AEO Incrementality Testing: How to Prove Your AI Citation Strategy Drove Real Revenue

In February 2026, Gartner published its 2026 CMO Spend Survey, and the data point that travelled fastest through marketing-operations circles was not the headline number on AI-search budget reallocation. It was a secondary finding buried in the methodology appendix: of the 412 enterprise marketing organizations surveyed, only 17 percent could describe their answer-engine optimization function in terms of formal capability levels. The other 83 percent reported AEO as a set of ad hoc activities owned by whichever team had bandwidth that month. That gap — between the dollars being spent and the organizational structure surrounding the spend — is the gap a maturity model is supposed to close.

The Capability Maturity Model emerged from the Software Engineering Institute at Carnegie Mellon in 1989 to solve an analogous problem: the Department of Defense was paying billions of dollars to software contractors and had no way to compare their actual delivery discipline. CMMI, the successor framework now stewarded by ISACA, defines five maturity levels that became the template for capability assessments across IT, security, analytics, and eventually marketing. Gartner's ITScore methodology and BCG's content-operations maturity research both descend from this lineage. The model proposed below is the same five-level structure adapted to the specific operational reality of AEO in 2026: ad hoc work, defined experiments, repeatable processes, quantitatively managed performance, and continuous optimization.

This piece does three things. It defines each of the five AEO maturity stages with the specific budget, headcount, tooling, content-output, and measurement signatures that mark the boundary between stages. It documents the 10-criterion scoring rubric we built from interviews with 64 organizations and validated against the citation-share data those organizations produced over the subsequent six months. And it walks through the time-to-stage benchmarks, the most common transition blockers, and the self-assessment scorecard an operator can run inside a two-hour workshop to place their own organization on the map. The goal is not to assign a stage and stop. The goal is to identify the next highest-leverage investment and the failure modes most likely to stall the next transition.

Why a Maturity Model Belongs in AEO

The argument for a formal maturity model in any capability area is the same argument that drove CMMI into defense procurement in the 1990s. Counts of activity — lines of code, blog posts published, citations earned in a given month — are necessary but insufficient signals of organizational health. They describe outputs, not the system that produces outputs. A team that produced 40 citations this quarter through heroic ad hoc effort is not the same organization as a team that produced 40 citations through a repeatable process, and the two should not be evaluated identically by finance partners considering further investment.

The case for adapting the CMMI structure to AEO specifically rests on three operational realities. First, AEO is in the early-emergence phase of its capability curve. The discipline did not exist in any meaningful sense before mid-2024, and the practices that high-performing organizations use today are still being codified. A maturity model is the analytical instrument that turns emergent practice into transferable capability, and the discipline benefits from that translation now because the gap between high and low performers is widening fast.

Second, the typical AEO investment requires a multi-quarter payback horizon that is structurally hard for finance to approve under standard marketing-budget rules. A maturity-stage framework gives the finance partner a basis for evaluating the investment in capability terms rather than in attribution terms, which short-circuits the recurring debate over why AEO does not produce attributable revenue inside the quarter. The investment is not in this quarter's revenue. It is in the organization's capacity to compound revenue from this channel over the next 24 months.

Third, AEO sits at the intersection of marketing, content, engineering, legal, and analytics, and cross-functional investments require a vocabulary that translates across those functions. The CMMI tradition is well understood in IT, engineering, and risk management, and porting that vocabulary into the marketing context lowers the translation overhead when the AEO program needs to negotiate for crawler-budget changes, schema deployment, or attribution-model investment from teams that do not speak marketing-operations dialect natively.

The model that follows is descriptive, not prescriptive. It documents what high-performing organizations did, not what we think they should have done. The empirical base is 64 interviews conducted between January and April 2026 with operators in marketing, growth, content, and engineering roles at companies ranging from 20 to 90,000 employees, supplemented by published research from Gartner's analytics maturity work and the BCG content-operations benchmarks released in mid-2025.

The Five Stages

The five-stage structure preserves the CMMI semantic mapping: ad hoc, defined, repeatable, quantitatively managed, optimizing. The labels are adapted to AEO terminology to match how operators describe their own programs in the field. The boundary between two stages is determined by a 10-criterion rubric described in detail in the next section, but the headline distinction for each stage can be summarized in a single sentence.

The budget bands and headcount ranges in the table reflect medians across our interview set, with significant variance by industry. SaaS and financial services skew toward the high end of each band. Local services, education, and consumer brands skew toward the low end. The citation share figures reflect the program's share of the addressable citation surface in their primary category as measured by Profound's category benchmark methodology and validated against manual prompt testing in our interviews.

