Large US companies run an average of 183 artificial intelligence tools and platforms, yet only 15% of their AI investments operate as a unified system, according to a white paper from WSJ Intelligence and Code and Theory. The study, publicised on October 7, 2026, surveyed 801 chief executives, technology chiefs and marketing chiefs at companies with at least $500 million in annual revenue, and found that one in four of those companies admit that no one is accountable for coordinating AI across the business.

In Short

Big American companies have bought a huge number of AI tools, but most of those tools do not talk to each other, and in a quarter of companies nobody is clearly in charge of making them work together. This matters for marketers because the same disconnection shows up where customers see the brand, including in what AI shopping and search assistants say about it. The study, paid for by an agency that sells this kind of fix, says companies need one owner, fewer overlapping tools and rules for AI agents, though its numbers are executives' own opinions rather than measured results.

The study and its date

The white paper is titled "The Orchestration Gap: Why AI Investment Isn't Driving Enterprise Performance." It runs to 23 pages and carries the line "This content was created by WSJ Intelligence, a unit of The Wall Street Journal Advertising Department." That distinction matters. WSJ Intelligence describes its own role as supporting "advertising sales client partners, product and integrated marketing teams, advertising clients and commercial partners." The research is a commercial collaboration with Code and Theory, not a piece of Wall Street Journal news reporting.

Three dates attach to the material. The survey was conducted in June 2026 and, according to the credits page, "the study was reported in July 2026." The PDF supplied to PPC Land is dated October 5, 2026, in both its file name and its document metadata. The public announcement, distributed as a press release from New York, went out on October 7, 2026, which this article treats as the announcement date.

The sample consists of 801 C-suite executives at US companies with at least $500 million in annual revenue. Titles were split evenly: 33% CEOs, 33% CTOs or CIOs, and 33% CMOs. The largest industry groups were technology and telecom at 23%, banking, finance and capital markets at 15%, manufacturing and construction at 13%, consumer packaged goods at 8%, and retail and wholesale trade at 8%.

Two subgroups recur throughout the findings. "Industry leaders" are the 174 respondents who rated their own organizations as industry-leading in both revenue growth and innovation. "AI proliferators" are the 350 respondents whose organizations had deployed at least 100 distinct AI tools and platforms. Both labels rest on self-assessment.

The methodology section contains a caveat that rarely survives into headlines. "Findings reflect respondents' reported perceptions and organizational practices," the credits page states. "They should not be interpreted as audited measures of business performance or as proof of causation."

Who is behind it

Code and Theory describes itself in the paper as "Stagwell's (STGW) digital transformation network," a group that also includes Create Group, Current, Instrument, Kettle, Left Field Labs and Truelogic. The agency says it operates with a "differentiated balance of 50% creatives and 50% engineers" and lists Amazon, IBM, JPMorganChase, Microsoft, NBC and the NFL among its clients.

The commercial context is explicit. The final content page of the PDF invites readers to scan a QR code "to request your complimentary assessment and see exactly where the orchestration gap is costing you." The paper diagnoses a problem that its co-author sells services to address. That does not make the survey numbers wrong, but it shapes which questions were asked and how the conclusions are framed.

Stagwell has been building out this kind of capability. The holding company's digital transformation unit grew 12% in results covered by PPC Land in August, while Stagwell's revenue overall rose 10%. At Cannes Lions in June, Code and Theory was named among Adobe's CX Enterprise partners, alongside Accenture Song, Omnicom and WPP, and said it was building a Content Operating System for Sports on that platform. Adobe's chief marketing officer, Lara Balazs, is one of two outside executives quoted in the white paper. The other is Jonathan Adashek of IBM, which Code and Theory lists as a client.

What "orchestration gap" means

The paper defines the term in its executive summary as "the distance between deploying AI and operating as a connected, agentic enterprise." In a letter that opens the report, Dan Gardner, co-founder and executive chairman of Code and Theory, frames it slightly differently: "the distance between what AI can do and what the organization is set up to deliver."

Gardner's letter argues that the obstacles are old ones that AI has made more visible. "These silos are not new. AI is forcing us to confront them," he wrote. His proposed remedy is organizational rather than technological: "buying more tools will not reveal that potential. Organizations first need to make what they already have work together."

