Agentic Brand Control · Win Rate, measured and moved

AI now makes the recommendation.
— we measure whether it picks you.

A mention is not a sale. Deep in a buying conversation the model stops listing and starts choosing — and one brand gets recommended while the rest get mentioned, or dropped entirely.

Meridian measures your win rate at that moment across ChatGPT, Gemini, Claude, Grok and Perplexity, then remediates the content, product data and technical layers that decided it. Across 1,427 measured buying conversations, only 12.7% of the brands named at the opening were still the recommendation at the decision.

Mentions are a vanity metric. Win rate is the revenue.

Request a Demo Fully managed. End to end. No analyst translation, no agency hand-off, no "here's what's wrong, good luck."

The discipline is Agentic Brand Control — the category we named and published in May 2026. The metric is AI Win-Rate: how often, in a buying conversation that reaches a decision, the model recommends you outright rather than listing you alongside everyone else. Meridian is a fully managed service, not a dashboard — we measure the win rate, remediate what suppresses it, and re-measure on a continuous cycle.

The shift

Most companies still treat AI as a search problem.

It is no longer a search problem.

It is now a recommendation infrastructure and AI ad-intelligence problem.

The brands AI recommends at the purchase decision point aren't the most-mentioned. They're the ones whose product data is internally consistent across every retailer page AI reads from, whose authoritative content sits in the public source diet AI models train on, and whose technical infrastructure is built specifically for AI crawlers rather than Google bots.

As AI platforms move toward monetized recommendation environments — sponsored placements inside ChatGPT, ad layers in Perplexity, agentic commerce systems making autonomous purchases — the brands that win this shift are the ones treating AI as infrastructure, not as marketing.

This is a different operating discipline from SEO. It is what Meridian was built for.

SHIFT 01

From mention & citation rate to win rate

Mentions and citations at the opening measure whether the model has heard of you. Almost every established brand passes that test — and it predicts nothing. The decision sits four to twelve turns deep, and that is the only turn where a brand is actually chosen.

SHIFT 02

From content gaps to infrastructure gaps

Your website is no longer the primary surface. AI reads from a public source diet — Wikipedia, retailer pages, trade authority, community discussion, peer-reviewed repositories. Most enterprise infrastructure isn't built for this.

SHIFT 03

From visibility tools to operational systems

Diagnostic dashboards tell you what's broken. They don't fix it. Meridian operationalizes the fix — content, structural, and technical remediation, executed as a managed enterprise system.

The intermediary layer

There is now a set of third parties standing between you and your buyers. Nobody hired them.

Your buyers no longer arrive at you first. They ask an assistant, and that assistant compares you against your competitors, forms a view, and makes a recommendation — before anyone on your sales or marketing team knows the conversation happened. These are independent intermediaries with a decisive say in your revenue, and no organization has any idea what they are telling people about the business.

GEO gives you mentions. A mention is not intelligence. Meridian tells you what these intermediaries actually say to your buyers at the decision, who they hand the sale to instead, and fixes it where it is wrong.

Your potential customers
The buyer with the budget
Asks in plain language.
Never types your name.
The intermediaries
Unhired. Unmanaged. Unmeasured.
ChatGPT Gemini Claude Grok Perplexity
compares · judges · recommends
The decision is made here.
Your sales & marketing
Finds out from the pipeline
Sees the deals that arrive.
Never sees the ones lost here.
No visibility of any of it.
What a mention tool sees

Whether your name appeared near the top of the answer. That is all it knows. It cannot tell you what was said about you, what the model compared you against, or who it recommended instead.

What Meridian sees

The whole conversation through to the decision — what the intermediary says about you, which competitor it hands the sale to, and why. Then we go and fix the sources it read.

What this is not

We are not a GEO tool. Try asking your GEO vendor what your win rate is.

Generative-engine optimization tools track mentions and citations at the opening of a conversation. That work is real, we do not argue with it, and if you are running a program you should keep it. But it answers one question — does the model know we exist? — and for an established brand the answer is almost always yes.

Being named is not the problem — it is the part almost everyone already passes. Getting named and then losing anyway is the norm. In our public AI Recommendation Ranking Tables, 95 of 180 brands never won a single one of their 27 buying conversations. Those are two different failures with two different fixes, and only one of them shows up on a mention chart.

So ask a second question the mention tools were never built to answer: when the conversation reaches a decision, how often does the model actually recommend us — and against whom do we lose? That number is your win rate. It is the number Meridian exists to move.

Our current product portfolio

How we can help you. Our current product portfolio.

01 · BRAND
Brand and the Role of the Brand
Win rate · role of brand in the decision

Measures how often AI recommends your brand at the point of decision — across the full multi-turn buying journey, on every major model — then remediates the content, product data and technical layers that decided the outcome. Includes role-of-brand measurement: what the model actually weighs when it chooses, and what happens to your position once price and features are held equal. The core Meridian platform, delivered fully managed.

02 · REPUTATION
Reputation
How the models characterize you

Measures how AI describes your reputation across trust, leadership, products, citizenship and workplace — scored cell by cell, tracked over time, and remediated where the models are wrong, thin or out of date. Available direct or through research partners, in multiple markets.

