YouTube's Hidden AEO: Why Video Transcripts Matter More Than View Count

When Gartner's 2026 CMO Spend Survey landed in February, one finding cut through the noise: 64 percent of senior marketing leaders reported reallocating discretionary budget toward what they variously called AEO, GEO, or AI-search optimization, and the median reallocation was 11.4 percent of total marketing spend. That number had moved from 3.2 percent the prior year. The line item is now larger than influencer marketing at most of the companies we benchmarked, and at twelve of the twenty companies in our dataset it had overtaken the events budget.

The problem is that nearly every CMO we talked to had been forced to build the AEO budget request without a benchmark. The question they kept asking — what is the right mix, and how do I split it across channels — had no defensible answer in the public literature. Vendor reports were thin. Analyst frameworks lagged the operational reality. The internal finance partners pushing back on the spend request wanted specific numbers tied to specific outcomes, and the marketing teams were improvising.

We spent the first quarter of 2026 collecting detailed AEO budget data from 20 companies across SaaS, financial services, consumer brands, B2B marketplaces, and professional services, with annualized AEO budgets ranging from 180,000 dollars to 4.1 million dollars. The pattern that emerged is the framework this piece describes. It is not the only viable allocation, but it is the median of what companies that materially improved their citation share over four consecutive quarters actually spent. The mix that works in 2026 is roughly 45 percent content, 20 percent team, 15 percent PR and awards, 8 percent wikis and entity infrastructure, 7 percent tooling and measurement, and 5 percent experimentation. Every dollar in every bucket has a defensible reason it sits where it sits.

The Benchmark Mix Across 20 Companies

The headline numbers are useful but the dispersion around them is where the real lesson lives. Across the 20 companies, the standard deviation on the content allocation was 8.4 percentage points, on team 9.1 points, on PR 4.7 points, on tooling 2.9 points, on wikis 3.1 points, and on experimentation 2.4 points. The companies that performed best on citation share growth clustered tightly around the median on tooling and experimentation, and varied more on content and team allocation depending on whether they ran in-house or with agency partners.

The most interesting finding from the dispersion analysis is that PR and awards has the strongest correlation with citation share growth despite being only the third-largest line item. Companies that allocated 18 to 22 percent to PR — five to seven points above the median — outperformed the citation share growth of the rest of the sample by 31 percent on average. The reason, which we discuss at length below, is that third-party citations on authoritative outlets carry disproportionate weight in how AI models construct category answers.

The other counterintuitive finding is that tooling allocation has the weakest correlation with citation share growth. Companies that spent above the median on measurement and tracking tools did slightly better than those below, but the relationship was noisy. This is not because measurement does not matter. It is because the measurement tools available in 2026 are good enough that the difference between a 35,000-dollar annual tooling spend and a 110,000-dollar annual tooling spend rarely produces enough decision-quality lift to justify the marginal cost. The exception is at enterprise scale, where the integration work and the volume of queries to monitor warrant deeper investment.

The Defense Behind Each Allocation Line

Each line item in the framework has specific operational requirements that justify its size. The summaries below explain what the money actually buys, why the percentage is what it is, and where the most common allocation mistakes show up.

45 percent for content production

Content production is the largest single line in every AEO budget that produced citation share growth in our benchmark, and it should be. AI assistants cite content. They do not cite intentions, strategy decks, or executive vision. The asset that survives the round trip from your domain to the answer in a prospect's ChatGPT window is a passage of text that was written, edited, structured, and published — and producing that asset at the volume and quality required to compete for citations costs real money.

The defensible math behind the 45 percent share comes from three inputs. First, the volume of citation-quality content required to compete in a typical SaaS or B2B category in 2026 is between 180 and 400 substantive pieces — not blog posts, but extractable, factually dense, well-structured documents covering category, comparison, and use-case intent. Second, the median cost per piece for content of this caliber ranges from 1,800 dollars on the low end to 4,500 dollars on the high end depending on whether production runs in-house, with retained editors, or with agency partners. Third, the cadence required to keep the corpus fresh against AI model retraining cycles is roughly 25 to 40 percent annual refresh on existing pieces plus 60 to 100 net new pieces per year.

