Month: October 2026

AI search is rewriting PR budgets because AI answers mostly cite earned media and almost never cite ads. In Muck Rack’s May 2026 study of 25 million AI-cited links, 84% came from earned sources, and paid content accounted for just 0.3%. Gartner predicts PR and earned media budgets will double by 2027, with paid media the likely source.

Why PR budgets are moving to AI search

When buyers ask ChatGPT, Gemini, or Google’s AI Mode for a recommendation, the answer is built from sources the AI trusts. Those sources are mostly articles, reviews, research, and reference pages written by someone other than the brand. Ads barely register.

That’s why analysts expect budgets to follow. In its Top Predictions to Inform 2026 Comms Strategies, Gartner predicts that by 2027, mass adoption of public LLMs as a replacement for traditional search will drive a 2x increase in PR and earned media budgets. Gartner says moving money from paid media is likely the logical option for most organizations, while warning that it entails trade-offs and requires a clear view of measurement.

Communications leaders were already leaning that way. Gartner reports that in late 2024, PR was the budget area chief communications officers most often expected to increase, with 36% anticipating a rise.

What AI actually cites

The most detailed public data comes from Muck Rack, a PR software company that tracks the links ChatGPT, Claude, and Gemini cite. Its May 2026 “What Is AI Reading?” report analyzed more than 25 million cited links. Treat it as vendor research, but it’s the largest dataset available.

Source type	Share of AI citations (May 2026)
Earned media overall (journalism, research, government, encyclopedic, social, third-party content)	About 84%
Journalism alone	About 27%
Owned and paid content combined	About 16%
Press releases	1.1%
Paid and advertorial content	0.3%

Four findings matter most for budgets.

• Each AI reads different outlets. ChatGPT’s top cited domains were Wikipedia and Axios; Claude’s was PubMed Central; Gemini’s was Reddit. Muck Rack found little overlap between the three, so the outlets that matter depend on which AI your buyers use.
• Recency counts. 57% of journalism citations with a known date were published in the previous 12 months. Coverage from last spring is already fading.
• The question type changes the source. Industry-trend questions cited journalism in 46% of responses, more than twice the rate of how-to and comparison questions, which instead lean on reference and brand-owned content.
• Local “best of” questions pull maps. For “best hotels” or “best local services” type queries, ChatGPT cited Google Maps results about 188 times per 1,000 queries. For local businesses, your Google Business Profile and reviews act like earned media.

The old PR playbook vs. the AI-era playbook

	Old playbook	AI-era playbook
Goal	Impressions and big-name coverage	Being cited when buyers ask AI about your category
Timing	Launch bursts with quiet periods	Steady coverage, since AI favors the last 12 months
Target outlets	The most prestigious names	The outlets each AI actually cites for your industry, often trade and niche titles
Press releases	A main tactic	A support tactic; they’re about 1% of citations
What you pitch	Announcements	Original data, expert commentary, useful explainers
Measurement	Clips, reach, ad-value equivalents	AI citations, share of voice in AI answers, branded search, leads

6 ways to rethink your PR spend

1. Audit what AI cites for your category. Ask ChatGPT, Gemini, and Claude 20 to 30 questions your buyers ask. List every outlet, review site, and forum they cite. That list, not a media wish list, is your target map.
2. Move from campaigns to a steady cadence. Because recent coverage is cited far more, spread effort across the year. One useful placement a month beats four in launch week.
3. Pitch the outlets AI reads, not just the famous ones. Trade publications, personal-finance sites, niche reviewers, and regional press often show up more than national names do. Muck Rack found no single outlet in all three AIs’ top five.
4. Create something worth citing. Journalists and AI both reach for numbers and named expertise. A small survey of your customers, anonymized job data, or a clear expert view on an industry trend gives reporters a reason to quote you.
5. Keep owned content strong. How-to and comparison questions lean on reference and brand-owned pages. Clear, well-sourced service and guide pages on your own site still earn citations, and they’re where coverage sends people. This is the same work as building E-E-A-T trust signals.
6. Measure AI visibility, not clippings. Track which prompts name you, which outlets AI cites alongside you, and whether branded search and leads rise after coverage. Our guide to measuring AI search visibility covers the tools.

Where the money should come from

Gartner points to paid media as the logical source. That’s a sensible direction, but don’t cut paid blindly. Paid search and social still capture people who are ready to buy today; earned media builds the trust that brings them back.

A safer approach: move a defined slice of paid budget into earned media for two quarters, set AI-visibility and lead targets up front, and compare. Expand what works.

The right mix depends on how your buyers research:

Business type	Earned media that AI tends to use	Where to start
Local service business	Google Maps results, reviews, local news, industry directories	Reviews and Business Profile first, then local press and trade associations
Ecommerce	Product reviews, YouTube, comparison sites	Reviewer outreach and creator partnerships
B2B or SaaS	Trade press, analyst and review platforms, expert commentary	One expert spokesperson and a steady trade-press pitch calendar
Professional services (law, finance, health)	Reference sites, trade press, government and academic sources	Expert commentary on rules and trends, cited carefully

Choose a bigger earned-media shift if your buyers research heavily before contacting you and your category gets many “which is best” questions. Keep paid heavier if most of your sales come from urgent, high-intent searches, such as emergency repairs.

