AI Traffic vs. AI Citations: The New SEO Reality Is a Two-Layer Journey (and Most Teams Are Measuring Only One Layer)

AI Search is reshaping how customers discover brands—but the biggest reporting mistake I’m seeing is painfully simple: teams are measuring only the click.

In 2026, that’s like measuring your brand only by the number of people who walked into your store, while ignoring everyone who looked at the shelf through the window, read the reviews, and decided you were “the best option” before they ever showed up.

AI systems increasingly do the “shopping around” inside the answer. That means the URLs that get cited (used as evidence, references, and supporting surfaces) often differ from the URLs that get visited (AI referral landings). If you optimize only for AI Referral traffic, you’re optimizing for the final step of a journey you didn’t measure.

This editorial is a practical playbook for businesses and agencies: what changed, why it matters, what can go wrong, and how to build an AI search measurement + execution system that actually drives outcomes.

Concise summary

• AI traffic and AI citations are different layers. AI traffic shows where a click lands; citations show where AI systems found and referenced information—even if nobody clicked.
• Clicks often land “later” in the journey. AI compresses discovery/evaluation inside the answer and pushes the click toward brand entry points and transactional pages.
• Most businesses are undercounting AI influence because they track referrals but don’t audit cited URLs, page types, and “non-standard” surfaces (subdomains, tools, locators, language variants).
• Reporting must segment by page type and intent. A homepage click, a rates page citation, and an OAuth redirect are not the same KPI.
• Execution is now the bottleneck. Winning requires continuously preparing pages AI systems rely on—and shipping approved site changes fast.

Table of contents

• What changed in search (and why this moment feels confusing)
• The two-layer AI search journey: “Visibility” vs. “Visits”
• Keep your feet on the ground: organic search is still the scale channel
• Why AI clicks often land “later” (and why that can fool you)
• Citations are a signal—not proof (and how to use them responsibly)
• Page-type reality: what AI tends to cite vs. what it tends to send traffic to
• The “weird URLs” problem: non-standard surfaces you can’t ignore
• What this means by vertical (and what SMEs can steal from each)
• A concrete SME scenario: local clinic vs. ecommerce store vs. SaaS
• How to report AI search without lying to yourself
• What can go wrong: the new failure modes in AI search
• A practical 30/60/90-day action plan
• Where AYSA.ai fits: monitor, prepare, request approval, execute
• What to do next
• Sources and further reading

What changed in search (and why this moment feels confusing)

The story most people repeat about AI search is simple: “AI traffic is growing, Organic search is shrinking, everything is changing.” It’s dramatic—and often directionally true in specific niches—but it’s not a useful operating model for most businesses.

Here’s what actually changed:

• The answer became a destination. In AI-assisted search experiences, users can get comparisons, shortlists, explanations, and recommendations without opening ten tabs.
• Links became optional. AI experiences may show fewer links than classic search results. And when links appear, they can be more conservative (homepages, canonical pages, known brand surfaces).
• “Visibility” separated from “visits.” Your page can be used to form an answer (and influence a decision) without receiving a click.
• Journey stages compressed. The traditional funnel (awareness → consideration → conversion) still exists, but AI can collapse the first two stages into the interface itself.

So the old mental model—“rankings lead to clicks; clicks lead to conversions”—needs an update. Not because SEO is dead, but because the measurable events have multiplied.

The two-layer AI search journey: “Visibility” vs. “Visits”

Aleyda Solis published one of the most practical datasets I’ve seen on this topic: a comparison of AI traffic versus AI citations using Semrush data across multiple verticals (Travel & Tourism, Finance, Real Estate, and Shopping/Retail). You should read it in full because it gives you a vocabulary to stop arguing in circles.

Here’s the key mental model from her research, translated into business terms:

• AI traffic = where the click lands (a measurable visit). It’s the “endpoint event.”
• AI citations = which URLs AI systems reference in answers. It’s an “influence/visibility event.”

These are not the same layer of the journey, and they don’t behave the same. The research shows that the pages getting clicks skew toward brand-entry and task/action pages, while citations skew toward discovery and evaluation surfaces (guides, categories, policies, support, locators, and other informational infrastructure).

That distinction is the difference between:

• Optimizing only the pages that benefit from AI clicks today, versus
• Preparing the pages AI systems use to trust you, understand you, and recommend you tomorrow.

