The CMO's AEO Dashboard: 7 Metrics That Actually Belong in a Board Deck

When Gartner published its 2025 CMO Spend Survey in October 2025, it found that 68% of CMOs had reduced their SEO budget over the prior 12 months while increasing investment in "AI-driven discovery." The problem: almost none of them had a measurement system that could tell a board whether the investment was working. The metrics on the slide were still organic sessions, keyword rankings, and share of voice — metrics built for a search paradigm that AI assistants are actively eroding.

This is the core problem facing every CMO going into a board presentation in 2026. AI search is a board-level topic — every board member has used ChatGPT, most are asking about AI strategy, and many are now asking directly whether the company is "winning" in AI search. But the CMO dashboard that answers that question does not exist in most companies. The legacy measurement stack produces vanity metrics. The AEO measurement stack — the one that maps to AI-influenced pipeline — is being built from scratch.

This piece is the practical build guide. It covers the seven AEO metrics that belong in board decks, how to measure each one, what the benchmarks look like across the B2B SaaS companies that are furthest along, and how to build the dashboard architecture that surfaces them clearly. It also covers the political infrastructure: how to get a board to care about metrics they have never seen before, and how to connect AI search visibility to the revenue narratives that boards actually respond to.

Why Organic Traffic and Share of Voice No Longer Belong in Board Decks

Before covering the seven metrics that should replace them, it is worth being precise about why the legacy metrics have broken down — not because they have become less accurate, but because they are measuring a surface that is no longer the primary driver of buyer discovery in most B2B categories.

Organic search traffic to B2B websites declined an average of 34% between January 2025 and March 2026, according to Signal's cross-industry analysis. The decline is not uniform — it is steepest in informational and comparison queries, which are exactly the queries that sit at the top of the funnel. For a CMO reporting organic sessions as a health metric, the dashboard is showing a market that is contracting. What it is not showing is the parallel market that is growing: the AI-assisted discovery channel where buyers are asking ChatGPT or Perplexity who to evaluate before they ever visit a website.

Share of voice — the traditional measure of brand visibility in paid and earned media — has the same structural problem. It measures presence in channels that buyers are increasingly not using for initial vendor discovery. A brand with 40% share of voice in Google paid search may have 8% share of model in AI assistant category responses. Those two numbers describe completely different competitive positions. The board is seeing the paid-media number. The one that predicts pipeline is the AI number.

The third legacy metric worth retiring from board decks is keyword ranking. As AI Mode in Google and AI Overviews now surface synthesized answers for the majority of commercial queries, ranking on page one for a head term no longer guarantees meaningful traffic. The correlation between page-one rankings and traffic that closed-loop analytics can actually measure has dropped substantially in 2025 and 2026. Reporting keyword rankings to a board in 2026 is approximately like reporting television ratings in 2015 — technically accurate, structurally irrelevant to where the audience actually is.

The Seven Metrics and What Each Measures

These seven metrics form a coherent measurement framework — they cover discovery (Metrics 1-2), positioning quality (Metrics 3-4), pipeline impact (Metric 5), information accuracy (Metric 6), and competitive standing (Metric 7). A CMO who can report all seven has built the most complete picture of AI search health that 2026 tooling makes possible.

Metric 1: Share of Model (Category Citation Rate)

Share of model is the primary AEO metric — the percentage of AI assistant responses to relevant category queries that include a citation or positive mention of your brand. It is the AI-era equivalent of aided brand awareness, except it is measured against the specific moment of purchase intent rather than general population recall.

How to measure it: Build a prompt set of 50 to 200 queries representing how your buyers ask AI assistants about your category. Include head-term queries ("best CRM for mid-market SaaS"), comparison queries ("alternatives to Salesforce for a 200-person company"), and job-to-be-done queries ("how should a VP of Sales manage a distributed SDR team"). Run these prompts across ChatGPT (web and API), Perplexity, Claude, and Gemini. Record which brands appear in each response. Calculate share of model as the percentage of responses in which your brand is mentioned or cited.

