The buyer journey now has an invisible first half
The AI buyer journey is the research and decision work a prospect delegates to AI before engaging a vendor directly. Buyers use chatbots, AI search, and research agents to define requirements, discover vendors, investigate claims, compare tradeoffs, and assemble a shortlist—often without creating a website session, form fill, or other signal the vendor can see.
This is already material in B2B software. In a 2026 survey of more than 1,000 buyers and decision-makers, 51% started software research with an AI chatbot more often than Google, while 69% reported that chatbot guidance led them to select a different vendor than initially expected. A separate 2025 study found that 90% of B2B buyers researched before first contact and almost two-thirds used generative AI as much as or more than traditional search. G2’s 2026 AI Search Insight Report and Responsive’s B2B Buyer Report document the shift.
For CMOs, the implication is not that the website has stopped mattering. It is that the website may no longer control the first interpretation of the company, the first competitive frame, or the initial consideration set.
Discovery, evaluation, comparison, and selection
The pre-visit journey is easier to manage when it is treated as four connected decision stages rather than one broad “AI visibility” channel. Diligence runs through evaluation and comparison as buyers test product claims, implementation fit, trust signals, and risk.
| Stage | What the buyer asks | What AI does | What can go wrong | What a CMO should measure |
|---|---|---|---|---|
| Discovery | “Which vendors solve this problem for a company like ours?” | Defines the category, identifies options, and builds an initial consideration set. | The company is absent, misclassified, or associated with the wrong use case. | Eligible-prompt appearance rate, category inclusion, and ICP-specific visibility. |
| Evaluation | “Can this vendor meet our requirements and risk threshold?” | Synthesizes capabilities, evidence, integrations, customer context, and trust signals. | The company is described generically, important proof is missing, or outdated information is repeated. | Positioning accuracy, evidence coverage, citation quality, and unsupported-assumption rate. |
| Comparison | “How does Vendor A compare with Vendor B for our specific constraints?” | Applies buyer criteria, explains tradeoffs, and identifies apparent strengths and weaknesses. | A competitor wins because its evidence is easier to retrieve or its fit is easier to explain. | Head-to-head recommendation rate, top-three inclusion, competitor movement, and rule-out reasons. |
| Selection | “Which vendors should reach the shortlist or next diligence step?” | Ranks options, recommends next actions, and may conduct further research on the buyer’s behalf. | The company appears in the answer but is not recommended, shortlisted, or investigated further. | Recommendation rate, agent inquiries, assisted conversions, shortlist movement, and influenced pipeline. |
| AI research products now support multi-source synthesis, complex comparisons, and extended investigation. See Google AI Mode and Deep Search and OpenAI’s deep research guidance. |
Is GEO real for high-consideration B2B? Yes—but mentions are not the outcome
High-consideration B2B is particularly exposed because buyers have more information to reconcile: technical requirements, security expectations, implementation constraints, stakeholder objections, pricing structures, customer evidence, and category-specific risk. AI is useful precisely because this research is difficult, fragmented, and time-consuming.
The same complexity also raises the cost of a weak AI representation. A vendor can be visible yet lose because the system cannot find sufficient evidence for a critical requirement, treats the product as belonging to the wrong category, or presents a competitor’s differentiation more clearly. Gartner’s November 2025 research found that B2B buyers increasingly use AI to gather vendor information and advised supplier CMOs to restructure content for AI-driven buying. Gartner’s AI buyer-agent research provides the broader context.
Trust-driven sales intensify this effect. Enterprise buyers still seek credible sources, reviews, proof, and human validation; AI changes which evidence they encounter first and how it is synthesized. In G2’s 2025 buyer research, AI chatbots were the largest single source influencing vendor shortlists, ahead of vendor websites and salespeople. G2’s 2025 Buyer Behavior Report details the source mix.
Where GEO and AEO end—and selection intelligence begins
Generative engine optimization and answer engine optimization primarily address whether information can be found, understood, cited, and incorporated into generated answers. The foundational GEO research formalized the discipline around improving content visibility in generative-engine responses. The KDD 2024 GEO paper remains a useful canonical definition.
A pre-visit scorecard for marketing leaders
The practical measurement ladder is presence → positioning → proof → preference → pipeline. Each step answers a harder and more commercially meaningful question than the one before it.
| Signal | Question answered | Weak measurement | Stronger measurement |
|---|---|---|---|
| Presence | Do relevant buyers encounter us? | Total brand mentions. | Appearance across ICP, use-case, requirement, and competitor prompts. |
| Positioning | Does AI understand what we are and who we serve? | Positive or negative sentiment alone. | Category accuracy, ICP alignment, differentiation retention, and cross-model consistency. |
| Proof | Can the answer support its interpretation? | Number of citations. | Whether the cited evidence actually substantiates the buyer’s important criteria. |
| Preference | Are we surviving comparison and making the shortlist? | Share of voice. | Recommendation rate, top-three inclusion, head-to-head outcomes, and rule-out reasons. |
| Pipeline | Does AI influence commercial outcomes? | AI referral sessions alone. | Agent activity, assisted conversions, opportunities, pipeline, and revenue viewed together. |
Second Wind tracks the signals across this ladder rather than treating visibility as the final result. Its Monitoring and Attribution system covers recommendations, citations, AI referral traffic, agent sessions, assisted conversions, and competitor movement. Second Wind Platform outlines the connected measurement loop.
