INSIGHT

The AI Research Gap: How to Measure AI-Influenced B2B Buyer Journeys

    Add a header to begin generating the table of contents
    Scroll to Top

    Executive takeaway: B2B buyers are increasingly researching companies inside AI systems before analytics ever sees them. If we credit only the visible website visit, we risk confusing the touchpoint we can measure with the influence that created demand. I do not believe the answer is another platform promising perfect attribution. The better approach is to connect the evidence we have and be clear about what it proves, what it suggests, and what remains unknown.

    Virbion AI Discovery-to-Revenue Evidence Chain connecting Semrush visibility, GA4 referral traffic, a visitor intelligence platform, and a CRM and revenue system.
    The Virbion framework separates directional AI visibility, confirmed referrals, associated visitor journeys, and known business outcomes.

    How AI Research Creates a B2B Attribution Gap

    A buyer asks an AI assistant to compare providers, explain a technical problem, identify credible vendors, or summarize an unfamiliar category. That answer can shape the shortlist before the buyer visits a single company website.

    The buyer may then search for a company by name, type the URL directly, return days later on another device, or ask a colleague to investigate. When the website visit finally occurs, analytics records the session it can see. Organic search, direct traffic, referral traffic, or another source receives the credit. The AI interaction that influenced the visit may leave no trace.

    The analytics is not necessarily wrong. It is incomplete.

    I refer to the missing influence as the AI research gap: the difference between what shaped the buyer’s decision and what our measurement systems can observe. Semrush describes a similar attribution gap in agentic search, including the common scenario in which an AI recommendation leads to a later branded search without a direct analytics trail.[3]

    AI did not create the entire problem. B2B attribution was already complicated by long buying cycles, buying committees, cross-device activity, privacy choices, offline conversations, partner influence, dark social, and inconsistent CRM capture. AI has added another important research environment outside much of the measurable journey.

    Why AI Attribution Is a Common Marketing Gap

    This is not a niche reporting problem. It is a more visible version of a gap marketers have been managing for years.

    CMO Alliance research shows how widespread the fragmentation remains. Data integration is a barrier to proving marketing ROI for 61% of marketers, while 53.7% cite accurate data collection. At the same time, 36.1% still rely primarily on first-touch attribution and 31.2% on last-touch attribution, even though very few B2B journeys are that simple.[1]

    The problem is not always a lack of technology. Many companies already own the essential platforms. What they lack is the connection between them. CMO Alliance reports that 54% of CMOs believe their stacks have meaningful gaps and specifically points to missing connections among attribution, content performance, and pipeline data.[2]

    That matches what I see in practice. Most marketing teams can measure pieces of the journey. The harder problem is connecting those pieces well enough to explain how visibility became engagement, how engagement became an identifiable relationship, and whether that relationship contributed to pipeline or revenue.

    How to Build an AI Discovery-to-Revenue Evidence Chain

    The fastest way to overstate marketing attribution is to ask one platform to answer a question it was not designed to answer.

    A visibility score is not revenue attribution. An observable referral is not always the buyer’s first exposure. A CRM source field is not a complete record of the research journey.

    Each platform answers a different question. I find it more useful to separate the evidence into four layers and connect them only where the data supports the connection.

    Virbion framework: The AI Discovery-to-Revenue Evidence Chain
    Evidence layerExample platformWhat it can demonstrateWhat it cannot establish alone
    Market visibilitySemrushMentions, citations, cited pages, prompts, competitors, trends, and perception signals.Whether a specific buyer saw or acted on an AI answer.
    Website acquisitionGA4Identifiable AI referrals, landing pages, behavior, engagement, and key events.AI influence that produced no attributable click or arrived later through another source.
    Known visitor journeySalesIQConsented visit history, referrer, pages viewed, and behavior associated with an identified lead or contact.Visitors who decline tracking, clear cookies, change devices, or never identify themselves.
    Pipeline and revenueZoho CRMQualification, ownership, meetings, opportunities, progression, closed revenue, and customer history.Undocumented influences that occurred before the person became known.

    How Semrush Measures AI Visibility Before the Click

    AI visibility platforms help answer a question that web analytics cannot: Is the company present when AI systems answer the questions its buyers are asking?

    Semrush’s AI Visibility Toolkit can benchmark brand presence, track mentions and citations, identify cited pages, analyze prompts and topic opportunities, compare competitors, evaluate perception and sentiment, and monitor visibility over time. Its Visibility Overview also exposes which sources and owned pages appear in monitored responses.[4][5]

    These signals matter because a buyer can encounter a brand without visiting its website. A cited article, a favorable mention, or inclusion in a comparison can influence the shortlist even when no click follows.

    But I would not present a Semrush visibility score, mention, or citation as proof of customer acquisition. The platform cannot identify everyone who saw an answer, confirm that they acted on it, or show that the exposure created pipeline. The composite score is a benchmark. The underlying prompts, mentions, citations, cited pages, competitors, platforms, and trends are where the more useful evidence lives.

    My measurement rule: Use Semrush to demonstrate market presence and directional influence. Do not turn visibility metrics into revenue claims they cannot support.