Source: Aleyda Solis — AI Traffic vs AI Citations: What Clicks and Cited Pages Show About the AI Search Journey.

Keep your feet on the ground: organic search is still the scale channel

One reason AI search reporting goes off the rails is that teams emotionally overweight what’s new. AI referral traffic is exciting because it’s novel and because it arrives with a sense of “the future.” But for most sites, AI referrals are still small compared to organic search.

Solis’ analysis (Semrush, April 2026, USA, top sites across four verticals) highlights that organic search remains vastly larger than AI traffic in aggregate. The exact percentages depend on the dataset and methodology, but the direction is what matters for operators: AI traffic is not replacing organic at scale—yet.

As a business owner, that should change your posture:

• Don’t panic and abandon traditional SEO that still pays the bills.
• Don’t chase AI traffic as your primary KPI if it’s 0.x% of your visits.
• Do build AI search readiness now—because citations and brand selection effects can move faster than traffic.

The point isn’t to dismiss AI search; it’s to stop using a single metric to make a strategic bet.

Why AI clicks often land “later” (and why that can fool you)

When an AI system answers a question like “best mortgage for first-time buyers,” “best family hotels near Disneyland,” or “best running shoes for flat feet,” it can do a lot of the work users previously did across multiple websites:

• Define the decision criteria
• Compare options
• Explain trade-offs
• Surface pros/cons
• Highlight availability, policies, or requirements

If that happens inside the AI interface, the eventual click is more likely to go to an action page: a booking page, a product page, an application start, a Store locator, a checkout flow, or sometimes just a homepage (the “safe canonical link”).

This creates two common illusions:

Illusion #1: “AI isn’t driving discovery, because we don’t see AI traffic to our guides.”

Reality: AI may be using your guides heavily (citations), but sending the click elsewhere (home/product/action). If you only report AI referrals, you’ll undercount AI’s influence at the top and middle of the funnel.

Illusion #2: “Our homepage is winning in AI, because it gets the AI clicks.”

Reality: homepages can become “AI traffic sinks” because they’re safe links, not because they’re the pages that built trust. If you optimize only the homepage, you might neglect the pages that earn the recommendation.

Also: it’s not only about journey stage

Solis correctly notes another operator reality: some AI referrals may be generated by assistants/agents following workflow links (logins, redirects, dashboards). That traffic is real, but it’s not a customer discovering your brand; it’s a system navigating a task. Treating that as “marketing success” is how bad decisions get funded.

Citations are a signal—not proof (and how to use them responsibly)

There’s a dangerous overreaction happening right now: “If my URL is cited, that means the AI answer is sourced from me.” Not necessarily.

Citations are still incredibly valuable, but you need to interpret them correctly:

• Citations are a measurable visibility signal. They show that your URL was surfaced/referenced alongside an answer for a set of prompts (in the dataset’s terms).
• Citations are not perfect Attribution. They don’t guarantee the AI used your page as the original factual source for every claim in the answer.
• Citations can be biased toward “safe” pages. Homepages, brand hubs, and canonical pages might be cited for navigational simplicity.

So what should a business do with citations?

• Use them to identify which page templates AI systems rely on.
• Use them to prioritize content hygiene, accuracy, and freshness.
• Use them to map “AI discovery surfaces” that need investment—even if traffic hasn’t moved yet.

In other words: citations are not the finish line; they’re a diagnostic tool.

Page-type reality: what AI tends to cite vs. what it tends to send traffic to

Let’s make this concrete for non-SEOs. Most sites have a limited set of page types (templates). AI systems tend to treat them differently depending on intent.

Based on the patterns summarized in Solis’ research, you can expect this split:

AI citations skew toward “answer infrastructure”

• Category/search/listing pages (help AI understand options and availability)
• Guides/editorials (help AI explain and compare)
• Support/policy pages (help AI handle edge cases and trust issues)
• Locators (help AI connect a query to a location/service footprint)
• Product/property details (help AI validate specifics)

AI traffic skews toward “action endpoints”

• Homepages / brand entry pages
• Transactional pages (book, buy, apply)
• Account/cart/checkout flows
• Operational URLs (logins, redirects) in some datasets

Why this matters for budgets

If your team only funds “conversion pages” because that’s where AI traffic lands, you may starve the pages that make you recommendable. Then you’ll wonder why your competitors keep showing up in AI answers.

