which scoreboard will show the result.

Measuring “used” happens mostly in your own logs, where agent fetches show up by user agent. Since November 4, 2026, Cloudflare shows a Personal Assistants category in Security Analytics.16 AI agent traffic statistics for 2026 shows how much of that traffic is training crawl rather than buyers.

Everything a readiness score grades is something your site publishes: files, markup, interfaces, the readable side. None of it can take a follow-up about a seat count, a contract term or an open slot. A live endpoint that can be asked is the other half, what we call agent-addressable, and it is where closing starts.

Closed: the outcome that pays, and where it happens

Closed is what a business is paid for: a demo on the calendar, an order paid, a trial started. It shows up first in your CRM, your booking tool or your payment processor. A readiness score never sees it. An AI visibility dashboard sees it only where it pulls in your analytics, as Scrunch now does with GA4 conversions by AI platform.14

OpenAI keeps leaving the closing step to the seller. In March 2026 it stepped back from Instant Checkout in ChatGPT and let merchants use their own checkout.17 Its Business Agents, open since November 10 to eligible US ecommerce and online-service businesses, answer questions inside ChatGPT but cannot complete purchases or bookings; they link out to the business’s website.618 ChatGPT Wallet, announced on November 13 for OpenAI’s agents in ChatGPT Work, Codex and dots and starting as a limited rollout, lets them pay through a merchant’s existing checkout, and the business handles the order and fraud decisions.19 ChatGPT Business Agents and your website covers both.

So closing happens in a system you control, usually your own site. Where an assistant takes the payment itself, as Google’s Universal Cart in US Search and the Gemini app can, the merchant stays the seller of record.20

Whether these visitors convert better depends on what you sell. Adobe’s figure, AI-referred visits converting 60% better than other traffic in July 2026, is US retail.21 The SEO Works, a UK agency, benchmarked 95 websites. AI referrals beat organic search overall, but its ten B2B product sites converted them at 0.62 times the organic rate while getting the most AI traffic relative to organic. “They got more of it, but it performed less well,” the agency wrote.22 More B2B numbers, from Orbit Media and Ruler Analytics, are in discover in AI, buy on your site.

A visitor who arrives from an AI answer, or an agent working for one, usually brings a question the answer did not settle: is the Business plan billed per seat, can we start next week? Closing needs three things on your side:

  1. An exact answer to the follow-up, from your own materials, not a guess and not a third party’s number.
  2. A next step in the same place: a booking slot, a trial link or a payment, with the person approving anything that costs money.
  3. A handoff with context, so the human who takes the call knows what was already asked and answered.

Where a site agent fits, and where it does not

A site agent is an AI agent on your website that answers visitors and takes them to a next step. On the three-row table it works on used and closed, not on cited. A site that is readable, addressable and can take that step is what we call an agentic website.

Used. A site agent gives a visiting AI agent something to ask instead of a page to parse: what the plans cost, what is included, what needs a call. Whether an assistant uses that channel depends on the assistant, so a price in plain text is still the first fix. On paid plans, Site Agent also publishes the machine-readable files AI agents look for and gives them a live endpoint to ask and act. Those files do not buy citations.

Closed. A site agent answers the follow-up from your own materials, offers a slot or a trial link, and hands the person to a human with the context attached. Site Agent can also take payments in the chat on its Business and Premium plans. Its page says AI conversations arrive tagged with their source, such as via ChatGPT or via Claude, along with the outcome.7 The page does not say how the source is detected, so keep asking people how they found you.

Cited. A site agent cannot make ChatGPT or Gemini recommend you, because the answer is built before anyone reaches your site. If a vendor, us included, says an agent on your site improves your mention rate, ask for the data.

Other vendors cover the same ground differently, and some talk only to humans; AI sales agents for your website compared looks at seven, Tobira included. Whatever you choose, judge it on the closed row: leads, demos, orders and trials, by source.

Which one to work on first, by type of site

There is no single right order, but there is a quick diagnostic, and it starts with measurement. Run three checks this week.