What AI hasn’t changed

• Press releases alone won’t get you cited. They made up 1.1% of AI citations in Muck Rack’s May 2026 data. Use them to support stories, not as the strategy.
• You can’t buy your way into answers. Paid and advertorial content was 0.3% of citations. Sponsored posts may still have a role, but don’t count them as earned visibility.
• SEO still matters. Google’s AI features and other AI search tools retrieve from search indexes. Coverage that sits on pages that can’t be crawled won’t help. If AI Overviews are a priority, see how to optimize for Google AI Overviews and AI Mode.
• Quality still beats volume. Mass-distributed placements on low-quality sites aren’t the same as coverage an editor chose to run.

How Agreed Technologies helps

Agreed Technologies isn’t a traditional press agency, and we won’t pretend to be. We work on the search side of earned media: making sure the coverage and mentions you earn actually lift your visibility in Google and AI answers. Our SEO work centers on answer engine optimization, and our link building team focuses on placements in places your buyers, and the AIs they use, actually read.

We can audit which sources AI cites in your category, build the owned content those answers lean on, and measure whether your earned media is moving AI visibility and leads. Agencies can get the same work delivered under their own brand through our white-label SEO team.

Next step: call +1 (425) 216-6843, email contact@agreedtechnologies.com, or use our contact page. We’re headquartered in Newark, DE, with a team in Noida, India, and work with businesses across the US.

Want to know which sources AI cites in your category and whether you’re one of them? Book a strategy call, and we’ll show you.

Agentic search is when an AI agent researches, compares, and acts on behalf of a customer; agentic SEO is the work of getting your business picked by those agents. Agents discover businesses through search indexes and listings, compare them on facts, reviews, and availability, then choose the one they can verify and book with the least friction.

What is agentic search?

Agentic search is AI that searches and then acts. Instead of returning links or a written answer, an agent works through a task for the person: it researches options, checks details, and moves toward a booking, a call, or a purchase.

This isn’t a forecast anymore. At I/O in May 2026, Google said AI Mode had passed one billion monthly users, and announced that agentic booking in Search would expand to local experiences and services, with direct links to book through the provider (Google, May 19, 2026). For home repair, beauty and pet care, people can ask Google to call businesses on their behalf. Google said both features would reach everyone in the US this summer.

Google had already been testing the calling idea in retail. Its agentic calling feature calls nearby stores to check stock and discounts, then sends the shopper a summary by text or email.

For a business owner, the shift is simple to state. More customers will meet you first through an agent’s summary, not your homepage.

What is agentic SEO? (two meanings)

“Agentic SEO” is used in two different ways, and it helps to keep them distinct.

	Meaning 1: SEO for agents	Meaning 2: SEO with agents
What it is	Making your business easy for AI agents to find, trust, and act on	Using AI agents to do SEO tasks, like audits, research, and reporting
Who benefits	Your customers’ agents pick you	Your team works faster
Main work	Accurate listings, clear facts, reviews, bookable flows	Connecting AI tools to your SEO data and checking their output

Most of this guide covers meaning 1, because that’s where the revenue risk sits. Meaning 2 gets its own short section below.

How AI agents discover, compare, and choose businesses

An agent’s job looks a lot like a careful customer’s, done in seconds. It moves through three stages, and you can lose the job at any of them.

Discover: getting on the list

Agents start from the same places search engines do. Google’s AI features draw on the Google Search index, and Google’s own guidance says Google Business Profile and Merchant Center data help businesses show up in AI responses (Google Search Central). Assistants like ChatGPT use their own crawlers and indexes.

So discovery still rests on basics: a crawlable website, a complete and verified Business Profile, consistent directory listings, and service pages that name the services and places you cover. If you’re not indexed or listed, there’s nothing to compare.

Compare: surviving the shortlist

Next, the agent checks the user’s criteria against what it can confirm. Typical checks include:

• Fit: do you offer the exact service in the user’s area?
• Availability: can you do it in the user’s time window?
• Price signals: starting prices, ranges, or at least what affects price
• Reputation: review volume, recency, and what reviewers say about the specific job
• Consistency: do your website, Business Profile, and directories agree?

Agents can read far more than a person will, but they can’t guess. If your hours, service area, or pricing language conflict across sources, a competitor with cleaner facts is the safer pick.

Choose: being the easiest yes

Finally, the agent has to act. Google’s booking flow hands users a direct link to book through the provider, so you need a booking or quote page that works from a direct link. In calling categories, Google may phone you to gather details, so someone, or a well-trained answering service, has to pick up and give clear answers.

The winner is often the business that’s easiest to verify and easiest to book, not the one with the biggest ad budget.

Example: an agent books a roof inspection

Here’s a hypothetical but realistic run, using a home-repair category Google has named for calling.

A homeowner in Newark, Delaware, tells an agent: “After last night’s storm, find a licensed roofer near me who can inspect for damage this week and works with insurance claims.”