Modern SEO/AEO/GEO planning needs two tracks:

• Citation readiness: pages AI uses to decide what to say
• Click readiness: pages users land on when they decide to act

The “weird URLs” problem: non-standard surfaces you can’t ignore

One of the most important operator insights from Solis’ analysis is that a huge share of cited URLs can live in a messy bucket of “other deep / non-standard URL surfaces.”

If you’ve run an SEO program, you already know this pain:

• Subdomains for locations, services, or specialized tools
• Language/region variations
• Legacy folder structures that don’t map cleanly to “blog” or “category”
• Offer pages, policy pages, destination pages, and tool pages with bespoke paths

In classic SEO, these pages often exist outside the “content team” workflow. In AI search, these are exactly the surfaces that can get cited because they answer narrow, practical questions.

Business implication: if your monitoring doesn’t include those surfaces, you’ll miss where AI visibility is actually happening.

Execution implication: if those surfaces are owned by different teams (ops, legal, product, engineering), you need a system that can coordinate changes across owners—with approvals—without stalling for months.

What this means by vertical (and what SMEs can steal from each)

Solis’ dataset covers four industries. You may not be in those exact verticals, but the patterns translate. Here’s how I’d apply the learnings as an operator.

Travel & Tourism: citations drive planning; clicks drive booking

Travel is a textbook example of AI compressing research. A traveler asks:

• “Best all-inclusive resorts in Cabo for families”
• “Best time to visit Tokyo with kids”
• “Hotels near the convention center with free parking”

AI can assemble a plan quickly. Your guide pages, destination pages, policy pages (cancellation, fees), and property details can get cited—even if the click ends up on a homepage or booking funnel.

What SMEs should steal:

• Create “decision pages” that match planning intent (neighborhood guides, seasonal travel, policy clarity).
• Maintain structured internal linking from those pages into bookable inventory.
• Keep policies and availability explanations updated—stale policy pages are AI trust killers.

Finance: traffic can be distorted by workflows; citations reveal true discovery

Finance is where bad measurement can do the most damage. Why? Because many finance platforms have heavy authenticated experiences: dashboards, payment flows, OAuth redirects, session-based URLs. Those can show up as “AI traffic” in third-party datasets, but they’re not necessarily new customer acquisition.

Solis points out Stripe as an example where AI traffic heavily hits workflow URLs, while citations skew toward educational resources. That illustrates the core lesson: traffic composition matters as much as volume.

What SMEs should steal (even outside finance):

• Separate “marketing AI traffic” from “operational AI traffic.”
• Build an educational layer designed for evaluation (fees, requirements, comparisons, definitions).
• Don’t let secure/operational URLs pollute reporting—filter, segment, and annotate.

Real Estate: homepages get clicks; local/tools get cited

Real estate search is inherently local and tool-driven: neighborhood pages, city pages, valuation tools, mortgage calculators, school district info. AI citations will naturally spread across these pages because they answer practical questions.

Meanwhile, clicks can consolidate on major brand entry points. That can mislead teams into over-optimizing the homepage while the real “AI trust surfaces” are local pages and tools.

What SMEs should steal:

• Invest in location coverage and consistent templates.
• Ensure tool pages (estimators, calculators) have clear explanations and constraints.
• Keep internal navigation clean so AI and users can traverse from discovery to action.

Shopping/Retail: citations spread; the click owner may not be you

Retail has a special problem: even when AI cites your product information, the click can go to a marketplace, an aggregator, or a different retailer with stronger fulfillment signals.

This creates a reality businesses don’t like to admit: you can “win” citations and still lose the transaction. So you need to optimize not only for mentionability, but for conversion ownership—where the purchase actually happens.

What SMEs should steal:

• Strengthen product detail clarity (specs, compatibility, sizing, policies).
• Create comparison and FAQ layers that AI can cite.
• Reduce friction on the pages that receive the click: shipping, returns, trust, and speed.

A concrete SME scenario: what this looks like in real life

Let’s run a realistic scenario with three businesses that might read AYSA.ai:

Scenario A: a local clinic (high trust, high stakes)

A clinic offers dermatology and cosmetic treatments. Patients ask AI:

• “What’s the difference between microneedling and laser resurfacing?”
• “How much does Botox cost in [city]?”
• “How to choose a dermatologist near me?”

What gets cited: treatment explainers, pricing ranges with caveats, aftercare pages, safety/contraindications, location/provider pages.

What gets the click: homepage, appointment booking, location page, provider profile.