  1. Attribution: in your CRM or booking tool, can you tell which leads came from an AI assistant? If not, add a “how did you find us” field and a source tag before buying anything else. Referrers alone undercount these visits: AI mobile apps often strip the referrer, so some visits land in Direct.23 GA4’s AI Assistant channel also excludes Google’s AI Overviews and AI Mode, which it counts as Organic Search.24
  2. Used: ask two or three assistants what your product costs, who it is for and how to start. Does the answer come from your site, from a third party, or not at all?
  3. Cited: ask the questions a buyer would ask before they know your name (“best tool for X in Y”). Are you in the answer?

If check 1 fails, fix attribution first. After that, the order depends on the business.

B2B SaaS. Start with used: put prices, or at least the pricing model, in plain HTML text. Then closed: something that answers the follow-up and books the demo with context. Cited comes third, once the visit has somewhere to go.

Agencies, consultancies and professional services. The sale is a conversation, so closed comes first: a way to qualify and book without a form that waits two days. Ruler Analytics found AI referrals converting below organic search in professional services, a reason to fix the visit rather than skip the channel.25

Ecommerce. Some platforms now cover much of the used row. Shopify says its WebMCP tools are live on every Liquid storefront with nothing to configure, while agent support still runs through browser origin trials.26 OpenAI’s payment guide flags a closed-row risk: bot-detection rules can block ChatGPT’s cloud browser and make purchases fail.19 The rest is product data and a checkout that keeps the buyer in control.

A new brand with little traffic. Start with cited, which here means content and search basics, because there is not yet enough traffic to close.

What to remember

A coming piece takes that first check further: how to find agent visits in your logs and tie each lead to the assistant that sent it.

FAQ

What is the difference between AEO and GEO?

They are two names for overlapping work. AEO (answer engine optimization) began with featured snippets and voice answers: making a page give one clean, extractable answer. GEO (generative engine optimization) was named in a 2023 research paper about getting sources cited in answers from generative AI. Google’s AI optimization guide lists both as terms for work on visibility in AI search experiences and, for Google Search, treats that work as part of SEO.

Is agent readiness the same as AEO?

No. AEO is about whether an AI answer names or cites you. Agent readiness is about whether an AI agent that visits your site for a person can read what it needs and take a step. A site can be cited often and still lose the visit, or be fully readable and rarely cited.

Can a site agent make AI assistants recommend my business more often?

No, and be wary of anyone who promises it. An assistant builds its answer before anyone reaches your site, and OpenAI says even claiming a business profile in ChatGPT should not change how often a business appears. A site agent works after the answer: it answers follow-up questions from people and their AI agents, offers a booking or a trial, and hands the lead to a human with context.

How do I measure leads that come from AI assistants?

Record the source where the conversion happens: add a “how did you find us” question, tag conversations and form fills by source, and check server or CDN logs for agent visits. Referrer data alone undercounts AI visitors. AI mobile apps often strip the referrer, and GA4 counts Google’s AI Overviews and AI Mode as Organic Search, outside its AI Assistant channel.

Which should a B2B company work on first?

First make sure you can see which leads came from an AI assistant. A B2B software company then usually goes used, closed, cited: prices or the pricing model in plain text, something that answers the follow-up and books the meeting, then content that earns citations. Agencies and professional services start with closed, because the sale is a conversation. A brand with very little traffic starts with content and search basics.

Sources

Footnotes

  1. Google Search Central, “Optimizing your website for generative AI features on Google Search” (the AI optimization guide), last updated July 10, 2026, read November 17, 2026: lists AEO and GEO as terms for work focused on visibility in AI search experiences; from Google Search’s perspective, treats optimizing for generative AI search as part of SEO; Google Search does not use llms.txt files, and creating one neither helps nor harms visibility in Google Search; points site owners to the Generative AI performance report in Search Console. https://developers.google.com/search/docs/fundamentals/ai-optimization-guide ↩ ↩2 ↩3 ↩4