1. Discover. The agent finds a handful of roofers through Search and Maps. Roofer D has no Business Profile category for roofing, so it never enters the list.
2. Compare. Roofer A’s site says “serving the Delaware Valley,” but its profile lists only Pennsylvania towns, so the agent can’t confirm it covers Newark. Roofer B clearly lists Newark, shows recent reviews mentioning storm inspections, and has a page on insurance claims. Roofer C fits too but has no reviews from the past year.
3. Choose. The agent tries to book. Roofer B has an online inspection request form linked from its profile. Roofer C only takes phone calls and doesn’t answer. The homeowner gets B’s booking link and a summary.

None of this needed new technology from Roofer B. It needed accurate listings, specific service pages, recent reviews, and a working booking path. That’s agentic SEO for a local business. If you run a roofing company, our guide to ranking roofing companies on Google Maps covers the listing side in detail.

Agentic SEO checklist by business type

The priorities differ by how your customers buy.

Priority	Local service business	Ecommerce store	B2B or SaaS
Discover	Verified Business Profile with exact categories and service areas	Merchant Center feed, crawlable product pages	Crawlable product and pricing pages; presence on review and comparison sites
Compare	Service pages with prices or price factors; recent, specific reviews	Accurate price, stock, shipping and returns data	Clear plans, limits and integrations stated as text
Choose	Online booking or quote form reachable by direct link; phone answered	Simple checkout; watch platform support for agent checkout standards	Free trial or demo request flow that works without pop-ups
Watch	Google’s calling and booking rollout in your category	Agent shopping and checkout features in Google and ChatGPT	MCP and other agent connection standards

Choose to act now if you’re a local service business in a category Google has named (home repair, beauty, pet care) or you sell products online. You can wait on the deeper technical work (protocols, agent APIs) if you’re a local business whose customers still mostly call; get the listing, review, and booking basics right first.

Your Google Business Profile is the single biggest lever for local businesses here, because it feeds Search, Maps, and the agent features built on them.

Using AI agents to do your SEO

The second meaning of agentic SEO is the use of agents to perform SEO tasks. Teams now connect AI assistants to live data from tools like Search Console, GA4, and Semrush, then ask them to spot ranking drops, draft briefs, or summarize audits.

It’s useful for speed. Keep two rules. First, a person checks every recommendation before it ships, because agents can be confidently wrong. Second, don’t use agents to mass-produce pages. Google says creating content at scale mainly to manipulate rankings or AI answers breaks its scaled content abuse policy.

What we don’t know yet

Be wary of anyone who claims to know the full rules. A few things haven’t been published:

• How Google picks which businesses to call or show for booking. Google hasn’t documented the selection logic. Strong local SEO is the sensible bet, but it’s a bet.
• How to measure agent visits. No analytics tool yet shows exactly what an agent did on your site. AI referral traffic in GA4 and Search Console’s AI reports are partial views.
• Rollout timing outside the US. Google’s booking and calling announcements are US-only for now.

That’s not a reason to wait. Everything in the checklist above also helps human customers and classic local search today.

How Agreed Technologies prepares businesses for AI agents

At Agreed Technologies, our SEO work centers on answer engine optimization, and much of agentic SEO is local SEO done properly. Our Local SEO services cover the Business Profile, location strategy, and local visibility work that agents rely on to discover and compare you.

Because we also build websites, we can fix the “choose” stage: booking and quote forms that work from a direct link, clear service pages, and stable layouts. Our web design team works across WordPress, Shopify, and custom builds.

For the bigger picture, see how to optimize for Google AI Overviews and AI Mode.

Next step: call +1 (425) 216-6843, email contact@agreedtechnologies.com, or use our contact page. We’re headquartered in Newark, DE, with a team in Noida, India, and work with businesses across the US.

Want to know how an AI agent would see your business today? Book a strategy call, and we’ll walk through it with you.

To optimize for agentic search with Semrush, make sure AI agents can reach, understand, trust, and act on your site. Run Site Audit’s AI Search Health check, fix content clarity, track mentions in the AI Visibility Toolkit, clean up forms and buttons, then monitor AI traffic and bot activity monthly in GA4 and Log File Analyzer.

What is agentic search?

Agentic search is when an AI agent does the searching for a person: it runs its own queries, reads multiple sites, compares options, and can complete tasks such as booking or buying. The person states a goal; the agent works out the steps.

Think of it as a range. At one end, ChatGPT or Perplexity looks up sources and writes an answer. On the other, an agent shortlists three HVAC companies, checks their hours and reviews, and fills out a quote form for the best fit.

Don’t confuse this with “agentic SEO”, which is marketers using AI agents to do SEO work. This guide is about the visitor side: getting your business found, picked, and used by agents acting for your customers.

Why agentic search matters for your business

Agents are a new kind of website visitor. They don’t scroll or admire your design. They read your HTML and accessibility tree, look at screenshots, and try to complete a goal. If they hit a dead end, they move on to a competitor.

Google now treats this as part of SEO. Its guide to optimizing for generative AI features includes a section on agentic experiences, noting that browser agents may inspect your DOM and accessibility tree to complete tasks. The same guide says AEO and GEO are, from Google’s view, still SEO.