Benchmarks:

What to report to the board: Share of model on your top 10 category head terms, trended quarterly, compared to your top two competitors. A 5-point quarter-over-quarter improvement is a strong result. Sustained decline signals pipeline risk that will show up in new ARR in two to three quarters.

Tooling: Profound, Otterly, and Peec all measure share of model at scale. Profound is the most comprehensive for multi-engine tracking. Budget $1,500 to $3,000 per month for enterprise-grade measurement. Manual prompt testing is viable at smaller scale but does not produce the statistical confidence needed for board-level reporting.

Metric 2: Citation Accuracy Rate

Citation accuracy rate measures what percentage of AI-generated claims about your product are factually correct. It is the quality complement to share of model — being cited frequently with wrong information is worse than being cited rarely with accurate information, because inaccurate AI citations generate a specific type of pipeline damage: prospects arrive at a sales call believing your product does something it does not, get corrected, and experience trust erosion that correlates with lower close rates.

How to measure it: Build a product-fact battery of 30 to 60 questions covering your most important features, pricing tiers, integration capabilities, use case fit, and limitations. Ask each question across your primary AI assistants. Compare the AI's response to ground truth from your product documentation. Score each claim as accurate, inaccurate, or hedged (when the AI appropriately expresses uncertainty rather than stating something wrong). Citation accuracy rate is the percentage of definitive claims (accurate + inaccurate) that are accurate.

What a 65% accuracy rate actually looks like in practice: In one mid-market HR software company's audit, AI assistants were accurately describing the product's core features 65% of the time. The 35% inaccurate claims clustered around three areas: a pricing tier that had changed six months earlier, an integration that had been deprecated, and a feature limitation the product had actually overcome in a Q3 release. The inaccuracies existed because the company's product pages had not been updated, its changelog was private, and its documentation still mentioned the deprecated integration. All three were fixable in 30 days. After fixes, accuracy climbed to 82%.

What to report to the board: Citation accuracy rate on a quarterly basis, with the specific claim categories that are failing. Frame it as brand risk: inaccurate AI descriptions of your product are the equivalent of incorrect listings on G2 or Capterra, except they reach buyers at the moment of highest intent with no correction mechanism.

Metric 3: Branded Versus Unbranded Citation Ratio

This metric measures whether AI search is pulling buyers toward your brand specifically (branded citations: "you should look at [Company X]") versus toward your category generally ("you should evaluate tools in this category"). The ratio is a proxy for brand equity in the AI-search layer — high branded citation rates indicate that AI assistants associate specific positive attributes with your brand name, not just with your category.

How to measure it: In your share-of-model measurement runs, tag each mention as branded (the AI uses your company name, product name, or a clearly attributable descriptor) or unbranded (the AI describes a capability or use case that fits you without naming you). The branded-to-unbranded ratio is simply the proportion of branded citations within your total citation mentions.

Why it matters: Unbranded citations are category-awareness citations — they imply that AI assistants know your product exists and belongs in a category but have not built strong entity associations that drive name-specific recommendations. Branded citations are the ones that actually drive direct navigation and brand-search lift. A company with 85% branded citation share is in a fundamentally stronger position than one with 40% branded share, even if total citation volume is similar.

The benchmark threshold: B2B SaaS companies that are running AEO programs effectively see branded citation shares of 60% or above within 12 months of program start. Below 40% suggests that content and documentation infrastructure has not created the entity associations that AI models need to recommend by name.

Metric 4: Comparison-Page Citation Rank

Comparison queries — "X vs Y," "alternatives to X," "best X for Y" — are the highest-intent queries in B2B AI search. They represent buyers who have already decided to change their tool and are now evaluating options. Where your brand appears in AI responses to comparison queries is one of the strongest leading indicators of pipeline quality.

How to measure it: Identify the 15 to 25 most common comparison queries in your category. Run them across AI assistants and record: (a) whether your brand is mentioned, (b) in what position (first, second, third, or lower), and (c) whether you are recommended as the preferred option or mentioned as a secondary alternative. Comparison-page citation rank is a composite of these factors — ideally expressed as the percentage of comparison queries in which your brand appears in position 1 or 2.