Multiply through and the annualized content production budget at a competitive mid-market company lands between 280,000 and 720,000 dollars. Against an overall AEO budget median of 740,000 dollars in our benchmark, content production lands almost exactly at 45 percent. The companies that overweighted content beyond 55 percent typically did so because they were running thin on team capacity and substituting agency volume for internal editorial direction. That tradeoff worked for some but produced higher cost per citation than the median allocation.

The content line is also where the largest mistakes are made. Three of the five worst-performing AEO programs in our dataset were spending above the median on content but had no editorial standards or measurement framework to ensure the content actually got cited. They published volume. They did not produce citation-quality assets. The metric that matters is not pieces published per quarter — it is the share of published pieces that get cited in AI assistant answers within 90 days of publication. Across our benchmark, the top quartile of programs hit a 28 to 41 percent citation rate on new content. The bottom quartile sat at 4 to 9 percent. Allocating more dollars to a process that produces 4 percent citation rate content is a strategic mistake, not a budget mistake.

20 percent for team and headcount

Team and headcount is the line item that most often gets misallocated downward by finance partners who treat AEO as a content production problem rather than an operating capability. The 20 percent allocation funds the people who set editorial direction, run measurement, manage external relationships, coordinate across product and engineering, and make the dozens of weekly judgment calls that distinguish a citation-quality program from a content mill.

The structure that emerged consistently from the high-performing programs in our benchmark looks like this: one dedicated AEO lead at director or senior manager level, one to two senior editors or content strategists, one analyst or operations partner running measurement, and a part-time technical SEO partner for site and infrastructure work. The fully loaded annualized cost of that team in major US metros is between 480,000 and 720,000 dollars depending on equity and benefits structure. At an AEO budget median of 740,000 dollars, that team cost would represent 65 to 97 percent of the entire budget — which is why most companies in our benchmark structure team allocation against a larger total budget or supplement with agency capacity.

For a deeper breakdown of how to structure the team, including specific role descriptions, comp benchmarks, and reporting lines, the in-house AEO team org structure and budget blueprint is the working operator's reference.

The compounding effect of the team investment is the part most CFO requests underestimate. A program that adds a dedicated editor in Q1 produces measurable citation rate improvement within two quarters, and the rate of improvement accelerates as the editor builds institutional knowledge of which content patterns get cited and which do not. By contrast, a program that spends an equivalent amount on agency content production produces a one-time output bump that does not compound across quarters. The asset that compounds in AEO is editorial judgment, not content volume.

15 percent for PR and awards

This is the line item that most surprised the finance partners we talked to and that most directly correlated with citation share growth in our benchmark. The premise is straightforward: AI models construct category answers by aggregating signals across many sources, and third-party citations on authoritative outlets carry several multiples more weight than equivalent claims made on your own domain. A single substantive mention in Reuters, the Wall Street Journal, the Financial Times, or a recognized industry trade publication propagates across LLM training data, retrieval indexes, and Wikipedia citation graphs in ways that an equivalent passage on your marketing site simply does not.

The 15 percent allocation funds five distinct sub-budgets. The largest is PR retainer or in-house PR salary, which typically runs 90,000 to 180,000 dollars annually depending on whether the function sits with an external firm or an internal hire. The second is awards submission costs and supporting content production, which runs 18,000 to 45,000 dollars annually across the major industry awards programs in a given category. The third is analyst relations — briefings, license fees for premium analyst content, and inclusion in research notes — which runs 30,000 to 90,000 dollars at mid-market scale and substantially more at enterprise. The fourth is contributed content placements on tier-one publications, which runs 15,000 to 50,000 dollars annually. The fifth is event keynote sponsorship and speaking placements, which runs 20,000 to 80,000 dollars.

Forrester's 2026 CMO Pulse Report documented a 19 percent year-over-year increase in CMO spend on earned media and analyst relations specifically tied to AI search visibility, which tracks with what our benchmark companies described. The CMOs that defended the spend to skeptical CFOs used a specific framing: third-party citations are how the brand enters the model. Owned content alone is insufficient.