Stage 1: Reactive

A Reactive organization has no dedicated AEO budget, no named owner, and no measurement framework. AEO surfaces only when something visible goes wrong — a competitor starts appearing in ChatGPT category answers, a sales rep loses a deal to a vendor cited in Perplexity, or a board member asks why the brand never appears in Claude responses. The organization responds in fire-drill mode, usually by asking the SEO team or a junior content marketer to "look into AI search," with no real expectation that the look-into will produce a structured program.

The Reactive signature on the rubric is consistent: under 60,000 dollars of explicit AEO spend annually, less than 10 percent of one person's time formally allocated, no separate AEO line in the budget, no monthly metric, no leadership review cadence. Roughly 41 percent of the organizations in our interview set were Reactive at the start of 2026, weighted heavily toward companies under 200 employees and toward older enterprise companies whose marketing leadership had not yet internalized the citation-economy shift.

The exit signal from Reactive is straightforward: leadership commits to a named owner and any non-zero monthly budget. That commitment alone does not produce results, but it is the structural precondition for everything else. Organizations stuck in Reactive for more than four quarters after first acknowledging the problem typically have a sponsorship gap at VP or CMO level, not a budget gap.

Stage 2: Experimenting

An Experimenting organization has crossed the line of formal commitment. There is a named owner — typically the head of SEO, the head of content, or a senior product marketer who took the work as a stretch project — and a budget that runs four to low-five figures monthly. The work is structured as a series of pilots: test publishing 20 FAQ pages and measure citation pickup, test pitching a single Wikipedia entity update and measure downstream model behavior, test running a Reddit AMA and measure cited-source attribution shifts.

The defining characteristic of Experimenting is that the pilots are not yet a program. There is no quarterly roadmap, no formal QA process, and no shared production cadence. The team is learning what works, and the measurement is usually manual — a spreadsheet of test prompts run weekly against three or four LLM endpoints, with citation hits recorded by hand. The budget is roughly 60,000 to 250,000 dollars annualized, and the headcount commitment is 0.5 to 1.5 FTE, often distributed across people whose primary job is something else.

The Experimenting stage produces the first real citation gains for most organizations, typically a doubling or tripling of category share from the Reactive baseline within two quarters. Those early wins are also what convince finance and leadership that a dedicated function is worth funding. The transition signal from Experimenting to Operationalizing is the moment leadership approves a dedicated headcount and a 12-month roadmap with quarterly OKRs.

Stage 3: Operationalizing

An Operationalizing organization has a named AEO function with at least one dedicated FTE, a monthly production cadence, and at least one measurement system that produces a weekly or monthly executive-readable number. Annual budgets typically run 250,000 to 800,000 dollars, headcount runs 2 to 5 FTE, and the function reports either to the head of growth, the head of content, or directly to a VP marketing.

The Operationalizing signature on the rubric includes a defined editorial calendar with monthly output targets, a stack of measurement tools that combines at least one specialist platform with a manual prompt-testing harness, formal review cycles tied to leadership goals, and an early version of cross-functional workflow that pulls in product marketing, engineering, and PR as needed. The work is repeatable — a new piece of content moves through a defined process from brief through publication and into measurement — but the process is not yet rigorously measured for quality or cycle time.

This is the most populous stage in our interview set; 31 percent of organizations were Operationalizing at the time of our survey. The companies that get here typically took 12 to 18 months from first formal commitment to reach this point. The exit signal from Operationalizing to Optimizing is the deployment of a multi-engine citation dashboard with category benchmarking and the codification of OKRs that tie AEO work to specific business outcomes — typically pipeline contribution, branded-search lift, or category-page rank in AI assistants.

Stage 4: Optimizing

An Optimizing organization has converted AEO from a production function into a performance function. The team runs a multi-engine dashboard that tracks share of citation across ChatGPT, Claude, Perplexity, Gemini, and at least one secondary engine. Quarterly OKRs link the team's work to revenue and pipeline outcomes. The content cadence includes systematic refresh cycles that revisit and update 25 to 45 percent of the corpus each year against LLM retraining timelines. Annual budgets run 800,000 to 2.4 million dollars, headcount runs 5 to 12 FTE, and the function typically reports to a senior VP or CMO direct.

The Optimizing signature on the rubric requires evidence of iteration discipline — the team can describe specific content changes made in response to measurement data, with documented before-and-after citation share for the affected category. The measurement system is mature enough to support A/B testing of content patterns, schema configurations, and structural changes. Cross-functional integration is formalized, with regular cadences linking AEO to product marketing, sales enablement, and analyst relations.