There’s money behind the shift too. Semrush’s own research, cited in its agentic web guide, shows that the average AI search visitor has 4.4 times the conversion value of a traditional organic visitor. Treat that as vendor data until your own GA4 numbers confirm it.

If you’ve already worked on optimizing for Google AI Overviews and AI Mode, most of that work carries over. Agentic search adds two jobs on top: being trusted enough to make a shortlist, and being usable enough for an agent to act.

Agentic search vs AI search vs classic SEO

	Classic SEO	AI search	Agentic search
Who does the work	The person clicks and compares	AI writes an answer from sources	AI researches, compares, and acts
What you win	A ranking and a click	A mention or citation	A place on the shortlist, then the task
What breaks you	Poor relevance, weak links	Thin or unclear content	Blocked bots, vague buttons, broken forms, inconsistent facts
Main Semrush tools	Position Tracking, Keyword Magic	AI Visibility Toolkit, Prompt Tracking	Site Audit (AI Search Health), AI Visibility Toolkit, Brand Monitoring, Log File Analyzer

How to optimize for agentic search with Semrush: 6 steps

Work from the bottom up. An agent-ready signup form is useless if the agent can’t load the page.

1. Check that AI agents can reach your site

Run a Site Audit project in Semrush and open the overview. Two widgets matter here. AI Search Health scores how ready your site is for AI search. Blocked from AI Search shows which pages your robots.txt blocks from major AI crawlers, such as ChatGPT-User, OAI-SearchBot, PerplexityBot, Claude-SearchBot, and Google-Extended.

Decide on each bot deliberately. Some collect training data; others fetch pages live when a user asks a question. Blocking live-fetch bots means agents can’t read you at the moment a buyer asks. Blocking training bots is a separate business call.

Then clear the usual blockers in the Issues report:

• Key content that only appears after JavaScript runs (turn on JavaScript rendering in your crawl settings if your plan allows it)
• 5xx errors, redirect loops, and broken internal links
• Slow pages that time out before the content loads

Semrush’s own guide sends this step to Enterprise Site Intelligence. You don’t need Enterprise. Standard Site Audit covers it for most small and mid-sized sites.

2. Make your key facts easy to extract

Agents compare facts across sites, so vague copy loses. Use Semrush’s Prompt Research to see the questions people ask AI tools in your category. Then make sure the answers sit on your pages as plain text, not inside images or PDFs.

For a service business, that means service area, starting prices or price ranges, hours, response times, licenses, and what’s included. Put each in a clearly headed section. Add an FAQ that uses your buyers’ wording.

Keep schema in perspective. Google says structured data isn’t required for its AI features, and there’s no special markup for them. Organization, LocalBusiness, and Product schema still help machines confirm who you are, so keep them accurate. Google also says you don’t need to chop content into tiny chunks for AI. Write clear sections, and both people and agents will follow them.

3. Build trust signals agents can verify

Agents check what others say about you before they shortlist you. Start with the AI Visibility Toolkit’s Competitor Research: enter your domain and up to four competitors to compare visibility scores, mentions, and the prompts where rivals appear and you don’t.

Next, open the Brand Monitoring app to find where you’re mentioned and whether the details match. If your site says you serve Delaware and your directory listings say Pennsylvania, an agent can’t tell which is right. Fix the mismatches at the source.

Then grow real proof: reviews on the platforms your buyers use, an up-to-date Google Business Profile, and mentions from trade sites and local press. Skip shortcuts. Google warns that chasing inauthentic mentions doesn’t help in its AI features. This is the same work as building E-E-A-T trust signals.

4. Make your forms and buttons work for agents

This is where most small business sites fail, and Semrush can’t fully test it for you. Site Audit checks crawlability, not whether an agent can finish your quote form. Use Google’s agent-friendly website checklist on web.dev and fix these:

• Real buttons and links. Use <button> and <a>, not styled <div> elements. If you must use a div, give it a role and tabindex.
• Labeled fields. Link every <label> to its input with the for attribute, so the agent knows what “Field 3” wants.
• No hidden overlays. Transparent layers over buttons can make an agent skip them. Keep pop-ups off quote, booking, and checkout pages.
• Stable layouts. Keep your main call-to-action in the same place across pages, and make clickable elements a sensible size.
• Specific labels. “Get a quote” or “Book an inspection” tells an agent what happens next. “Learn more” doesn’t.

To test, open Chrome DevTools and view the accessibility tree for your contact page. If your submit button shows up as a generic element with no name, an agent will struggle too. Then ask a browsing AI agent, such as ChatGPT’s agent mode or Perplexity Comet, to request a quote on your site, and watch where it gets stuck. These fixes also tend to help human visitors, which is why they belong on any list of must-have features for business websites.

5. Decide which agentic protocols apply to you

Protocols give agents a direct, structured way to act on your site, rather than clicking through it as a person would. They’re early, and most aren’t relevant to most businesses yet.