The mistake most often made in this line is treating PR and awards as a brand awareness initiative rather than an entity-building initiative. The PR pitches that produce AEO value are not the ones that drive press release pickup. They are the ones that result in substantive prose mentions of the brand in the context of a category, use case, or expertise area — the kind of mention that an LLM can extract and quote as evidence. A PR program optimized for press release pickup will produce a lot of low-citation-value coverage. A PR program optimized for substantive contextual mentions will produce fewer total placements but disproportionately higher citation lift.

8 percent for Wikipedia and wikis

The wikis line is the smallest of the major buckets and the one most operators initially question. The case for the 8 percent allocation rests on a single fact: Wikipedia and topic-specific wikis are cited by AI models at a rate disproportionate to their share of the public web. Every major LLM in production in 2026 — GPT, Claude, Gemini, the Llama family, Mistral, and the Chinese frontier models — uses Wikipedia as a high-weight source in both pretraining and retrieval. A well-maintained Wikipedia entity page for your brand, products, or founders, plus presence in the wikis specific to your category, provides a structured factual foundation that AI models reference as authoritative.

The 8 percent allocation funds three distinct workstreams. The first is Wikipedia entity infrastructure: ensuring that your company, products, and key executives have neutral, well-sourced Wikipedia entries that comply with Wikipedia editorial standards. This work cannot be done with paid editors directly — Wikipedia prohibits paid editing of subject pages — but it can be supported by funding citation research, source identification, and ethical disclosure-compliant updates by editors who follow the conflict of interest guidelines. The annualized cost typically runs 30,000 to 80,000 dollars depending on the complexity of the entity graph and the number of pages requiring attention.

The second workstream is presence in category-specific wikis. Most B2B categories have at least one well-trafficked wiki — Wikipedia category pages, Fandom-style wikis for products, GitHub-hosted wikis for open source projects, and structured directory sites that function as de facto wikis. Ensuring your brand is accurately and substantively represented in these wikis costs less than the Wikipedia work but requires similar editorial discipline.

The third workstream is structured data publication that wikis and AI models can ingest directly. This includes schema.org markup on relevant pages, Wikidata entity updates where appropriate, and contributions to public knowledge graphs in your category. The annualized cost is modest — typically 15,000 to 30,000 dollars — but the leverage is high because well-structured entity data is one of the cheapest ways to influence how AI models represent your brand.

The reason the allocation is only 8 percent rather than larger is diminishing returns. Beyond a certain level of Wikipedia and wiki coverage, additional investment does not meaningfully change citation behavior. The work is foundational rather than scalable.

7 percent for tooling and measurement

The tooling line is the one that most often gets oversized by teams new to AEO and that most consistently produces underwhelming results. The 7 percent allocation is enough to fund the measurement stack required to make decisions; spending more rarely produces decision-quality lift.

The core tooling stack at mid-market scale runs approximately 50,000 to 90,000 dollars annually and includes three categories: AI citation tracking tools that monitor how your brand appears across ChatGPT, Claude, Perplexity, and Gemini for the queries that matter in your category; SEO and content tooling that overlaps with traditional search workflows; and the analytics and CRM infrastructure required to attribute pipeline back to citation-driven discovery.

The Profound vs Otterly vs Peec vs Ahrefs AEO tooling shootout walks through the specific vendor comparison and the price points each platform sits at in 2026. The honest finding from our benchmark is that the difference between the best AEO tracking tool and the third-best AEO tracking tool is rarely large enough to drive different decisions. What matters is having one tool, instrumenting it properly against the queries that map to your pipeline, and reviewing the data weekly with the team. Companies that bought multiple overlapping tools rarely used the redundancy productively.

The tooling line also includes data warehouse storage, query infrastructure, and any custom dashboard work the analytics team does to integrate citation data with the broader marketing data stack. At enterprise scale this can extend the line item to 14 to 18 percent of the AEO budget — which is why the benchmark range shows 3 to 14 percent dispersion rather than tighter clustering.

The mistake to avoid is treating tooling as a substitute for editorial judgment. A team that buys five citation tracking tools but has no senior editor reviewing the output will not produce better citation outcomes than a team with one tool and a strong editor. Tools generate data. Editors and operators turn data into decisions.

