Kyle Poyar's recent Growth Unhinged piece opens with a question that should be on every marketing team's agenda: when AI agents start making purchasing decisions on behalf of users, will your brand be in the consideration set? The question is not theoretical. Ramp has launched Agent Cards. Mastercard and Google have partnered on a standard to verify AI-initiated transactions. Stripe has been running its Agentic Commerce Protocol with OpenAI for six months. The infrastructure for AI buying already exists.
Poyar focuses on what this means for pricing strategy — and his analysis is sharp. But pricing legibility is the second problem. The first problem is brand legibility. And it is the one that most teams are not yet addressing.
The shortlist problem
An AI agent evaluating vendors for a procurement decision does not start at your pricing page. It starts with a broader question: which brands in this category are worth evaluating at all? That initial filtering — the shortlist — is determined by citability, not by pricing transparency.
If an agent asks "which AI visibility platforms should I evaluate?" and your brand is not in the AI's training data, not cited by authoritative sources, not described consistently across the surfaces the model is pulling from — you are not on the shortlist. The pricing evaluation never happens. The structured documentation Poyar recommends never gets read. You have been excluded before the conversation started.
This is the citability problem. And it precedes every other optimisation a brand might make for agentic commerce.
What agents actually evaluate
Poyar identifies something important about how AI agents differ from human buyers: they are not susceptible to the psychological mechanisms that traditional pricing strategy relies on. Charm pricing ($9.99) does not work on an agent. Artificial scarcity does not create urgency. A "contact sales" CTA does not create intrigue — it creates exclusion. The agent moves on.
Agents evaluate based on structured, available, machine-readable information. They synthesise across sources — your own documentation, third-party reviews, Reddit discussions, analyst reports, competitor comparisons — and reach a conclusion based on what the aggregate of available information says about your brand. As Poyar notes, citing AirOps research, third-party sources account for 85% of brand mentions in AI search. Your own website is a minority voice in the agent's evaluation.
This is where citability becomes the foundational issue. The Citability Score framework measures exactly what an agent is evaluating when it forms a picture of your brand: the consistency of your entity definition across sources, the quality and structure of your content, the coherence of your positioning across platforms, and the degree to which external sources describe you accurately. A brand with a Citability Score of 35 — which is the average before optimisation — is being described inconsistently, incompletely, and often inaccurately across the surfaces an agent will consult.
The Sentiment gap in agentic evaluation
There is a specific dimension of the citability problem that becomes acute in an agentic buying context: the gap between how a brand positions itself and how AI systems actually describe it. This is what the Sentiment sub-score within the AI Visibility Index measures.
For human buyers, a Sentiment gap is a brand problem — it creates confusion, dilutes positioning, and reduces recall. For AI agent buyers, a Sentiment gap is a qualification problem. If the agent's synthesis of available information produces a description of your brand that does not match the buyer's criteria, you will be filtered out — even if your actual product is a perfect fit.
Consider a brand that positions itself as a premium, enterprise-grade solution, but whose external footprint is dominated by early-stage coverage, freemium comparisons, and user-generated content that emphasises accessibility and ease of use. An agent evaluating enterprise solutions may not include it in the shortlist, not because the product does not qualify, but because the available information does not signal enterprise readiness clearly enough. The brand's own positioning is being overridden by the aggregate of what third parties say.
Correcting this is not a matter of updating the pricing page. It requires a systematic audit of how the brand is represented across all surfaces an agent will consult — and a structured programme to bring that representation into alignment with actual positioning.
What Poyar gets right — and what comes before it
The practical recommendations in the Growth Unhinged piece are sound. Transparent pricing documentation, structured and machine-readable. FAQ content designed for how agents retrieve information in chunks. Dedicated landing pages for each tier. External pricing consensus across review sites, Reddit, and analyst sources. These are the right moves for the pricing layer of agentic readiness.
But they assume the brand has already solved the layer beneath: being in the agent's awareness in the first place, being described accurately enough to qualify for evaluation, and being positioned consistently enough to match the agent's buying criteria.
The sequence matters. An agent that has never encountered your brand — or that has formed an inaccurate picture of it from inconsistent sources — will not navigate to your pricing documentation. The AEO work Poyar recommends for pricing only generates value if the brand-level citability work has been done first.
Pricing legibility is the second problem. The first is whether the agent knows your brand exists — and whether it describes you accurately enough to put you in the room.
The measurement layer
Poyar recommends tracking AI influence at different stages of the buying journey — from discovery through evaluation to purchase. This is the right framework. The metrics he suggests for the discovery stage — visibility in relevant AI queries, attribution tracking — map directly onto what the AI Visibility Index measures systematically and continuously.
The AVI tracks citation frequency, position, and sentiment across all major AI platforms for the queries that matter to a brand's category. This is not a one-time audit — it is the baseline measurement infrastructure that allows a brand to monitor its citability in real time, observe how representation changes as content and external sources evolve, and detect early signals of the kind of Sentiment drift that would cause an agent to misclassify the brand.
For brands preparing for agentic commerce, this measurement layer is not optional. Without it, the optimisation work Poyar recommends — and the citability work that precedes it — has no feedback loop. You are making changes without knowing whether the agent's picture of your brand is actually improving.
The window before it becomes standard
Agentic purchasing is early. Poyar is correct that we are still in the infrastructure phase — the protocols exist, the payment rails are being built, but autonomous B2B purchasing at scale is not yet the norm. For most categories, AI agents currently function as influencers on the buying committee, shaping which products get evaluated rather than completing transactions autonomously.
This is precisely why now is the right time to act. The brands that establish strong citability before agentic buying reaches mainstream adoption will have a compounding advantage — they will already be in the agent's mental model, already represented accurately, already consistent across sources. The brands that wait will face a harder problem: correcting a representation that has already been ingested and reinforced across multiple training cycles, against competitors who have been building citation authority for months or years.
The question Poyar asks — "when AI starts buying, will you be in the consideration set?" — has an answer that depends on work that needs to start well before the first agentic transaction happens. That work is citability. And the measurement infrastructure to track it already exists.
Source: Growth Unhinged — Your next customer might be an AI agent, by Kyle Poyar (April 2026).