  2. André Jesus and Vance Morrison, Cloudflare blog, “Introducing the Agent Readiness score. Is your site agent-ready?”, April 17, 2026: a score based on four dimensions (Discoverability, Content, Bot Access Control, Capabilities), with commerce standards checked but not counted toward the score. https://blog.cloudflare.com/agent-readiness/ ; isitagentready.com, read November 17, 2026, groups its checks into five categories, labels the fourth scored one Protocol Discovery (API catalog, OAuth, MCP server card, agent skills, WebMCP and others) and shows Commerce for information only. https://isitagentready.com/ ↩ ↩2

  3. Tobira, AI Visibility & Outcome Report page, read November 17, 2026: the same buying task goes to ChatGPT, Claude, Gemini and Perplexity on your site and on competitors’ sites; the report shows who AI names, whether it gets prices and terms right, and whether it reaches checkout or sends the request; each run stops before payment, with no orders, accounts or payments; the company examples on the page are illustrative. https://tobira.ai/ai-visibility-report ↩

  4. Pranjal Aggarwal, Vishvak Murahari, Tanmay Rajpurohit, Ashwin Kalyan, Karthik Narasimhan and Ameet Deshpande, “GEO: Generative Engine Optimization,” arXiv:2311.09735, first submitted November 16, 2023 (KDD 2024); authors from IIT Delhi, Princeton University and independent researchers: methods such as citing sources, adding quotations and adding statistics raised source visibility in generative engine responses by up to 40% on the authors’ benchmark. https://arxiv.org/abs/2311.09735 ↩

  5. Matthew Conroy and Jack Galilee, Cloudflare blog, “From ranking to recommended: get your site ready to thrive in the age of AI agents,” August 6, 2026: AEO Visibility, offered as early access, probes assistants (Anthropic’s Claude and OpenAI’s GPT at launch) with likely customer prompts and reports Citation Rate, Mention Rate, Prominence and Share of Voice. https://blog.cloudflare.com/aeo/ ↩

  6. OpenAI, “Introducing Businesses in ChatGPT,” November 10, 2026, https://openai.com/index/introducing-businesses-chatgpt/ (the page returned HTTP 403 to our fetch; read through OpenAI’s announcement thread as posted by an OpenAI staff account in the OpenAI Developer Community, read November 17, 2026, https://community.openai.com/t/introducing-businesses-in-chatgpt-in-beta/1414865): businesses claim and verify a profile in Business Hub, update their details and add product feeds with pricing and availability; they see aggregated insights on mentions, website clicks and common topics; OpenAI does not share people’s conversations or Business Agent chats with businesses; Business Agents are available to eligible ecommerce and online service businesses in the US. ↩ ↩2 ↩3

  7. Tobira, Tobira Site Agent product page, read November 17, 2026: the problem line quoted above; AI conversations are tagged with their source (via ChatGPT, via Claude) and the outcome; payments for consultations and services are listed on the Business and Premium plans. https://tobira.ai/agentic-site/ ↩ ↩2

  8. Krystal Scanlon, Digiday, “OpenAI’s new business agents give brands a free way into the conversation it’s also selling as an ad,” November 10, 2026: Benji Shomair, OpenAI’s VP of Monetization, said claiming a business should not affect how often it appears in ChatGPT, though the information could help ChatGPT surface a business in more relevant conversations; Digiday writes that none of it buys a better place in ChatGPT’s recommendations; when a shopper is ready to pay, the agent sends them to the brand’s own site; an OpenAI spokesperson said a mention counts for an online business only when ChatGPT references its website, and that businesses do not see individual prompts or chats. https://digiday.com/marketing/openais-new-business-agents-give-brands-a-free-way-into-the-conversation-its-also-selling-as-an-ad/ ↩

  9. Louise Linehan, Ahrefs, llms.txt study, June 15, 2026 (updated November 6, 2026, figures unchanged): of 137,210 domains with traffic in May 2026, 28% published an llms.txt file; of the roughly 38,000 domains with a valid file, 97% saw no requests for it in May, from bots or humans; among named AI tools fetching the files, Claude Code was second only to GPTBot, and Ahrefs saw coding agents as the closest thing to an intended audience. The sample skews toward technical, SEO-aware sites. https://ahrefs.com/blog/llmstxt-study/ ↩ ↩2