03 · POLITICAL
Political & Advocacy
Candidates · elected officials · PACs · policy

How AI represents a candidate, elected official, PAC or policy position — scored on visibility, accuracy, framing and consensus. Built for campaigns, PACs, advocacy organizations and public affairs teams, and run non-partisan. Measurement only: we never remediate a political entity. That is a standing principle, not a missing feature — intervening would forfeit the neutrality that makes the measure worth having.

04 · NPD
New Product Development
White-space & expansion intelligence

Multi-persona, multi-lens AI research that finds where category demand is unmet — the white space to expand into and the products to build, grounded in what the models and real buyers actually ask. Delivered as a scoped research engagement.

Delivered in your vertical

Every product is run to the evidence and governance standards your sector demands.

Healthcare & pharma Financial services & payments Technology & B2B software Beauty & CPG Retail & commerce Travel & hospitality Automotive Political & public affairs
The research foundation

Built on seventeen months of continuous primary AI research.

Since March 2025 we've mapped how ChatGPT, Gemini, Perplexity, Claude and Grok reason about brands across every major consumer and B2B category — anti-aging skincare, pharmaceuticals, financial services, CPG, luxury, automotive, travel, SaaS, and emerging agentic commerce systems.

The patterns are consistent, measurable, and specific to each platform: Gemini's educational drift arc, Perplexity's runtime retrieval displacement, ChatGPT's training-data anchoring, Grok's recency bias, Claude's authority anchoring. Measurement runs on ChatGPT — in two conditions, closed-book and live-search, so any difference is attributable to retrieval rather than the model — alongside Gemini, Claude, Grok and Perplexity, weighted to real assistant usage rather than counted equally. Remediation is built for all of them, because the content and product data we fix is what every one of them reads.

22,000+
AI buying journeys analyzed
380+
Enterprise brands tested
17
Months of continuous measurement
5
AI platforms measured and remediated
14
Sectors measured longitudinally
The number that matters most

Research scale earns trust. But the number that ends the meeting is the one almost nobody has been shown: how often a model actually picks you when it has to choose.
12.7%of brands named at the opening of a buying conversation are still the recommendation at the decision — n=1,427 probes across ten industries

Why this is defensible

The methodology is public. The system that runs it is ours.

Measurement only counts if someone can check it. Our instruments are published, versioned and citable — and the software that operates them is wholly owned and built in-house, not licensed, resold or wrapped around someone else's API.

01

Published methodology and prior art

Every instrument is written up, versioned and deposited with a permanent DOI — including the full AIVO measurement methodology and the scoring behind our public AI Recommendation Ranking Tables. A client, a competitor or a journalist can reconstruct how a number was produced. Most of this category will not show you that.

02

At the standards table

AIVO is working with the Media Rating Council AI Standards Working Group, alongside the IAB Tech Lab and other members, on how AI brand measurement should be audited and accredited. We are contributing to the yardstick, not waiting to be measured by it.

03

Wholly owned technology

Meridian and everything around it is our own build: the probe and diagnostic engine, the remediation and publication system, the reputation platform, role-of-brand measurement, PIM diagnostics, the market knowledge architecture, and the political representation system. Developed in-house, live in production, owned outright.

The platform

Six instruments. One platform.

Meridian operates as a single continuous system. Every instrument feeds the next. Diagnosis flows into interpretation, interpretation flows into remediation, remediation flows into publication, and publication flows back into the next measurement cycle.

INSTRUMENT 01

Buying Journey Probe

Turn-by-turn journey mapping across 4–12 turn flows. Captures DIT detection, handoff analysis, and named competitor displacement in directed and agentic modes.

INSTRUMENT 02

PSOS Baseline

Platform Stability of Organic Scoring — breadth, depth, resilience, sentiment, decay. Fragile, moderate, or strong banding across every major AI platform.

INSTRUMENT 03

Reasoning Paths

How AI reasons its way to a recommendation. The logic chain from query to answer — which sources it draws on at each turn, how the reasoning shifts when buyers probe deeper, and where your authority sits in that path.

INSTRUMENT 04

Ad Intelligence

Detects sponsored placements and AI-native ad behavior across ChatGPT, Perplexity, and Gemini. Per-platform pre-spend verdict before any budget commits.

INSTRUMENT 05

PIM Diagnostic

Cross-retailer entity resolution and product data fragmentation analysis. Identifies the field-level inconsistencies driving AI displacement at SKU and category level.

INSTRUMENT 06

Agentic Readiness

Measures your brand's survival through autonomous AI shopping flows — agent handoff capture, purchase recommendation defense, agentic ad receptivity.

The full remediation suite

We don't just tell you what's broken. We fix it.

Most AI visibility platforms hand you a dashboard and walk away. Meridian operates as a managed enterprise system, executing remediation across three pillars: the content layer AI cites from, the structural layer AI reads your product data from, and the technical layer AI crawlers ingest. All published into the public source diet, all measured for revenue attribution, all running on a continuous monthly or quarterly cycle.