Common mistake: measuring only AI referrals and concluding “AI isn’t working,” then cutting content investment—while competitors become the cited source.

What to do: build a “citation-ready” library with medically reviewed content, clear authorship, updated policies, and internal paths to booking.

Scenario B: an ecommerce brand (mid consideration, high comparison)

Customers ask AI:

• “Best standing desk under $500”
• “Standing desk pros/cons”
• “Desk size for dual monitors”

What gets cited: comparison guides, sizing guides, warranty/returns, shipping pages, category pages.

What gets the click: a specific product page, a bundle page, or the homepage.

Common mistake: optimizing only product pages for “AI traffic,” while leaving policies and guides thin, outdated, or contradictory.

What to do: treat guides and policies as revenue infrastructure—because in AI search, they are.

Scenario C: a SaaS tool (evaluation-heavy, trust signals matter)

Prospects ask AI:

• “Best invoicing software for freelancers”
• “Stripe vs PayPal fees”
• “How to set up recurring billing”

What gets cited: documentation, pricing explanations, integration pages, glossary posts, security pages.

What gets the click: pricing page, signup, homepage.

Common mistake: shipping product changes but not updating docs, security pages, or pricing clarifications—creating AI-visible inconsistency.

What to do: create a monitoring loop that flags cited pages for freshness and consistency reviews.

How to report AI search without lying to yourself

If you’re a founder or a marketing leader, you don’t need more dashboards—you need fewer metrics with clearer definitions.

Here’s the reporting system I recommend building (and what AYSA customers should push their teams toward).

1) Separate the layers: AI presence, AI citations, AI referral traffic

• AI referral traffic: what visits are attributed to AI referrers in your analytics.
• AI citations: which URLs are being cited in AI answers (from the best dataset you have access to).
• AI presence: broader concept—mentions without explicit citations, brand inclusion, and recommendation patterns (harder to measure, but still important).

Don’t mash these into one chart. It will produce bad strategy.

2) Segment by page type before making strategic claims

At minimum, report AI traffic and AI citations by page template:

• Homepage / brand entry
• Category / search / listing
• Product / service detail
• Guides / editorial
• Support / policy / FAQ
• Location / locator
• Account / checkout / booking
• Auth / redirect / operational

This prevents the classic “our AI traffic went up because logins went up” mistake.

3) Track traffic ownership (who actually gets the click)

In many markets, AI answers may recommend “the best option” but the click goes to a marketplace, an aggregator, or a directory. If you’re a brand that sells through channels, you need to monitor where the click is going—because your “AI visibility” can translate into someone else’s revenue.

Even without perfect data, you can start with a practical proxy:

• Which of your pages are cited most?
• Which competitor/marketplace pages get cited alongside yours?
• When you do receive AI traffic, which landing pages get it?

4) Build KPIs that match reality

Examples of smarter KPIs:

• Citation-ready coverage: % of priority topics with a high-quality, maintained page.
• Template health: freshness and consistency scores across “AI cited” templates (policies, guides, docs).
• AI landing readiness: speed, clarity, and conversion friction on top AI landing pages.
• Overlap analysis: which pages are both cited and clicked (high leverage) versus only cited (influence) versus only clicked (possible “safe link” effect).

What can go wrong: the new failure modes in AI search

AI search introduces new ways to lose that classic SEO didn’t punish as quickly.

1) Stale pages become trust landmines

Support pages, policy pages, and definitions are heavily used in AI answers. If those pages are outdated, AI can surface incorrect guidance (or avoid you entirely). Businesses often under-resource these pages because they don’t “drive conversions.” That logic is now obsolete.

2) Inconsistent information across templates

If your pricing page says one thing, your FAQ says another, and your help doc says a third, you’re feeding AI a contradiction. That can reduce citation likelihood and increase the risk of misrepresentation.

3) Operational URLs pollute strategy

Logins, redirects, session URLs, and checkout tokens can appear in AI traffic datasets. If you don’t segment them out, you’ll optimize the wrong thing and misread growth.

4) Homepage obsession

Yes, homepages can receive a lot of AI clicks. But that does not mean the homepage is doing the persuasive work. Over-focusing on the homepage often leads to under-investment in the pages that actually earn citations.

5) Execution lag becomes a competitive moat (for your competitor)

AI search optimization is not just content creation. It’s continuous maintenance across many page types. Teams that can ship changes weekly will outpace teams that ship quarterly.