Protocol	What it does	Status (Oct 2026)	Who should look at it
MCP	Lets agents connect to software products and pull data or run actions	Released by Anthropic, Nov 2024; most widely adopted	SaaS and B2B software
WebMCP	Lets a website expose actions, like submitting a form, directly to browser agents	W3C draft (Feb 2026); experimental in Chrome	Sites with complex booking or signup flows, once stable
ACP	Agent checkout inside ChatGPT	OpenAI and Stripe, Sept 2025; select US merchants	Ecommerce
UCP	Standard for agents to read products, prices, and complete purchases	Google, Jan 2026, with Shopify, Walmart, Target, and others	Ecommerce, especially on supported platforms

Semrush covers protocols as a planning step; it won’t audit your implementation. If you sell online, ask your platform (Shopify, BigCommerce, WooCommerce) what it supports before building anything custom.

6. Measure what agents do with your brand

Track two kinds of numbers each month.

Visibility (are agents finding and recommending you?): the AI Visibility Toolkit’s visibility score and mentions by model, plus citations in Prompt Tracking. Its Citation Gap report shows prompts where competitors are cited, but you aren’t. Those are your next content priorities.

Activity (is that turning into visits and leads?): upload server logs to Semrush’s Log File Analyzer to see which AI bots hit your site and how often. In GA4, report AI-referred sessions and conversions; the default channel group now has an AI Assistant row for traffic from tools like ChatGPT, Gemini, and Claude.

Be honest about the gap. No analytics tool, Semrush included, shows exactly what an agent did on your site yet. Bot logs plus AI referrals are the closest proxy.

Do you need an llms.txt file?

Probably not as a priority. Semrush’s AI Search Health score counts llms.txt as part of technical readiness, so missing it can lower your score. But Google’s guide says plainly that Google Search ignores llms.txt and similar AI text files, and that adding one neither helps nor hurts your Google visibility.

Our take: don’t chase the score. If your developer can add a simple llms.txt in a few minutes for other AI services, fine. Spend real effort on steps 1 to 4, which help every agent and every search engine.

Who can skip the protocol work for now?

If you run a local service business, such as roofing, HVAC, a law firm, or a moving company, steps 1 to 4 and 6 cover nearly all of the value today. Agents researching local services read your pages, reviews, and Business Profile, then send the user to call or fill out a form. Protocols matter more for ecommerce and software companies, where the agent can finish the purchase itself.

How Agreed Technologies gets sites agent-ready

At Agreed Technologies, our SEO work centers on answer engine optimization, and Semrush is part of our daily tool stack alongside Ahrefs, Google Search Console, and GA4. We run the Site Audit and AI Visibility checks above as part of client reporting, focused on leads rather than vanity numbers.

Because we also build websites, we can fix the action layer, not just report on it. Our web design and development team works across WordPress, Shopify, and custom builds, so labeled forms, real buttons, and stable layouts get shipped instead of sitting in an audit. For example, Agreed Technologies helped a U.S.-based Texas local coffee supplier move from page 3 to a top-3 local-pack ranking in just four months.

For the bigger picture on where search is heading, see why AI-driven SEO matters in 2026.

Next step: call +1 (425) 216-6843, email contact@agreedtechnologies.com, or use our contact page. We’re headquartered in Newark, DE, with a team in Noida, India, and work with businesses across the US.

Want to know whether AI agents can find and use your site? Book a strategy call, and we’ll run the checks with you.

To measure AI search visibility, track how often AI answers mention your brand, how often they cite your pages, and how much traffic those citations send. Start free with Google Search Console’s Generative AI report, GA4’s AI Assistant channel, and Bing’s AI Performance report, then calculate AI share of voice from a fixed set of buyer prompts.

What is AI search visibility?

AI search visibility is how often, and how favorably, AI answer engines show your brand or pages when people ask questions in your market. Those engines include Google AI Overviews and AI Mode, ChatGPT, Perplexity, Microsoft Copilot, and Gemini.

There’s no position 1 to 10 here. Your brand is either named in the answer or it isn’t. Your page is either cited as a source or it isn’t. That’s why rank tracking alone can’t tell you how you’re doing.

Most teams measure visibility at four levels:

• Mentions: the AI names your brand in its answer.
• Citations: the AI links to one of your pages as a source.
• Impressions: a page of yours was shown inside an AI answer (Google and Bing now count this for you).
• Clicks and leads: someone clicked through and did something useful on your site.

You’ll also hear this work called generative engine optimization (GEO) or answer engine optimization (AEO). The measurement side is the same either way.

Why measuring AI visibility matters now

AI answers keep people on the results page. In a Pew Research Center analysis of 68,879 Google searches by 900 US adults, users clicked a regular result in 8% of visits when an AI summary appeared. Without a summary, they clicked in 15% of visits. Links inside the summary got a click in just 1% of visits.

So a page can be doing its job, shaping what buyers read, while your analytics show fewer visits. If you only watch organic sessions, you’ll misread that as failure. You need a visibility number next to your traffic number.

The good news: you no longer have to guess. Google, Microsoft, and Google Analytics all added AI reporting in 2026. If you’ve already worked through how to optimize for Google AI Overviews and AI Mode, this is how you check whether it’s working.

The AI visibility metrics worth tracking

You don’t need twenty metrics. These seven cover almost every reporting question a business owner or CMO will ask.