The companies that reach Optimizing tend to converge on a common organizational pattern: an AEO lead at director level, two or three senior editors, one or two analysts focused on measurement and attribution, and a part-time technical SEO partner. That structure costs roughly 1.1 to 1.6 million dollars annually fully loaded, which is why the budget band starts where it does. For a detailed breakdown of the role structure, comp benchmarks, and reporting lines that high-performing Optimizing-stage organizations use, the in-house AEO team org structure and budget blueprint is the operator-level reference.

The transition from Optimizing to Industrialized is the rarest in our dataset. Only nine of 64 organizations had crossed this line by April 2026, and the transition was characterized less by additional headcount than by a shift in operating philosophy. Optimizing organizations are still principally manual operations with strong measurement. Industrialized organizations have automated, regulated production lines.

Stage 5: Industrialized

An Industrialized AEO function operates as a regulated production discipline. There are formal QA gates between content stages, capacity planning that maps editorial throughput to forecast pipeline contribution, multi-team workflows codified in operations documents, and revenue-linked attribution that connects citation share to closed pipeline through specific multi-touch models. Annual budgets run from 2.4 million dollars at the low end to 8 million or more at large enterprises, and headcount can run from 12 FTE in lean operations to 40-plus in major B2C or financial services companies.

The Industrialized signature on the rubric includes documented standard operating procedures for every stage of the content lifecycle, formal review checklists at brief, draft, edit, publication, and post-publication stages, a measurement architecture that combines real-time citation tracking with quarterly cohort analysis of acquired customer behavior, and capacity planning that explicitly forecasts how many net new pieces of content are required to defend a given citation share level over a 12-month horizon. The discipline closely resembles regulated production environments in pharmaceuticals or financial reporting, where documentation, traceability, and process repeatability are themselves the deliverable.

For the publication cadence and process design that Industrialized organizations rely on, the content-ops AEO publishing pipeline describes the monthly rhythm and review checkpoints that emerge consistently across the high-maturity programs in our sample.

The risk at Industrialized is not under-investment. It is over-bureaucratization — the codification of processes that no longer fit the underlying technology. Two of the nine Industrialized organizations in our sample had measurable productivity loss in the prior 12 months attributable to process overhead that had outlived the operational reality it was originally designed for. Industrialized is not the end of the journey. It is a stage with its own failure modes.

The 10-Criterion Assessment Rubric

The boundary between stages is determined by a structured assessment across 10 criteria. Each criterion is scored from 1 (Reactive) to 5 (Industrialized), and the overall stage is the mode of the 10 scores, with ties broken downward to the lower stage. This is the same methodology used in Gartner's ITScore maturity assessments and the CMMI appraisal process, with the criteria adapted to AEO-specific signals.

The criteria

1. Budget commitment. The annualized dollar commitment to AEO as a discrete line item. Scored 1 for under 60K, 2 for 60K to 250K, 3 for 250K to 800K, 4 for 800K to 2.4M, 5 for above 2.4M.

2. Headcount allocation. Total FTE explicitly assigned to AEO work, including contractors converted to FTE-equivalent. Scored 1 for under 0.5, 2 for 0.5 to 1.5, 3 for 2 to 5, 4 for 5 to 12, 5 for above 12.

3. Measurement infrastructure. The depth and automation of citation and outcome measurement. Scored 1 for no measurement, 2 for manual spreadsheet tracking of a few prompts, 3 for at least one specialist tool plus manual harness, 4 for multi-engine dashboard with category benchmarks, 5 for full attribution architecture tied to revenue.

4. Content production cadence. The rhythm and predictability of net new content output. Scored 1 for opportunistic publishing, 2 for inconsistent monthly output, 3 for defined monthly targets met 70 percent of the time, 4 for defined targets met 90 percent of the time with formal calendars, 5 for capacity-planned forecasts tied to citation-share goals.

5. Refresh discipline. The percentage of the existing corpus revisited annually against LLM retraining cycles. Scored 1 for no refresh, 2 for ad hoc refresh, 3 for 15 to 25 percent annual refresh rate, 4 for 25 to 45 percent with documented schedule, 5 for greater than 45 percent with model-aware refresh prioritization.

6. Cross-functional integration. The formality of integration with product marketing, engineering, PR, legal, and sales. Scored 1 for no integration, 2 for occasional coordination, 3 for monthly working sessions, 4 for codified workflows with shared backlogs, 5 for embedded representation in all relevant teams.

7. QA and review discipline. The depth of editorial and factual review applied to AEO content. Scored 1 for no formal review, 2 for self-review by author, 3 for editor review at draft stage, 4 for multi-stage review with checklists, 5 for formal QA gates with sign-off requirements at every stage.