The dynamics that drive this metric: As covered in depth in the AEO citation tracking playbook, comparison-page citations are heavily influenced by the quality of vendor-published comparison content. Companies that publish detailed, fair, substantive "X vs Y" and "alternatives to X" pages on their own domain see those pages cited by AI assistants in the answers to competitor queries — meaning their brand appears in conversations they previously could not reach. This is one of the highest-leverage AEO investments available.

What to report to the board: Your comparison-page citation rank on the top 10 competitive queries, compared to competitors, trended quarterly. A CMO who can show that the company's brand is now mentioned in 60% of "alternatives to [Competitor]" queries — up from 20% a year ago — has told a clear story about competitive positioning shift.

Metric 5: AI Dark Funnel Pipeline Estimate

The AI dark funnel is pipeline influenced by AI search that arrives in your CRM as direct navigation, branded search, or unattributed inbound — with no AI referral source recorded in GA4 or your attribution tooling. It is the fastest-growing unattributed revenue segment in B2B in 2026, and it is systematically invisible in standard analytics setups.

How to measure it: Three methods, combined:

Method 1 — Closed-won survey. Ask a sample of recently closed-won customers: "How did you first become aware of us?" Include "AI assistant (ChatGPT, Perplexity, Claude, etc.)" as an explicit response option in post-sale NPS surveys or customer calls. The percentage who select an AI assistant is your AI discovery rate. Apply that rate to total closed-won ARR to produce your AI dark funnel revenue estimate.

Method 2 — Branded search correlation. Track the correlation between your share-of-model scores (Metric 1) and branded search volume in Google Search Console. In companies with established AEO programs, a 10-point share-of-model improvement correlates with a 12-18% branded search volume increase over the following 60 to 90 days. This correlation gives you a leading indicator: branded search volume trends are a lagging proxy for AI discovery volume trends.

Method 3 — Direct navigation lift analysis. When AI assistants recommend a brand, the most common user behavior is to open a new browser tab and navigate directly to the recommended domain. Track direct session volume month over month and look for inflections that correlate with share-of-model changes.

What to report to the board: A single AI dark funnel ARR estimate, methodology-noted, with the survey sample size and confidence interval. Most boards respond well to a conservative estimate ("we believe AI search is influencing at least $X in annual ARR, based on a sample of Y closed deals") rather than an unsourced large number. For most B2B SaaS companies running the closed-won survey for the first time, the initial estimate comes in between 12% and 28% of new ARR. That number tends to surprise boards — in a useful way.

Metric 6: LLM Accuracy on Product Facts

This metric extends citation accuracy (Metric 2) to the specific category of product facts that most directly influence purchase decisions: pricing, feature availability by tier, integration compatibility, and use case fit by customer segment. AI assistants are frequently wrong about these facts in ways that are consequential — not just embarrassing — because buyers use these facts to filter vendor shortlists.

The most common accuracy failures in 2026:

Why this matters beyond accuracy: When an AI assistant tells a prospect that your product includes a feature that is actually only in a higher tier, and that prospect arrives at a demo expecting that feature, the demo dynamic is adversarial from the first minute. Sales teams in companies with poor LLM accuracy scores report higher "feature gap" objections in demos — objections that are not actually about missing features but about AI-generated misaligned expectations. Tracking LLM product-fact accuracy is, in part, a sales productivity metric.

How to improve it: The remediation is documentation-first. Product pages with declarative, tiered feature descriptions, structured pricing tables, and up-to-date integration lists are the single most effective lever. Companies that implement comprehensive schema markup and entity context on pricing and feature pages see accuracy improvements of 15 to 25 percentage points within 60 to 90 days of implementation.

Metric 7: Competitor Citation Gap

The competitor citation gap is the delta between your share-of-model score and the category leader's share-of-model score on the same prompt set. It is the most board-readable of the seven metrics because it directly answers the question boards are increasingly asking: "Are we winning or losing in AI search?"