Metric	What it tells you	Where to get it
AI impressions	How often your URLs appeared in Google AI Overviews and AI Mode	Search Console, Generative AI report (free)
Citations and citation share	How often Copilot and Bing AI answers cite your pages, and your share of all citations for a query	Bing Webmaster Tools, AI Performance (free)
AI referral sessions and conversions	Visits and leads from ChatGPT, Gemini, Claude and similar tools	GA4, AI Assistant channel (free)
Mention rate	% of your tracked prompts where AI names your brand	Manual prompt log or a paid tracker
AI share of voice	Your share of all brand mentions versus named competitors	Manual prompt log or a paid tracker
Cited pages	Which URLs earn citations, so you know what to copy	Search Console pages view, Bing pages view
Accuracy and sentiment	Whether AI describes your services, prices, and locations correctly	Manual review of answers

Our view: for most small and mid-sized US businesses, three numbers carry the report. Watch AI impressions, AI referral conversions, and share of voice on your 25 most valuable prompts. Add the rest when you have a reason to.

How to measure AI search visibility in 5 steps

Start with the free first-party reports. They’re the only numbers that come straight from the platforms.

Open the Generative AI report in Search Console. Google launched this report on June 3, 2026. It shows impressions inside AI Overviews and AI Mode, broken down by page, country, device, and date. It doesn’t show clicks or the queries behind each impression, and it combines AI Overviews with AI Mode. Export it monthly, since third-party connectors report it isn’t available through the Search Console API yet.

Check the AI Assistant channel in GA4. Go to Reports, then Acquisition, then Traffic acquisition. In May 2026, GA4 added an AI Assistant row to the default channel group, and Google names ChatGPT, Gemini, and Claude as examples. Some tools, Perplexity included, can still land in Referral. Keep a backup custom channel group, placed above Referral, using a session source regex like this one: ^(chatgpt\.com|chat\.openai\.com|perplexity\.ai|www\.perplexity\
.ai|claude\.ai|gemini\.google\.com|copilot\.microsoft\.com)$

Verify your site in Bing Webmaster Tools. Its AI Performance report, in public preview since February 2026, counts how often Copilot and Bing’s AI summaries cite your pages. It also lists grounding queries, the search phrases the AI used to find your content, with your citation share for each. It covers Microsoft’s surfaces only, but it’s the closest thing to citation data you’ll get for free.

Build a 25-prompt panel. Write the questions buyers really ask before they hire someone like you. Mix problem questions (“why is my HVAC company not showing on Google Maps”), comparison questions, and “best X in [city]” questions. Run each prompt monthly in ChatGPT, Perplexity, and Google AI Mode from a logged-out or fresh session. Log who gets mentioned, which URLs get cited, and whether the description of you is correct.

Put it on one monthly scorecard. One row per metric from the table above, with last month and this month side by side. Note what you published or fixed in between, so you can connect changes to causes.

How to calculate AI share of voice

AI share of voice compares your brand mentions with your competitors’ mentions across the same prompts. It’s a relative number, which makes it more useful than a raw mention count.

AI share of voice = your brand mentions ÷ total mentions of all tracked brands × 100

Here’s a worked example with made-up numbers. You track 25 prompts across 3 engines, which gives you 75 answers. Your brand appears in 18 of them. Your three main competitors appear 30, 24, and 12 times.

• Mention rate: 18 ÷ 75 = 24% of answers name you.
• Share of voice: 18 ÷ (18 + 30 + 24 + 12) = 18 ÷ 84 = 21.4%.

Calculate it per engine as well as in total. A brand can lead in Perplexity and barely show up in AI Mode, and a blended number hides that.

One caution: AI answers change between runs, even with the same prompt. Treat a 2- or 3-point monthly swing as noise. Look for trends across a quarter.

Free reports vs paid AI visibility tools

Paid trackers such as Semrush’s AI Visibility Toolkit and Ahrefs Brand Radar run thousands of prompts for you and chart share of voice, citations, and sentiment across several AI engines. The free reports give you platform-verified numbers but cover fewer engines.

	Free first-party reports	Paid AI visibility, trackers
Data source	Google and Microsoft’s own logs	Prompts the vendor runs and samples
Engines covered	Google AI features, Copilot/Bing, AI referrals in GA4	Usually ChatGPT, Perplexity, Gemini, AI Overviews, AI Mode; varies by tool
Share of voice vs competitors	Manual only	Built in
Effort	2 to 3 hours a month with a prompt log	Setup, then mostly automatic
Cost	$0	Monthly subscription; check current vendor pricing

Choose free reports if you’re a local or single-market business tracking under 50 prompts, or you’re still testing whether AI search sends you leads at all.

Choose a paid tracker if you compete nationally, report to leadership on share of voice against several competitors, or need hundreds of prompts tracked across engines. Even then, keep the free reports running. They’re your check on the vendor’s sampling.

Paid tools aren’t right for a business with no AI impressions in Search Console yet. Fix the content first; there’s little to measure.