PILLAR 01

Content remediation

Editorial, community, retail, comparison

Our proprietary Radar and Orbit systems pinpoint the exact content AI reads from at the decision. Then we make sure it’s there — identified, sourced, and placed.

  • Editorial authority content placed in trade publications
  • Wikipedia and Wikidata entity accuracy and references addressed
  • Reddit, Medium, and community-source presence built
  • Retailer review pages strengthened
  • Comparison content ("brand X vs brand Y") authored
  • Parent-brand anchor content for SKU-displacement recovery
  • Agentic handoff recovery content for turn-8 survival
PILLAR 02

Structural remediation

Product data, retailer alignment, PIM

We resolve the product data fragmentation that AI sees across your retailer footprint, and push the corrections into your PIM platform.

  • Canonical product record generated per SKU base
  • Per-retailer field-level correction calculator
  • Direct integration into Salsify, Akeneo, Stibo, inRiver
  • Internal master spec consolidation
  • Shade, size, and variant naming standardization
  • Cross-retailer field alignment program
  • SKU-displacement pattern resolution
PILLAR 03

Technical remediation

Infrastructure, crawlers, publication

We build the technical infrastructure AI crawlers actually need — and publish your authority into the public research corpus AI models train on.

  • Structured atoms generated and deployed per platform
  • Machine-readable site indexes built and maintained
  • AI crawler access configured at the infrastructure layer
  • Schema markup tuned per AI platform
  • DOI-backed publication into Zenodo, GitHub, HuggingFace
  • Provenance tier metadata embedded at every atom
  • Wikidata entity registration and maintenance
Fully managed. End to end. As a system. No analyst translation. No agency hand-off. No "here's what's wrong, good luck." Continuous monthly or quarterly cycle inside one platform, with full revenue attribution.
Built for enterprise scale

Multi-brand portfolios. Defensible governance.

Meridian is designed for organizations operating multiple brands across multiple categories, multiple geographies, and multiple regulatory environments. The platform scales from a single-brand pilot to full group programs spanning forty or more brands across all major divisions.

It is built to ISO 42001 and NIST AI RMF 1.0 standards, with DOI-assigned methodology and complete SHA-256 audit trails — the only AI visibility platform built to enterprise governance requirements out of the box.

E01

Multi-brand portfolio

Every brand in one workspace. Cross-brand drift detection. Portfolio-level governance.

E02

ISO 42001 aligned

Built to international AI management system standards. Defensible to procurement, compliance, and board.

E03

DOI-backed methodology

Peer-reviewed measurement framework published on Zenodo. Every claim traces to source probe turns.

E04

Continuous monthly or quarterly cycle

Diagnostic, remediation, publication, attribution, re-measurement. As a managed operational system.

Typical deployments

Designed for organizations where AI recommendation behavior materially impacts revenue.

Meridian is deployed across sectors where AI-driven recommendation is becoming the dominant discovery and decision channel — and where governance, brand positioning, and competitive survivability cannot be left to chance.

D01
Global beauty & skincare
Multi-brand portfolios at parent-group level. Retailer fragmentation and SKU displacement are critical risks.
D02
Enterprise retail & commerce
Large catalogs across multiple retailers. PIM consistency is the dominant AI visibility lever.
D03
Pharmaceutical organizations
Regulated authority claims, clinical evidence anchoring, governance defensibility at scale.
D04
Financial institutions
Composite+ measurement, regulatory alignment, board-level reporting on AI brand position.
D05
Multi-market consumer brands
Cross-geography AI behavior, multi-language source diet, regional retailer ecosystems.
D06
Travel & hospitality
Long consideration journeys with heavy review and citation dependence. Agentic booking is the fastest-emerging displacement risk in any category.
The next shift

The category is about to change again.

Agentic commerce arrives at scale through 2026. AI-native advertising is already launching on ChatGPT. Gemini Shopping is in market. Perplexity is running monetized recommendation. When an AI agent buys on behalf of your customer, the decision is made by the model — not the person — and the rules of that decision are exactly what Meridian has spent seventeen months decoding.

Brands that measure now will be positioned when the shift happens. Brands that wait will find out, too late, that they were invisible at the moment that mattered.

The AIVO ecosystem

Meridian is the enterprise platform from the team behind the AIVO Standard — the research program establishing the measurement standards for AI-era brand discovery.

Fourteen months of primary research ran before a line of the platform was written — across financial services, pharmaceuticals, consumer goods and beauty — and produced the methodologies now operationalized inside Meridian: Buying Journey Probe, PSOS, Reasoning Paths, Ad Intelligence, PIM Diagnostic, and Agentic Readiness. Measurement has run continuously in the seventeen months since. Published in the AIVO Journal, deposited with DOIs, and referenced in Fortune, AdAge, Business Insider, and American Banker.

Request a Demo

See how AI recommendation behavior is shaping your category right now.

Thirty minutes. Mapped against AI recommendation behavior, AI ad influence, and competitive survivability in your sector — using the same continuous measurement infrastructure Meridian operates for enterprise clients across beauty, pharma, finance, retail, travel, and consumer brands.

Request a demo or diagnostic

We respond within one business day. No sales sequence.