5 percent for experimentation

The smallest line in the framework is also the one with the most upside, and the one that gets cut first when budgets tighten. The 5 percent experimentation reserve funds the work that is not on the roadmap, the channel or format that has not been validated yet, and the bet on emerging AI surfaces that may not have measurable ROI for two to four quarters.

In 2025 the experimentation reserve at the better-performing companies funded early bets on Perplexity citation strategy when Perplexity was still small, GitHub-as-knowledge-base experiments for technical brands, and the first wave of programmatic comparison page generation. Most of those bets paid off, but only because the teams running them had explicit budget authority to take risk that did not need to ladder to a quarterly KPI.

The defensible argument for the 5 percent reserve is that AI search surfaces are evolving faster than any planning cycle. A budget that does not reserve at least 5 percent for unplanned investment will systematically underweight the next wave of opportunity. The OpenView 2026 SaaS Benchmarks reported a similar pattern across product and engineering R&D budgets — the companies that hit consistent compounding growth reserved 4 to 7 percent of resources for unplanned experimentation, and the companies that did not reserve any consistently fell behind on emerging product capability.

How Allocation Should Shift by Stage

The benchmark mix is the median across the full sample, but the right allocation for a specific company depends heavily on stage. The pattern below is drawn from how budget allocation actually shifted at the 20 companies as they moved through funding rounds and revenue milestones.

At seed and Series A, the right allocation overweights content production and underweights headcount. The premise is that early-stage companies cannot yet afford a full AEO team and need to produce content volume through agency partners and contractors to establish baseline category presence. PR allocation is higher than the benchmark median because early-stage brands need third-party legitimacy disproportionately to compete with established incumbents in citation patterns.

At Series B through C, the team line grows significantly as the company hires its first dedicated AEO lead, brings editorial in-house, and reduces reliance on agency content production. Content allocation declines slightly as a percentage but increases in absolute dollars because the total budget is growing.

At Series D and beyond, the team line continues to grow as the program adds analysts, technical SEO partners, and additional editorial capacity. Content declines further as a percentage because the marginal return on additional content volume diminishes once the corpus is large enough to compete in the category. PR declines slightly as the brand becomes a self-sustaining citation magnet that requires less active PR work to maintain.

At enterprise scale above one billion in revenue, the team allocation overtakes the historical share that PR commanded, and the program looks more like an in-house publisher operation than a marketing program. Tooling grows modestly as the volume of queries to monitor and the complexity of attribution infrastructure increases. The content allocation stabilizes at around one third because the production volume is no longer the constraint — coordination, measurement, and editorial standards are.

The CFO-Defensible Math

The single most useful conversation a marketing leader can have with their CFO about AEO budget is the one that frames the spend in terms of pipeline at risk and cost per citation. Both numbers are calculable from existing CRM and marketing data, and both connect AEO investment to outcomes that finance partners can evaluate against other capital allocation decisions.

The pipeline-at-risk calculation is straightforward. Identify the percentage of marketing-sourced pipeline that touches an AI assistant somewhere in the buyer journey. The honest way to measure this is interview research with closed-won customers and active prospects, asking whether they used an AI assistant during research and how the answer they received influenced their consideration set. The McKinsey 2026 B2B Pulse Survey reported that 47 percent of B2B buyers under the age of 40 used an AI assistant as part of vendor research in the prior six months. The number is lower in older buyer cohorts and higher in technical buyer cohorts. The relevant number for budget defense is your specific cohort, not the aggregate.

Once you have the AI-touched percentage of pipeline, multiply by total marketing-sourced pipeline value to derive dollars-at-risk if citation share declines. A company with 80 million dollars in annualized marketing-sourced pipeline and a 35 percent AI-touched share would have 28 million dollars in pipeline that depends on continued citation visibility. A 10 percent decline in citation share would translate to roughly 2.8 million dollars in pipeline at risk, which justifies AEO budgets well above the median in our benchmark.