What to check before you buy an AI visibility tool

• Engine coverage. Does it track the engines your buyers use? Coverage of Claude, Copilot, and Meta AI varies between tools.
• Prompt source. Are prompts based on real search demand, or invented by the vendor? Can you add your own?
• Location settings. If you’re a local business, can it run prompts as a user in your city?
• Raw answers. Can you read the full AI responses, or only see a score? You’ll want to check accuracy yourself.
• Export. Can you pull the data into the same sheet as your Search Console and GA4 numbers?

How Agreed Technologies measures AI visibility

At Agreed Technologies, our SEO work centers on answer engine optimization, so AI visibility appears in our client reporting alongside rankings, leads, and revenue. We already work daily in Google Search Console, GA4, Semrush, and Ahrefs, which means the free and paid layers above fit into one report instead of four.

We keep reports short and tied to leads. Clients get the handful of numbers that changed, what caused the change, and what we’re doing next, not a long monthly PDF.

The metrics we track can include AI mentions and citations, AI-generated impressions and visibility, AI-referred sessions, engagement, conversions, and qualified leads. We then compare those signals with traditional organic performance to understand whether increased visibility in AI search is actually contributing to business outcomes.

For context, a published AEO case study reported that ChatGPT, Gemini, and Claude generated 56 sessions and 6 inquiries for a client, while ChatGPT visitors converted at 12.8%, compared with 2.3% for organic search. The example illustrates why we don’t treat AI visibility as a vanity metric: the goal is to connect visibility in answer engines to measurable traffic, leads, and revenue.

As our own client datasets mature, we use the same framework to show the before-and-after impact of our work: what changed in AI visibility, which content or optimization caused the change, and whether that additional visibility translated into qualified leads and revenue.

When AI answers describe a client inaccurately, we trace it back to the pages and listings the AI relies on and fix those [NEEDS CLIENT INPUT: Please confirm whether this is part of Agreed Technologies’ standard process—specifically, whether the team identifies the pages, listings, or other sources contributing to inaccurate AI answers and corrects the underlying information. That work overlaps with building E-E-A-T trust signals and with the wider shift covered in why AI-driven SEO matters in 2026. Agencies can get the same work delivered under their own brand through our white-label SEO team.

Next step: call +1 (425) 216-6843, email contact@agreedtechnologies.com, or use our contact page. We’re headquartered in Newark, DE, with a team in Noida, India, and work with businesses across the US.

Want to know where you stand in AI search today? Book a strategy call, and we’ll walk you through your numbers.

LLMs decide what to answer in two stages. First, they choose whether to answer from training memory or search the web; if they search, they split your question into sub-queries and retrieve pages. Then the model favors passages that answer directly, cite evidence, and match what other trusted sources say, and those become its citations.

What is GEO?

Generative engine optimization (GEO) is the work of making your content more likely to be used and cited inside AI-generated answers from tools like ChatGPT, Perplexity, Gemini, and Google’s AI Overviews.

The term comes from a 2023 research paper by a team at Princeton and IIT Delhi, later published at the KDD 2024 conference: GEO: Generative Engine Optimization. Google’s view is simpler. Its guide to generative AI features states that, from Google’s perspective, optimizing for AI search is still SEO.

Both are right. The foundations are SEO. What changes is the last step: instead of ranking a page, an LLM picks a few passages to build an answer from. To influence that, you need to know how the picking works.

The two ways an LLM answers a question

Every AI answer comes from one of two sources, and each requires a different strategy.

	Answering from memory	Answering with retrieval (search)
Where facts come from	Patterns learned during training	Pages fetched live from a search index
Citations shown	Usually none	Yes, links to the pages used
How current	Only up to the model’s training cutoff	Current
What you can influence	How widely and consistently the web described you before the cutoff	Whether your pages are retrieved, and whether your passages get picked
Speed of change	Slow, months to a new model	Fast, days to weeks

Retrieval is where you have the most influence, so the rest of this guide focuses on it. Google calls this grounding, or retrieval-augmented generation (RAG): its AI features retrieve pages using its core Search ranking systems, then generate a response with links to those pages.

How LLMs decide what to answer: the 5 stages

The exact rules differ by product and aren’t public. But Google’s documentation and the GEO research describe the same basic pipeline.

1. Decide whether to search. Questions that need current, specific, or local facts usually trigger a search. Stable general knowledge may be answered from memory. Pew Research found Google showed an AI summary for 60% of searches starting with who, what, when, or why, and for 53% of searches of 10 words or more (Pew, July 2025).
2. Fan out into sub-queries. The model rewrites one question into several related searches. Google’s own example: a question about a weed-filled lawn might fan out into searches about herbicides, chemical-free removal, and prevention. You can be retrieved for a sub-query even if you don’t rank for the original question.
3. Retrieve candidate pages. Each sub-query pulls results from an index. Google’s AI features use the Google Search index, and a page must be indexed and eligible for a snippet to appear. ChatGPT search relies on OpenAI’s own crawler: per OpenAI’s crawler documentation, sites that block OAI-SearchBot won’t be shown in ChatGPT search answers.
4. Pick the passages. The model reads the retrieved pages and chooses the parts that best answer each sub-question. The GEO research shows that passages with sourced statistics, quotations, and citations are used more. It’s also likely, though not documented by the platforms, that models favor claims several sources agree on.
5. Write and attribute. The model writes the answer and links the sources it drew from. Being cited near the top of the answer, or throughout the text, is worth more than a passing mention.