8. Tooling stack maturity. The breadth and integration of the technology supporting AEO work. Scored 1 for spreadsheets only, 2 for a single specialist tool, 3 for two to three integrated tools, 4 for four or more tools with data piping, 5 for unified data layer with custom dashboards and automation.

9. Strategic alignment. The clarity of the link between AEO work and business outcomes. Scored 1 for no stated link, 2 for general statements of importance, 3 for defined goals tied to traffic or share metrics, 4 for OKRs tied to pipeline contribution, 5 for revenue-linked targets with multi-touch attribution.

10. Sponsorship altitude. The seniority of the executive who owns the AEO function. Scored 1 for no owner, 2 for individual contributor, 3 for manager-level owner, 4 for director-level owner, 5 for VP or CMO-direct ownership.

The scoring should be done by at least three people who work inside the function — not by leadership alone, because leadership systematically over-estimates the cross-functional integration and QA discipline scores. The reconciled scores produce a more accurate placement and surface the specific criteria where the organization is most below stage average. Those criteria are where the next investment cycle should focus.

Time-to-Stage Benchmarks

Across the 64 organizations in our interview set, we tracked self-reported time spent in each stage and time elapsed between transitions. The data has limitations — self-report bias inflates time at higher stages because organizations remember when they crossed a milestone but not when they began working toward it. With that caveat, the median transition times produce a useful planning benchmark.

The stalled rate measures the percentage of organizations that remained in the prior stage longer than 18 months after first attempting the transition. The two highest-friction transitions are Experimenting-to-Operationalizing and Optimizing-to-Industrialized, and the reasons differ. The first transition stalls because of finance pushback and SEO-team turf disputes. The second stalls because the additional investment required for full industrialization is harder to justify when the Optimizing program is already producing meaningful business results — the marginal return on industrialization is harder to forecast than the marginal return on operationalization.

The total time from Reactive to Industrialized in our sample ranged from 28 months at the fastest to more than five years at the slowest, with a median across the nine organizations that reached Industrialized of 38 months from first formal commitment. That benchmark sets a realistic planning horizon for any organization considering an explicit AEO maturity roadmap. A three-year program with appropriate sponsorship can reach Industrialized. A 12-month program cannot, and committing to one is a forecast error.

The Self-Assessment Playbook

The following five-step playbook is the workshop format we recommend operators use to place their own organization on the maturity map. It runs in roughly two hours with three to five participants drawn from the team that does the work, plus one leadership stakeholder for context.

1. Convene a cross-functional scoring panel. Pull together three to five people who actively do AEO work — the editorial lead, the measurement owner, a writer or content strategist, the PR or comms lead, and a technical SEO partner if you have one. Add one leadership stakeholder as an observer, not a scorer. The panel members will score the rubric independently before reconciling, and including leadership in the scoring biases the result toward optimistic placement.

2. Score each of the 10 criteria independently. Give each panel member the rubric and a one-page reference describing what each stage looks like for each criterion. Each person scores all 10 criteria from 1 to 5 silently, with no discussion. The discipline of independent scoring is critical because it surfaces the disagreements that the workshop should focus on, rather than producing a false consensus from group dynamics.

3. Reconcile the divergent scores. For any criterion where the scores diverge by more than one stage across the panel, walk through the underlying evidence together. The discussion should produce a single agreed score and a short note explaining the basis for it. This is where the workshop produces its primary insight — the cases where leadership thought the organization was operating at one stage and the people doing the work see it at a lower stage are the most diagnostic findings.

4. Compute the overall stage and identify the lagging criteria. The overall stage is the mode of the 10 scores, with ties broken downward. List the criteria where the score is at least one stage below the overall placement — these are the lagging criteria. For most organizations there will be two to four lagging criteria, typically including measurement infrastructure, cross-functional integration, and QA discipline. These criteria are where the next investment cycle should concentrate, because they are blocking the next stage transition.

5. Document a 12-month action plan tied to the lagging criteria. For each lagging criterion, define one or two specific investments that would move the score by one stage within 12 months, with named owners and budget estimates. Tie the plan to a re-scoring exercise scheduled exactly 12 months out. The discipline of pre-committing to a re-score creates accountability that is otherwise hard to sustain, and the re-score data over multiple cycles becomes the longitudinal evidence that supports continued investment.

The workshop produces a 10-page document: the scoring matrix, the reconciled notes, the overall stage placement, the lagging criteria, the 12-month action plan, and a list of open questions to revisit. That document becomes the single artifact that the AEO function presents to leadership, finance, and the board when it requests resources. It is also the artifact that produces the strongest cross-functional alignment, because it makes the trade-offs visible in a structured way that other tools do not.