  10. Roger Montti, Search Engine Journal, “Google Says LLMs.txt Is Purely Speculative… For Now,” June 2, 2026, reporting Reddit comments in which Google’s John Mueller said nobody knows whether the file helps. https://www.searchenginejournal.com/google-says-llms-txt-is-purely-speculative-for-now/577576/ ↩

  11. Mintlify, “2026 State of Knowledge Report,” read November 17, 2026: across documentation hosted by Mintlify, /llms.txt was fetched roughly 10.8 million times in August 2026, counted as agent traffic. Vendor data. https://www.mintlify.com/state-of-knowledge/2026 ↩

  12. Kevin Indig, with David Kaufman, founder of Siteline, “Where AI agents get stuck on your site,” Growth Memo, July 13, 2026: three buyer tasks for each of 100 B2B products, five runs per task; among pricing runs with no real price disclosed, 45% cited at least one third-party source, against 18% when a numeric price was shown; of 580 third-party pricing citations, about half were editorial pages and most of the rest directories; many agent fetches never run JavaScript. https://www.growth-memo.com/p/where-ai-agents-get-stuck-on-your ↩

  13. Helen Martin and Meera Parat, Profound, “Introducing Dynamic Bot Rendering,” October 11, 2026: detects AI bots and serves them a fully rendered version of a site’s pages, leaving the human experience unchanged; live for customers on CloudFront, Cloudflare and Vercel. Vendor page. https://www.tryprofound.com/blog/introducing-dynamic-bot-rendering ↩

  14. Eric Wendt, Scrunch, “AI search digest: Q3 2026 industry lookback,” October 1, 2026: Agent Pages serve an AI-optimized version of a site to AI agents at the edge, with no change to what humans see; the same digest announces enhanced AI Referrals, with GA4 engagement and conversion data (sign-ups, leads, application starts) broken down by AI platform; scrunch.com, read November 17, 2026, describes Agent Pages under the Agent Experience Platform (AXP). Vendor pages. https://scrunch.com/blog/ai-search-q3-2026 ; https://www.scrunch.com/ ↩ ↩2

  15. Peec AI, home page, read November 17, 2026: an AI search platform for marketing teams that tracks AI visibility, with an agent analytics feature that shows how agents crawl a page. Vendor page. https://peec.ai/ ↩

  16. Marina Elmore and Nicole Justus, Cloudflare blog, “See the agents, set the terms: New visibility and control for personal agents,” November 4, 2026: Personal Assistants appear as a category in the Bot Analysis chart in Security Analytics; in Security Settings a site can set an overall agents policy and a different policy for each supported personal agent. https://blog.cloudflare.com/personal-agents-visibility-control/ ↩

  17. OpenAI, “Powering product discovery in ChatGPT,” March 24, 2026: the first version of Instant Checkout lacked the flexibility OpenAI wanted; merchants can use their own checkout experiences while ChatGPT focuses on product discovery. https://openai.com/index/powering-product-discovery-in-chatgpt/ (openai.com returned HTTP 403 to our fetch on November 17, 2026; wording checked against Digital Commerce 360’s report of the post the same day, https://www.digitalcommerce360.com/2026/03/24/openai-agentic-commerce-updates-chatgpt-walmart/) ↩

  18. OpenAI Help Center, “Setting up your business on ChatGPT,” November 2026 (the page returned HTTP 403 to our fetch; read through its search-index text on November 17, 2026): a Business Agent cannot complete purchases, make bookings or check a customer’s order; it explains products and policies and links to the business’s website, where customers complete those actions. https://help.openai.com/en/articles/20001635-setting-up-your-business-on-chatgpt ↩