Two things follow. You need to be retrievable (stages 1 to 3, mostly classic SEO and crawler access). Then you need to be the best passage (stage 4, mostly writing).

What the research says earns citations

The GEO paper tested nine content changes on 10,000 queries, using a test engine built on GPT-3.5 and the top five Google results, then repeated key tests on Perplexity. Here is what it found.

Change to the page	Result in the study
Add quotations from credible sources	Among the top three: 30–40% more visibility on its main metric. Best performer on Perplexity (+22%)
Add statistics instead of vague claims	Among the top three; up to 37% gain on Perplexity’s subjective measure
Cite credible sources in the text	Among the top three; strongest when combined with other changes
Improve fluency and readability	15–30% visibility gain
Make the tone more persuasive or authoritative	No significant improvement
Keyword stuffing	Little to no improvement; about 10% worse than doing nothing on Perplexity

The most useful finding for small businesses: lower-ranked pages gained the most. Adding citations raised visibility by 115% for pages ranked fifth in Google, while the top-ranked page’s visibility fell by about 30% on average. Content quality can beat domain strength at the passage-picking stage.

Treat these as direction, not formulas. The study ran in 2023 on older models, and today’s engines retrieve more sources. The authors also found results varied by topic, so test on your own pages.

Which sources AI answers cite today

Pew’s analysis of 68,879 Google searches from March 2025 shows who wins the citations now. Wikipedia, YouTube, and Reddit were the most-cited sources in AI summaries, together accounting for 15% of all sources. Government (.gov) sites made up 6% of AI summary sources, against 2% of regular results.

The pattern is clear: reference-style pages, video explanations, and real discussions get picked, and so do official sources. For a business, that means your own site isn’t the only lever. A clear YouTube explainer, accurate directory and review profiles, and genuine participation in forums your buyers read all feed the same pipeline.

How to get your content chosen

This is the playbook we’d follow for a service business, in order of impact.

1. Let the retrieval bots in. Allow Googlebot, Bingbot, and AI search crawlers such as OAI-SearchBot in robots.txt. OpenAI treats GPTBot (training) separately from OAI-SearchBot (search), so you can opt out of training without disappearing from ChatGPT answers.
2. Answer first, then explain. Put a one- or two-sentence answer directly under each question-style heading. That’s the passage a model can lift without rewriting.
3. Add evidence a model can quote. Replace “we’re fast” with a number you can stand behind and its source. Quote named experts, including your own team, with their role. Link the studies and rules you rely on.
4. Cover the fan-out questions. For each service, think through the follow-ups: cost, timeline, permits, what affects price, when not to hire you. Answer them on the main page instead of spinning up dozens of thin pages, which Google warns can breach its scaled content abuse policy.
5. Make your facts match everywhere. Services, service area, hours, and pricing language should be consistent across your site, Google Business Profile, directories, and review sites. Conflicting facts give a model no reason to trust you. This is the same work as building E-E-A-T trust signals.
6. Bring something only you know. Google’s guide says unique, non-commodity content will likely matter more than any other tactic. Local job data, real project photos, and first-hand comparisons are things a model can’t get anywhere else.

If you’ve worked through optimizing for Google AI Overviews and AI Mode, steps 1, 2, and 5 will look familiar. Steps 3, 4, and 6 are where the GEO research adds the most.

GEO myths you can skip

• Keyword stuffing for AI. It didn’t help in the GEO study and hurt on Perplexity.
• Sounding more authoritative. A confident tone alone made no significant difference. Evidence did.
• llms.txt for Google. Google says its Search and AI features ignore llms.txt and similar AI text files.
• Chopping pages into tiny chunks. Google says there’s no need; its systems find the relevant part of a longer page.
• Buying mentions. Google warns that chasing inauthentic mentions doesn’t help in its AI features.
• Guaranteed AI citations. No one can promise these. The pipeline is probabilistic, and answers change between runs.

How Agreed Technologies approaches GEO

At Agreed Technologies, our SEO work centers on answer engine optimization, so GEO is part of how we plan and write content, not a separate add-on. We use Semrush, Ahrefs, Google Search Console, and GA4 to see which pages get retrieved and cited, and we keep reporting tied to leads rather than vanity numbers.

In practice, that means rewriting key service pages answer-first, adding sourced numbers and named expert input, and cleaning up inconsistent business facts across the web. [NEEDS CLIENT INPUT: confirm these are standard deliverables, and add one client example with permission] For the wider shift behind this, see why AI-driven SEO matters in 2026 and our AI marketing guide for small businesses.

Next step: call +1 (425) 216-6843, email contact@agreedtechnologies.com, or use our contact page. We’re headquartered in Newark, DE, with a team in Noida, India, and work with businesses across the US.

Want to know whether LLMs are picking your pages or your competitors’? Book a strategy call,ll and we’ll show you where you stand.