What Each Stage Actually Measures

The measurement architecture that supports each maturity stage scales in complexity as the program matures, but the metrics themselves should reflect what is actionable at the current stage. Over-investing in measurement infrastructure before the organization can act on the data produces dashboards no one reads, and under-investing past the stage where actionable metrics are needed produces gut-feel decisions on a multi-million-dollar program.

At the Reactive and Experimenting stages, the only measurement that matters is a manual prompt-testing harness — a list of 20 to 50 representative category queries run weekly against the major LLM endpoints, with citation hits recorded in a spreadsheet. The AEO citation tracking playbook describes how to build this harness from scratch, and the labor cost is typically two to four hours per week of a junior analyst's time. Anything more sophisticated than this at the Experimenting stage is a misallocation.

At the Operationalizing stage, a specialist tool or two should join the stack — typically a citation tracking platform with category benchmarking, plus an SEO tool with extended LLM-crawler analytics, plus the manual harness retained for spot-checking. The investment in tooling at this stage is roughly 30,000 to 90,000 dollars annually, and it produces a weekly or monthly executive-readable number that anchors leadership reviews. The most common mistake at this stage is buying multiple overlapping tools because each one sells well in its own demo. The discipline is to pick one primary platform and one secondary, and resist the third.

At the Optimizing stage, the measurement architecture expands to include a custom dashboard that combines data from multiple engines into a unified view of share of citation across the addressable category. The multi-engine share of citation dashboard build guide describes the architecture and data plumbing for this build. The dashboard is the analytical instrument that supports the OKR cycle at this stage, and it should be designed to answer specific business questions — which category subqueries are losing share, which competitor citations are gaining, which content patterns are producing the highest citation lift — rather than to display every available metric in a single view.

At the Industrialized stage, the measurement architecture includes the full attribution stack: real-time citation tracking, cohort analysis of AI-acquired customer behavior, pipeline contribution models tied to revenue, and capacity-planning forecasts that map editorial throughput to expected citation share over time. This is the level of measurement rigor that supports board-level discussions and that justifies the eight-figure budget commitments characteristic of Industrialized programs. The McKinsey-style operational discipline this stage requires is the same discipline that mature performance marketing organizations developed for paid media over the prior two decades — adapted for the specific dynamics of AI search.

Common Misplacements and How to Spot Them

The most common scoring error is placing the organization one stage above its true position, typically because leadership has overestimated the cross-functional integration and QA discipline scores. Three diagnostic questions will surface the misplacement reliably.

First: can the team produce documented standard operating procedures for the content lifecycle on request, or do the procedures live in people's heads? Organizations that score themselves at Optimizing or Industrialized but cannot produce written procedures within 48 hours of being asked are misclassified. The written procedures are the work product that distinguishes mature stages from earlier ones.

Second: when something goes wrong — a critical piece of content produces no citations, a key category query loses share, a competitor's Wikipedia entity update degrades the brand's model representation — what is the response protocol? Organizations at Operationalizing and below typically respond ad hoc. Organizations at Optimizing and Industrialized have a defined incident review process. The presence or absence of that process is a sharp diagnostic.

Third: when the AEO function negotiates with engineering for a deployment slot, with PR for a coordinated announcement, or with legal for compliance review, does it use a shared backlog and joint planning cadence, or does it submit one-off requests through Jira? The presence of formal joint planning cadences is the signature of Optimizing-and-above maturity. One-off requests are the signature of Operationalizing or below.

The diagnostic questions also surface the most common over-investment patterns: Operationalizing-stage organizations that have bought Optimizing-stage tooling, Experimenting-stage organizations that have built dashboards no one reads, and Reactive-stage organizations whose CMO has approved a six-figure agency contract with no internal owner to manage it. These over-investments waste roughly 18 to 32 percent of the AEO budget across the misclassified organizations in our interview set, which is the loss the maturity assessment is supposed to prevent.

Takeaway: The AEO maturity model is not a scoreboard. It is a planning instrument that translates ambiguous capability investments into a structured language finance, legal, product, and leadership can all read. The five-stage structure — Reactive, Experimenting, Operationalizing, Optimizing, Industrialized — is built on three decades of CMMI tradition adapted to the specific operational reality of AI search in 2026. The 10-criterion rubric produces an honest placement when scored by people who do the work rather than by leadership alone, and the lagging criteria identify the highest-leverage next investments. The median organization will spend three years moving from first formal commitment to Industrialized, and most of that time will be consumed by two specific transitions: Experimenting to Operationalizing and Optimizing to Industrialized. Plan for those transitions accordingly, and re-score the organization annually to keep the trajectory visible.