    How GA4 Identifies Measurable AI Referral Traffic

    When an AI platform passes a recognizable referrer and the visitor clicks through, GA4 can identify the source or referring domain. Marketers can then examine the landing page, engaged sessions, page flow, key events, and conversions associated with those observable visits.

    Google distinguishes between user acquisition, which describes how a user was first acquired, and traffic acquisition, which identifies the source of an individual session. That difference matters when an AI referral starts the measurable relationship but later visits arrive through direct, organic, email, or another channel.[6]

    This makes GA4 useful for answering practical questions: Which AI platforms send measurable traffic? Which pages attract it? What do those visitors do next? Do they engage or convert differently from organic and direct visitors?

    The limitation is just as important. GA4 captures only the AI referrals that produce an observable trail. If the platform withholds the referrer, the buyer copies a URL, or the buyer returns later through another source, the AI influence may appear as direct, organic, or unassigned traffic.

    How the Visitor Intelligence Platform Connects Consented Visitor Journeys to the CRM

    Virbion AI Discovery-to-Revenue Evidence Chain connecting Semrush visibility, GA4 referral traffic, a visitor intelligence platform, and a CRM and revenue system. Other platforms may provide comparable capabilities, but the consent, identity and continuity requirements remain the same.

    Aggregate traffic data becomes more useful when we can connect a visitor’s behavior to a known business relationship. This is also where we need to be precise about the roles of GA4 and Zoho SalesIQ.

    SalesIQ does not normally take GA4’s source value and simply transfer it into CRM. It records its own visitor data, which can include the referrer, campaign source, landing page, pages viewed, time spent, and other visit activity. Through the Zoho CRM integration, information such as First Visited URL, Referrer, and visit history can be associated with leads and contacts.[7][9]

    When the visitor permits the applicable tracking cookies and later identifies themselves through a form, chat, login, or another supported method, SalesIQ can associate that tracked browser journey with the matching CRM record. This gives us a potential bridge between an identifiable AI referral and a known prospect or customer.

    That connection is real, but it is conditional. The referral has to survive. Cookie continuity has to be maintained. The visitor has to identify themselves, and the CRM match has to work. SalesIQ provides APIs for managing analytics and performance cookie consent, so the design also has to respect the visitor’s choices.[8]

    The distinction: GA4 reports the acquisition and behavior it can observe. SalesIQ can create the visitor-level bridge to CRM when consent, identity, and technical continuity are available.

    How the CRM Connects Known Journeys to Revenue

    The CRM is where the evidence becomes commercially meaningful. It allows us to connect known journeys to qualification, sales activity, meetings, opportunities, revenue, and customer status. Without that connection, AI reporting can become another collection of visibility metrics that leadership cannot evaluate economically.

    I also do not recommend relying on one CRM source field. At minimum, the revenue system should distinguish the original source, observable conversion source, self-reported discovery source, influential content, campaign history, and the sales team’s validated understanding of why the buyer engaged.

    The objective is not to award all the credit to one platform. It is to understand enough of the sequence to make better decisions: which topics deserve investment, which pages influence evaluation, which sources produce qualified demand, and which follow-up process moves that demand forward.

    A Practical B2B Reporting Example

    We built a custom reporting environment for a B2B client using Semrush, GA4, Zoho Analytics, SalesIQ, and Zoho CRM. It identifies measurable visits from AI sources and reports the referring platform, landing content, visitor behavior, page flow, engagement, and relevant conversion activity.

    The reporting confirms that AI platforms can send identifiable, commercially relevant visitors. It also makes the boundary of that evidence visible. GA4 can confirm an AI-referred session when the referral survives. It cannot identify every buyer who encountered the company in an AI answer and later returned through branded search, direct navigation, email, or another device.

    With the appropriate SalesIQ consent and successful visitor identification, we can extend the journey. SalesIQ can preserve its referrer and behavioral history and associate that activity with the corresponding CRM lead or contact. CRM can then show whether the known record became qualified, entered an opportunity, or produced revenue.

    That does not eliminate the attribution gap. It narrows the gap and shows us exactly where the remaining uncertainty begins. I would rather give an executive that honest boundary than manufacture certainty the data cannot support.

    How to Classify AI Attribution Evidence

    Not every signal deserves the same confidence. A credible executive report should show the strength of the evidence instead of flattening everything into one number.

    Evidence classifications and the claims each level can support
    Evidence classificationMeaningAppropriate claim
    Confirmed referralA recognizable AI source sent an observable website session.The AI platform referred this session.
    Associated journeyA consented tracked visit was connected to an identified CRM record.This known prospect had an observable AI-referred journey.
    Self-reported influenceThe buyer names an AI platform, source, or resource during a form submission or sales conversation.The buyer identified this influence, even if analytics did not capture it.
    Directional visibilityMentions, citations, prompts, cited pages, branded search, or direct traffic move in a meaningful pattern.The evidence suggests increased AI-assisted visibility; it does not prove individual causation.
    Unknown influenceThe research occurred without a preserved referral, identity, or self-report.The influence may exist, but the current system cannot verify it.