How to present it:

A simple visualization works: a bar chart showing share-of-model scores for your top four competitors and your own brand, across the same head-term prompt set. The gap between you and the category leader — expressed both as percentage points and as an implied pipeline-risk multiple — is the single most impactful number you can show a board.

The pipeline-risk multiple calculation: if AI-influenced discovery accounts for an estimated 20% of new ARR (your dark funnel estimate), and the category leader has a 3x share-of-model advantage (e.g., 60% vs 20%), then you are reaching roughly one-third the AI-influenced pipeline that the leader is reaching. That is a $X million annual pipeline deficit at your current ARR scale. That number gets budget approved.

The competitive intelligence value: Competitor citation gap measurement also tells you where competitors are getting cited that you are not. Running the detailed prompt-level analysis — not just the aggregate share score but the specific queries where the competitor appears and you do not — produces a content gap map. Each gap is a specific comparison page, documentation section, or use case essay that you do not have and they do. That map is the AEO content roadmap.

Building the Dashboard: Architecture and Tooling

Having seven metrics is only useful if they are surfaced in a format that drives action rather than just report cards. The dashboard architecture that works for board-level AEO reporting has three layers.

Layer 1 — Executive summary (board deck): One page, four numbers: share of model (your score and the category leader's), citation accuracy rate, AI dark funnel ARR estimate, and competitor citation gap. Trended quarterly for the last four quarters. Color-coded against thresholds (green: on track, yellow: monitoring required, red: intervention needed). This page belongs in every board deck from Q2 2026 onward.

Layer 2 — CMO operational dashboard (weekly): Share of model by engine (ChatGPT, Perplexity, Claude, Gemini separately), citation accuracy rate by fact category, branded vs unbranded citation ratio, comparison-page citation rank by query. Updated weekly via automated tooling. Used for prioritizing content and documentation investments.

Layer 3 — AEO analyst working view (daily): Full prompt-level citation data, specific inaccurate claims requiring remediation, new competitor content detected in citation results, changelog from AI model updates that may have shifted citation behavior. Used by the AEO team for day-to-day optimization.

Tooling stack for 2026:

The total tooling cost for a complete seven-metric AEO measurement stack is $2,300 to $5,300 per month in software, plus internal labor for survey administration and audit runs. For a company with $20M or more in ARR, that investment is well below the threshold that would require board approval — and the business case for the investment is built directly from the metrics it produces.

The Reporting Cadence That Works

The CMOs reporting AEO metrics most effectively in 2026 are running a three-tier cadence.

Monthly: Full seven-metric dashboard review with the VP of Marketing and demand gen leadership. Identify the two or three metrics that need intervention. Assign owners and 30-day improvement targets.

Quarterly: Board presentation of the executive summary layer. Lead with the dark funnel ARR estimate (the revenue story), follow with share of model and competitor citation gap (the competitive story), close with citation accuracy rate (the operational quality story). This ordering — revenue, then competitive, then operational — follows the hierarchy of what boards actually care about.

Annually: Full competitive AEO audit. Run the complete prompt battery across every major competitor. Map the citation gaps. Update the 12-month content and documentation roadmap based on findings. Present the annual audit as the strategic context for the next year's AEO budget request.

The board narrative arc: In Q1, you introduce the metrics and establish the baseline ("here is where we are"). In Q2, you show the first movement ("share of model on head terms improved from 22% to 28%, correlated with a 14% branded search volume increase"). In Q3, you connect the movement to pipeline ("dark funnel survey now shows 23% of closed-won citing AI assistant as first touchpoint, up from 18%"). By Q4, you have a credible four-quarter data story that links AEO investment to pipeline conversion. That data story is the foundation for the following year's budget conversation.

Getting Buy-In: The Political Infrastructure

The seven metrics require buy-in from functions beyond marketing to measure and improve. Share of model is primarily influenced by documentation (product or engineering), comparison pages (content marketing), and changelog quality (product and engineering). Citation accuracy is owned by whoever maintains the product pages and documentation. The CMO who tries to run this program as a marketing-only initiative will be blocked at every turn.