A practical 30/60/90-day action plan

Here’s how I’d operationalize this as a founder, a head of marketing, or an agency owner.

Days 1–30: Build the measurement spine

• Inventory your templates (homepage, category, product/service, guide, support/policy, location, tools, checkout, docs).
• Separate AI referrals from other traffic in your analytics reporting.
• Create an “operational URL” filter list (login, oauth, /authorize, /session, /redirect patterns) so these don’t distort interpretation.
• Start a cited-URL watchlist using the best available external dataset you have (Solis’ work highlights why this layer matters).

If you need a starting point for AI visibility and monitoring concepts, we keep this framework updated here: https://aysa.ai/ai-search-visibility/ and https://aysa.ai/monitoring/.

Days 31–60: Fix the pages AI systems use to decide

• Pick 10–20 priority topics where you must be recommendable (your money pages + your trust pages).
• Upgrade “citation readiness” for each topic: definitions, constraints, policies, comparisons, and clear internal links to next actions.
• Eliminate contradictions across pricing, policies, FAQs, and support docs.
• Improve “non-standard surfaces”: subdomains, tools, locators, language variants that are likely to be cited.

AYSA’s tooling for AI SEO workflows can help teams structure this work without drowning in manual audits: https://aysa.ai/ai-seo-tools/.

Days 61–90: Optimize click endpoints without starving discovery

• Improve top AI landing pages (often homepage/product/service/pricing): clarity, speed, trust blocks, friction removal.
• Strengthen internal “handoff” paths from cited pages (guides/policies/categories) to action pages.
• Ship changes faster by moving from “recommendations” to “approved execution” workflows.

This is where most teams stall—because execution requires coordination across stakeholders. That’s why AYSA is built around a controlled change process: monitor, prepare, ask for approval, then execute accepted website changes.

Where AYSA.ai fits: monitor, prepare, request approval, execute

AI search optimization is quickly becoming an execution game, not a theory game.

Most businesses already have enough SEO advice to last a lifetime. The bottleneck is:

• knowing what to change first,
• making changes safely (especially on revenue pages), and
• shipping consistently.

AYSA is designed to operate like an execution system—not just an audit generator:

• Monitor your AI/SEO visibility and key site surfaces (monitoring).
• Prepare specific changes by page type (metadata, internal links, content structure, technical fixes, template improvements).
• Request approval so you keep human control over brand/legal/compliance-sensitive edits.
• Execute the accepted changes so strategy becomes reality.

For teams evaluating whether execution support is worth it, start here: https://aysa.ai/pricing/. For ongoing editorial context and updates, we publish more on our blog.

What to do next

• Stop using AI referrals as the sole KPI. Separate citations and clicks.
• Audit your top cited page types (guides, policies, categories, locators, tools)—even if they don’t get AI traffic.
• Segment “operational” URLs so you don’t mistake workflow traffic for marketing impact.
• Build two backlogs: citation readiness + click readiness.
• Increase shipping velocity using an approval-based execution workflow (so changes don’t die in tickets).
• Set a 90-day review cadence: what got cited, what got clicked, what converted, what changed after you shipped.

Sources and further reading

• Aleyda Solis — AI Traffic vs AI Citations: What Clicks and Cited Pages Show About the AI Search Journey
• Semrush (tool referenced in the research dataset)
• Aleyda Solis — The 10 Key Characteristics of AI Search Winning Brands
• SEO FOMO newsletter hub (AI/SEO news context)
• AYSA — AI Search Visibility
• AYSA — AI SEO Tools
• AYSA — Monitoring
• AYSA — Blog

Related AI SEO resources

Continue the AI search topic inside AYSA.

Use these pages to connect the article with AI SEO tools, AI visibility monitoring, AI Overviews and approved website execution.

AI SEO tools The commercial hub for AI SEO, AI search optimization and approved execution. AI search monitoring Track AI visibility, answer-engine mentions and the next actions worth approving. AI Overviews SEO Prepare clearer, more structured pages without promising guaranteed inclusion. SEO, AEO and GEO visibility A plain-English explanation of the search visibility layer AYSA prepares for.

Written by

Marius Dosinescu

Marius Dosinescu is the founder of AYSA.ai, an entrepreneur focused on SEO automation, ecommerce growth, authority building and approved website execution for businesses that want organic growth without specialist overhead.

Author profile Dosinescu.ro LinkedIn

Share in X f @