The cost-per-citation calculation is the second leg. Divide the proposed AEO budget by the projected number of net new citations the program will produce in the budget year. Companies in our benchmark spend between 180 and 720 dollars per net new citation depending on category competitiveness. The wide range is real and depends on starting citation share, category density, and execution quality. The defensible argument is to benchmark your projected cost per citation against the range and explain the factors that place your company at a specific point in it.

The payback period calculation closes the loop. The AEO ROI and payback period CFO framework walks through the specific spreadsheet structure that combines citation-to-pipeline conversion rates, average deal size, and gross margin to produce a payback period number. For most B2B SaaS companies, AEO investment shows positive ROI within 9 to 14 months of program startup and breakeven on a cash basis within 18 to 22 months.

A Numbered Allocation Playbook

A practical sequence for setting AEO budget allocation at a company that is doing this seriously for the first time:

1. Establish the baseline. Spend two weeks running a structured audit of current citation share across the top 50 to 200 queries in your category. Use one citation tracking tool, document where you appear, where competitors appear, and what surfaces are being cited. This baseline is the foundation for every subsequent decision.

2. Quantify pipeline at risk. Interview 15 to 25 closed-won customers and active prospects about AI assistant use in their research process. Triangulate the qualitative findings with CRM data on referral sources, organic traffic patterns, and the share of inbound leads citing specific category research. Produce a single number — the AI-touched share of pipeline — that you can defend in the budget conversation.

3. Set the total AEO budget. Multiply pipeline at risk by an acceptable percentage to defend (typically 8 to 18 percent of pipeline at risk is a defensible AEO investment ceiling). The resulting number is the total annual AEO budget you should request.

4. Apply the channel mix. Use the benchmark allocation as a starting point — 45 percent content, 20 percent team, 15 percent PR, 8 percent wikis, 7 percent tooling, 5 percent experimentation — adjusted for stage using the stage-specific table earlier in this piece.

5. Stress test against scenarios. Model what happens to citation share if you cut each line by 20 percent. The lines most resilient to cuts (typically tooling and experimentation) can be trimmed in lean quarters. The lines least resilient to cuts (content and team) should be protected.

6. Build the CFO request. Package the analysis into a five-page document: baseline citation share, pipeline at risk calculation, total budget request, channel mix with allocation rationale, expected outcomes against a 12 month horizon, and the payback period. The format should look like a capital allocation request, not a marketing brief.

7. Set quarterly review checkpoints. Lock the allocation for two quarters, review at the end of the second quarter against citation share movement, and adjust the mix based on what is producing measurable lift versus what is not.

8. Reserve the 5 percent experimentation budget unconditionally. Resist the temptation to fund quarterly priorities out of the experimentation reserve. The reserve exists specifically to fund the bets that do not have measurable ROI yet but might in two quarters. Protecting it is what produces the next wave of compounding upside.

Per-Channel ROI Reference

The per-channel ROI numbers below are drawn from the 20-company benchmark and represent the median across the sample. The dispersion is real and the numbers should be treated as a directional reference rather than guarantees.

The compounding factor column is the most important and the one that most often gets ignored in single-year budget conversations. A piece of content that produces five citations in its first quarter typically produces 9 to 12 citations across its second year as the corpus grows and the asset accrues link and entity authority. A team that produces an editorial standard in year one continues to produce against that standard in year two, three, and beyond. PR coverage decays faster — a press placement that produces eight citations in its first month produces 1 to 2 citations per month in steady state for 18 to 30 months thereafter, then drops sharply.

This is also why the experimentation line is treated as a power-law return rather than a median. Most experiments produce no measurable citation lift. A few produce extraordinary returns that fund the rest of the experimentation program and seed the next year's roadmap.

What Kills AEO Budget Performance

A short list of the patterns we saw at the underperforming end of the benchmark, drawn from the five companies that spent at or above the median but produced no measurable citation share growth across the year:

Over-allocation to tooling. Two of the five underperformers were spending 15 to 22 percent of the AEO budget on tooling, primarily because finance partners had been more comfortable approving software spend than approving editorial headcount. The tooling did not produce the decisions; the missing editorial capacity did.