The organizational move that works is framing AEO as a revenue-protection initiative at the executive team level, not a marketing optimization. The competitor citation gap is the entry point: "We are invisible in 35% of the AI-assisted buying conversations in our category. Our primary competitor is visible in 78% of them. Here is the revenue implication." That framing gets an executive-level working group. The working group is what gets documentation updated, changelogs published publicly, and comparison-page programs staffed with the people who actually understand the competitive landscape.

For a detailed treatment of how to engineer citations from the content and technical side, ChatGPT citation engineering is the companion read. For how to measure AI search influence before your tooling stack is fully built, the AEO citation tracking playbook has the manual methodology. And for the broader picture of what is happening to B2B pipeline as AI search matures, the share of model measurement framework is essential context.

The Benchmark Table Every CMO Should Have

The most requested output from CMOs building their first AEO board presentation is a benchmarking reference — something that lets them contextualize their current scores against what good looks like. The following table is based on Signal's analysis of 47 B2B SaaS companies across seven categories, measured over 12 months through Q1 2026.

These benchmarks are directional, not precise — the variance within each category is substantial, and vertical market, company size, and documentation quality are all significant moderating factors. But they give a CMO a defensible reference frame for telling the board whether the current scores represent a critical problem, a developing position, or a mature program.

The Playbook for CMOs Starting From Zero

If your company has no AEO measurement in place today, the sequence that gets you to a credible board presentation in 90 days:

1. Establish the baseline (Days 1–30). Run a manual share-of-model audit on 50 head-term and comparison queries across ChatGPT, Claude, Perplexity, and Gemini. Score your citations and your top two competitors'. Run the product-fact accuracy battery on 40 questions. Calculate your branded citation share. You now have the baseline that every subsequent metric will measure against.

2. Run the closed-won survey (Days 15–45). Add an AI assistant option to your post-sale customer calls or NPS survey. Collect responses from 30 to 50 recent closed-won deals. Calculate your AI dark funnel discovery rate. Apply it to the last 12 months of closed-won ARR to produce your dark funnel revenue estimate.

3. Identify the top 5 accuracy failures (Days 30–45). From the product-fact battery, identify the five most common inaccurate AI claims about your product. Assign remediation owners. Fix the underlying documentation, product pages, or structured data within 30 days.

4. Build the executive summary slide (Days 45–60). Four numbers: share of model, citation accuracy, dark funnel ARR estimate, competitor citation gap. Add a trend note ("baseline established; Q3 targets set"). This is slide-ready for the next board meeting.

5. Deploy tooling for ongoing measurement (Days 60–90). Sign up for Profound or an equivalent. Configure the prompt set as a standing weekly run. Build the GA4 custom channel groupings that surface AI-referred traffic (the GA4 AEO configuration guide has the step-by-step setup). You now have a measurement infrastructure that produces the seven-metric dashboard automatically.

6. Present at the next board meeting with a three-quarter roadmap. Frame it as a new measurement capability, not a remediation report. "We have built visibility into AI-assisted discovery for the first time. Here is where we are, here is the competitive gap, and here is the 9-month roadmap to close it." That framing positions the CMO as ahead of the problem rather than behind it.

The CMOs who are presenting these metrics in Q2 2026 board meetings are consistently reporting one of two outcomes: either the board reacts with recognition ("this is exactly what we have been asking about") or with surprise at the competitive gap ("why are we at 22% and our competitor is at 65%?"). Both reactions produce the same outcome: budget approved, cross-functional resources aligned, and an executive mandate for the program. The data, once visible, tends to make the case for itself.

Takeaway: The CMO who can report AI search visibility in a board meeting has a narrative advantage that no amount of organic traffic reporting can match in 2026. The seven metrics — share of model, citation accuracy, branded citation ratio, comparison-page citation rank, AI dark funnel pipeline, LLM product-fact accuracy, and competitor citation gap — form a coherent framework that maps from content investment to competitive position to pipeline influence. Building the baseline takes 30 days. Getting to board-ready reporting takes 90. The companies running this measurement program today will have four quarters of trend data by the time their competitors start building it, and that measurement lead compounds into a strategic advantage that is genuinely hard to close.