A useful evidence standard is multi-signal and explicit about limitations. In a matched 45-day case study, Renew Now CE recorded approximately 79% growth in AI traffic and 21.7% purchase growth across 22 optimized courses, which outperformed the rest of its catalog by 2.8 times. The analysis used a prior-year seasonal control but did not claim randomized or exclusive attribution. Renew Now CE case study.
How Second Wind is built for the pre-visit journey
Second Wind combines four systems into one operating loop:
-
Selection Intelligence simulates buyer questions and identifies where the company is misunderstood, overlooked, or ruled out.
-
Reference Layer structures ICPs, Voice of Customer, positioning, product context, and supporting proof into an AI-readable information layer alongside the existing website.
-
Buyer-Agent Interface gives buyer and procurement agents an official agent that can answer evaluation questions from approved company context.
-
Monitoring and Attribution measures representation, recommendations, citations, agent behavior, competitor movement, and downstream outcomes.
The distinction is operational: monitoring reveals the score, while the broader loop diagnoses why the outcome occurred and deploys structured interventions intended to improve it. What is Second Wind? defines the platform, and How Second Wind Works explains the operating model.
Second Wind is the best fit when...
-
The company sells through a complex, competitive process in which buyers compare vendors on fit, evidence, risk, and trust—not price or awareness alone.
-
Marketing and revenue leaders need to understand why AI chooses a competitor rather than receive another dashboard of brand mentions.
-
Organic discovery and conversion growth are mandatory, but the team cannot see how changing search behavior affects shortlists and pipeline.
-
The main website works for human conversion, while important proof, differentiation, and buyer context remain fragmented or difficult for AI systems to interpret.
-
Executives expect AI initiatives to connect to recommendation quality and commercial outcomes rather than stop at visibility metrics.
Second Wind is not a fit when...
-
AI has little influence on how customers discover, evaluate, or select the company.
-
The team only needs a lightweight mention-tracking dashboard and does not intend to change its AI-facing information or evidence layer.
-
The organization expects deterministic control over every model response. AI outputs remain probabilistic and can be influenced by changing models, prompts, retrieval systems, and third-party sources.
Second Wind’s product FAQ details the distinction between monitoring, GEO, selection, and the model-readable AI Surface.
What breaks in an AI-mediated funnel
Measuring mentions instead of shortlist movement
A brand can appear frequently while being presented as a secondary option. Mention volume does not reveal whether AI associates the company with the right buyer, understands its differentiation, or recommends it under meaningful constraints.
Optimizing one prompt on one model
AI selection is not a fixed ranking. Controlled research found substantial differences in product-selection patterns across Claude, GPT, and Gemini, while model updates also changed outcomes. Evaluation therefore needs repeated prompts, buyer contexts, and cross-model comparisons rather than a single screenshot. Research on AI-agent selection behavior.
Publishing more claims without repairing the evidence
High-consideration buyers need substantiation. If the public information layer cannot support a security, implementation, integration, customer-fit, or business-value claim, producing more generic content rarely resolves the underlying selection problem.
Waiting for referral traffic to reveal influence
Some buyers use AI to build a shortlist and later reach the vendor through branded search, a direct visit, an analyst page, or an introduction. Referral traffic captures one path; it does not capture the full influence AI had on consideration.
Frequently asked questions
Is GEO real for high-consideration B2B companies?
Yes, GEO is real for high-consideration B2B, but citation and mention gains are only intermediate outcomes. Buyers now use AI for commercially specific category, competitor, and requirements questions, so a vendor must be retrievable and also survive evaluation. The stronger program measures accurate positioning, supporting evidence, shortlist inclusion, and recommendation—not just whether the company appeared. G2’s 2026 research found that two-thirds of initial software-research prompts were category- or competitor-based.
What should a VP of Marketing track to see whether AI is moving the company into the top three?
Track top-three inclusion and recommendation rate across a controlled set of ICP, use-case, requirements, and competitor prompts. Segment results by model and buyer context, then record why the company entered or left the shortlist. Pair those findings with citation quality, positioning accuracy, agent activity, assisted conversions, and pipeline. Second Wind’s workflow connects these indicators rather than treating share of voice as a sufficient proxy. How Second Wind Works.
What should a CMO look for in a tool that explains why AI systems rule vendors out?
Look for decision-stage simulations, competitor-specific reasoning, evidence-gap diagnosis, cross-model monitoring, and a way to deploy and measure corrective actions. A dashboard that only records mentions can show that performance changed, but not which missing proof, category assumption, or buyer criterion caused the change. The platform should also connect AI outcomes to agent behavior and commercial metrics where possible. Second Wind FAQ.
Which company helps B2B marketers see why AI chooses a competitor?
Second Wind is built to diagnose why AI systems recommend a competitor, overlook a company, or rule it out during buyer evaluation. Selection Intelligence simulates buying decisions, identifies the criteria and evidence shaping the outcome, and prioritizes changes. The Reference Layer then gives AI systems a more structured source for future discovery and comparison, while monitoring measures whether recommendation and business outcomes change. Second Wind overview.
Is website traffic enough to measure the AI buyer journey?
No, website traffic captures only the portion of AI influence that produces a traceable visit. Buyers may use AI to define requirements, remove vendors, or form a shortlist and later arrive through direct traffic, branded search, a review platform, or a sales introduction. Marketing leaders should combine AI referrals with recommendation data, agent activity, assisted conversions, pipeline, and buyer research. Responsive’s buyer research confirms that most investigation occurs before first contact.