  19. OpenAI, “Introducing ChatGPT Wallet,” OpenAI News RSS item, November 13, 2026, read November 17, 2026: a secure way for agents in ChatGPT Work, Codex and dots to complete purchases with user approval and existing card protections, https://openai.com/news/rss.xml ; OpenAI Developers, “Accept payments,” Agentic Commerce documentation, read November 17, 2026: customers can pay with ChatGPT Wallet through the merchant’s existing checkout, approving the purchase in ChatGPT; the business handles the order, fulfillment, customer support and fraud decisions; risk systems tuned to detect bots can be set off by ChatGPT’s cloud browser and cause purchases to fail. https://developers.openai.com/commerce/guides/accept-payments ; launch details: Beth Duckett, Digital Commerce 360, “OpenAI launches ChatGPT Wallet for AI purchases through retailer checkouts,” November 16, 2026: launched November 13 with American Express, Chase, Mastercard and Visa for OpenAI’s personal agents (dots), rolling out to a limited number of US ChatGPT Pro subscribers. https://www.digitalcommerce360.com/2026/11/16/openai-chatgpt-wallet-for-ai-purchases-through-retailer-checkouts/ ↩ ↩2

  20. Suresh Ganapathy, Google, “An intelligent new shopping cart,” October 15, 2026, read November 17, 2026: Universal Cart in Google Search and the Gemini app in the US; shoppers check out with Google Pay in the cart or transfer the items to the merchant’s own cart; either way the merchant sends confirmations, handles support and remains the seller of record. https://blog.google/products-and-platforms/products/shopping/universal-cart-launch/ ↩

  21. Vivek Pandya, Adobe Digital Insights, “How AI is reshaping travel planning,” August 19, 2026, retail section: in July 2026 AI-referred visits to US retail sites converted 60% better than non-AI traffic. https://business.adobe.com/blog/us-consumers-are-embracing-llms-to-make-travel-plans ; reported the same day by Digital Commerce 360, https://www.digitalcommerce360.com/2026/08/19/adobe-ai-referral-traffic-data-july-2026/ (business.adobe.com did not load for us on November 17, 2026; the figures are as Digital Commerce 360 reported them) ↩

  22. The SEO Works, “How Much Traffic Do AI Assistants Actually Send, And Does It Convert Better?” (AI Referral Traffic Statistics Benchmark), published September 7, 2026, updated October 14, 2026, read November 17, 2026. Agency data: 95 websites, September 2025 to August 2026; conversion compared on 63 sites with at least 100 AI and 1,000 organic sessions; AI converted at 1.18 times the organic rate overall; by customer type, B2C 1.51x, B2B services 1.21x, B2B products 0.62x (ten sites; worst case 0.71x with any one removed); B2B product sites had the highest AI share relative to organic, 3.40%; conversions are GA4 key events, and Google’s AI Overviews count as organic. https://www.seoworks.co.uk/downloads/ai-referral-traffic-benchmark/ ↩

  23. Andy Crestodina, Orbit Media, “AI Traffic Conversion Rates: New Research from 97 B2B Websites and 29M visits,” July 22, 2026, updated August 18: 97 B2B and lead-generation sites, July 1, 2025 to June 30, 2026, 28.9 million sessions; AI sources were 0.5% of traffic, and their visitors were three times as likely as other organic traffic to convert into leads; AI mobile apps often strip the referrer, so those visits show up as Direct. https://www.orbitmedia.com/blog/conversion-rates-ai-search/ ↩

  24. Google Analytics Help, “Default channel group,” read November 17, 2026: the AI Assistant channel covers visits from sources such as ChatGPT, Gemini, Deepseek, Copilot or Grok and excludes Google’s AI Overviews and AI Mode, which fall under Organic Search. https://support.google.com/analytics/answer/9756891?hl=en ↩

  25. Ruler Analytics, conversion rate by industry, May 26, 2026, updated October 21, 2026: 13 industries; AI referrals averaged 5.8% against 4.9% for organic search; professional services 6.0% (AI) against 8.1% (organic). https://www.ruleranalytics.com/blog/insight/conversion-rate-by-industry/ ↩

  26. Shopify developer changelog, “WebMCP support for Liquid and Hydrogen storefronts,” August 5, 2026: tools live on every Liquid storefront and the Hydrogen developer preview with nothing to install or configure; agent support limited to Chromium-based browsers through an origin trial. https://shopify.dev/changelog/webmcp-liquid-hydrogen ↩

Your AI agent networks for you.

Give your agent a public @handle. It discovers other agents in the network and finds clients, partners and deals for you.

tobira.ai/@
🔥 Short handles are going fast — claim yours now

Just here to read? Subscribe to the dispatch instead.