    How to Capture AI Influence with Self-Reported Attribution

    As more research happens inside systems we cannot fully observe, we need to ask buyers better questions.

    The standard ‘How did you hear about us?’ field is usually too shallow. A buyer may first encounter the company in an AI answer, validate it through Google, read a LinkedIn post, and then submit a form. One forced-choice field will not explain that journey.

    I prefer two concise questions: ‘Where did you first become aware of us?’ and ‘What most influenced your decision to contact us?’ Sales can validate the answers during discovery and record useful details such as the AI platform, prompt topic, article, referral, event, or colleague involved.

    Self-reported attribution is not perfect. Buyers forget details and simplify their answers. But it provides an independent signal that cookies and referral data often cannot—especially when we compare it with GA4, SalesIQ, content engagement, and CRM progression.

    What Should Leadership See in an AI-Era Marketing Report?

    An executive does not need another dashboard filled with platform activity. The report should be organized around the business questions leadership is actually trying to answer.

    Discoverability: Are we appearing for priority buyer questions, and how do mentions, citations, cited pages, and competitive share change?

    Observable demand: Which AI sources send measurable traffic, which pages attract it, and how does that traffic behave?

    Known journeys: Which consented visitors can be connected to CRM records, and what did they view before identifying themselves?

    Pipeline contribution: Which known journeys became qualified leads, meetings, opportunities, or revenue?

    Evidence quality: Which conclusions are confirmed, self-reported, directional, or still unknown?

    This structure gives leadership a more honest view of marketing contribution. It also creates a practical feedback loop: visibility data shapes content priorities, visitor behavior improves conversion paths, CRM outcomes refine targeting, and sales feedback clarifies which messages and sources influence real buyers.

    How to Measure AI Influence Without Claiming Perfect Attribution

    Marketing attribution is often discussed as if one sufficiently advanced model will eventually explain the entire customer journey. I do not think that is a realistic standard.

    AI research, privacy choices, cross-device behavior, offline influence, peer recommendations, and delayed conversion all create legitimate blind spots. Better technology can reduce those blind spots. It cannot erase them.

    The standard I use is decision-grade evidence: enough reliable, clearly classified information to decide what to continue, change, test, or stop. Semrush, GA4, SalesIQ, CRM, self-reported attribution, and sales feedback each contribute a different part of that evidence.

    The principle: Do not ask one metric to explain the buyer journey. Build the evidence chain, preserve the limitations of each signal, and connect the strongest available evidence to business outcomes.

    The Executive Takeaway

    AI is changing how B2B buyers discover, compare, and validate companies. Some of that activity will create a measurable referral. Much of it will not.

    That is not a reason to abandon attribution. It is a reason to stop overstating what any one platform can prove.

    Marketing leaders should connect market visibility, observable referrals, consented visitor journeys, self-reported influence, and CRM outcomes into one evidence system. The result will still contain uncertainty. But it will be more useful than a first-touch field, more credible than a composite visibility score, and more actionable than a dashboard that stops at website traffic.

    Perfect attribution is not the goal. Better decisions are.

    Is your AI visibility connected to revenue evidence?

    At Virbion, we help growth-stage and mid-market companies connect AI visibility, website analytics, CRM, revenue operations, sales follow-up, and decision-grade reporting. We determine what can be measured, connect the evidence that matters, and improve the revenue system around it—without manufacturing certainty the data cannot support.

    Request a Revenue Systems Review

    Sources and Further Reading

    1. How CMOs Can Justify Their Financial Investments. CMO Alliance. Current research on attribution models, data integration, collection accuracy, and ROI evidence.
    2. What CMOs Must Know About Martech in 2026. CMO Alliance. Research on stack gaps, fragmented data, integration barriers, and the missing connection between traffic and pipeline.
    3. Attribution Gap in Agentic Search: How to Close It. Semrush. Discussion of invisible AI influence and multi-layer measurement.
    4. AI Visibility Toolkit. Semrush Knowledge Base. Current toolkit capabilities for brand visibility, competitors, prompts, perception, and monitoring.
    5. AI Visibility Overview Report. Semrush Knowledge Base. Definitions of visibility score, mentions, citations, cited pages, sources, and topic opportunities.
    6. User Acquisition Report. Google Analytics Help. Definitions of first-user acquisition, session acquisition, source, medium, engagement, and key events.
    7. Visitor Tracking with Zoho SalesIQ. Zoho CRM Help. How SalesIQ visit information and referrer fields can be associated with CRM leads and contacts.
    8. SalesIQ Cookie Consent API. Zoho SalesIQ Help. Controls for SalesIQ analytics and performance cookie preferences.
    9. SalesIQ Visitor Object. Zoho SalesIQ Help. Visitor attributes including referrer, campaign source, landing page, previous page, and CRM status.

    About the Author

    Jason Steinbach

    I focus on building and running the revenue infrastructure behind growth: go-to-market architecture, pipeline and forecast governance, CRM and analytics alignment, and decision-grade reporting. My free time is spent with the incredible wife, family and community God has blessed me with, mentoring youth, producing my youngest son's band "Anthem Eterna", and trying to keep up with my oldest son's gaming development projects.
    Scroll to Top