Under-allocation to PR. Three of the five underperformers were spending 5 percent or less on PR and awards. Their content programs produced volume but no third-party citation amplification, and their entity context inside AI models remained weak relative to the competitive set.

No experimentation reserve. Four of the five underperformers had no experimentation budget. Every dollar was committed to known quarterly priorities. When emerging surfaces opened (Perplexity early-2025, GitHub knowledge bases for technical brands, video citation surfaces), these companies could not move quickly because they had no unallocated budget.

Content volume without editorial standards. All five underperformers were producing high content volume — above the benchmark median in pieces published per quarter — but had no senior editor or editorial standard ensuring the content was citation-quality. The result was high cost per citation and low citation rate on new content.

No measurement of citation share movement. Three of the five did not have citation share measurement instrumented at all. They were producing AEO work without tracking whether it was producing citation share growth, which made it impossible to reallocate budget toward what was working.

The HBR analysis of marketing measurement maturity frames this exact pattern: marketing programs that ship without measurement infrastructure systematically underperform because the feedback loop required to learn from the spend does not exist.

How to Stage the Allocation Through the Year

The benchmark mix is annual, but the actual cadence of spending within the year shifts by quarter in predictable ways. The pattern across the 20 companies looks like this:

Quarter 1 is heaviest on tooling, baseline measurement, and the wikis and entity infrastructure work that requires elapsed time to compound. Quarter 1 typically runs 35 to 40 percent of annual tooling spend, 25 percent of annual wikis spend, and front-loaded content production to seed the year's editorial calendar.

Quarter 2 is the heaviest content production quarter and the heaviest PR retainer quarter. Editorial teams are at full capacity, PR firms are running active campaigns, and the awards submission cycle peaks in late Q2 for fall award programs.

Quarter 3 shifts toward measurement, mid-year audit, and experimentation. The pattern that emerged from the better-performing programs is that Q3 is when the team takes the data from H1, identifies what is and is not working, and reallocates the experimentation budget toward the bets that warrant deeper investment.

Quarter 4 is heavy on year-end PR and awards push, planning work for the following year, and the editorial calendar reset that positions the program for Q1.

The MarketingProfs 2026 B2B Content Marketing Benchmarks report a similar quarterly cadence pattern across B2B content programs in general, with the AEO-specific overlay being that Q1 and Q3 carry more measurement and infrastructure investment than traditional content programs do.

The Honest Limits of the Benchmark

The 20-company sample is not representative of the full marketing landscape. The companies that participated in the benchmark all had at least 18 months of AEO program history, had at least one dedicated AEO operator on staff or under retainer, and were tracking citation share with at least one purpose-built tool. The framework should be treated as a starting point for serious operators, not a universal allocation rule.

Categories that are heavily regulated (financial services, healthcare, legal) tend to overallocate to wikis, compliance review, and editorial oversight relative to the median. Consumer brands tend to overallocate to PR and underallocate to team because the brand context work that drives consumer AEO performance is closer to traditional PR than to B2B editorial production. Open source software brands often spend less on PR and more on community and developer relations work that does not fit cleanly into the framework above.

The other honest limit is that the framework is calibrated for English-language AI assistants and primarily North American and European markets. Non-English AEO programs face different citation patterns, different competitive density, and often different tooling availability that shifts the optimal allocation meaningfully.

Takeaway: The 2026 AEO budget allocation that produced citation share growth across the 20 companies we benchmarked is 45 percent content, 20 percent team, 15 percent PR and awards, 8 percent wikis, 7 percent tooling, and 5 percent experimentation. The numbers are defensible against finance scrutiny because each is anchored in a specific operational requirement: content because AI models cite content, team because editorial judgment compounds, PR because third-party citations carry disproportionate weight, wikis because entity infrastructure is foundational, tooling because measurement enables reallocation, and experimentation because AI surfaces evolve faster than planning cycles. Stage matters — early-stage companies should overweight content and PR, enterprise companies should overweight team and tooling — but the mid-range allocation is the median that worked across the sample. The CFO-defensible math runs through pipeline at risk and cost per citation, and the operators who treat the budget as a portfolio allocation rather than a content marketing line are the ones compounding their lead.