The specific remedy the paper keeps returning to is an orchestration layer, a technology tier that "sits across the existing environment" and "connects what already exists, brings AI and agentic capabilities into existing workflows, and makes the underlying technology easier for people to use."

The numbers on tool sprawl

The headline figure is 183. That is the average number of distinct AI tools and platforms run by organizations in the sample, counting across all locations and business functions. Among industry leaders, the average rises to 190.

The distribution behind that average is wide. According to Figure 1 in the report, 20% of all respondents run between 1 and 25 tools, 37% run between 26 and 99, 31% run between 100 and 499, and 13% run 500 or more. Industry leaders skew heavier: 12% at 1 to 25, 43% at 26 to 99, 33% at 100 to 499, and 13% at 500 or more.

The paper rounds these into two shorthand claims - "Four in five large enterprises manage at least 26 AI tools, and 2 in 5 run more than 100 tools." The chart adds up to 44% at 100 tools or more, which also matches the 350 AI proliferators out of 801 respondents.

Volume has not produced integration. Only 15% of AI investments "currently operate as a unified system, with information and insights that can be used across business functions," according to the white paper. Eighty-five percent of organizations describe their AI stacks as disconnected or siloed. Two-thirds routinely discover redundant AI investments or overlapping pilots. And more than two-thirds have not issued a request for proposals for a centralized system spanning products, functions and people.

Agreement on the principle is near universal. Ninety-four percent of respondents agree that better coordination of people and tools will help grow the business, and 85% say AI must now deliver measurable return on investment. The gap between those two figures and the 15% unified figure is the paper's central argument.

Faster output, slower decisions

A further finding concerns pace. Three in five executives report that final decision-making and strategic execution are "stagnant, bottlenecked or slower amid systemic noise." Even among the self-described industry leaders, 52% say they are stuck in what the report calls "execution paralysis."

The paper's explanation is that individual productivity gains do not carry through the organization. "AI can make one department seem more productive while leaving the next team with more work to reconcile," it states. "Outputs still have to cross systems, approval chains and functional boundaries before they become decisions."

It also describes a supervision burden that grows with output. Work can "pass through dozens of AI-assisted decisions, iterations and handoffs until the output is no longer clearly connected to objectives that started it." People, the report argues, end up acting "not as creators, but as evaluators and curators." That observation tracks closely with research Typeface published in June, which found that 34% of senior marketers now need one to two months to launch a campaign, up from 5% in 2025, and that 92% say campaigns require 10 or more stakeholders.

Culture outranks budget

When respondents were asked what prevents cross-functional AI orchestration, money and skills came last. According to Figure 2, the barriers ranked as follows:

  • Corporate culture and mindset - 61%
  • Data fragmentation, including incompatible data models and no shared intelligence - 58%
  • Tech and infrastructure, meaning no central architecture to connect AI tools - 56%
  • Legacy workflows not designed for the speed of agentic AI - 49%
  • Leadership and governance alignment - 43%
  • Skillset and talent gaps - 38%
  • Budgetary constraints - 29%

The pattern holds across sectors. Between 82% and 87% of respondents in every major industry group describe their stacks as disconnected or siloed, and at least half in every group report stagnant or bottlenecked strategic execution. "No sector has broadly solved the connectivity, execution and coordination problems," the paper states.

The report's reading of these numbers is that "additional talent or funding alone will not close the orchestration gap." Respondents were asked to choose barriers rather than rank them by severity, so the percentages measure how often a barrier was cited, not how much each one costs.

Nobody agrees who is in charge

The most striking section for marketing leaders concerns ownership. Respondents were asked who in their organization is "primarily accountable for AI orchestration across functions." The answers diverged sharply by the job title of the person answering.

According to Figure 3, CEOs split their answers as follows: 42% named the CEO, 36% the CIO or CTO, 17% said there was no single clear owner, 4% named the COO and 1% the CMO. Technology chiefs saw it differently: 57% named the CIO or CTO, 8% the CEO, 28% said no single clear owner, 3% named the COO and 4% the CMO. Marketing chiefs gave 43% to the CIO or CTO, 20% to the CEO, 25% to no single owner, 5% to the COO and only 7% to the CMO.

Put plainly, 42% of CEOs think the job belongs to them, 57% of technology leaders think it belongs to them, and only 7% of CMOs place it with their own function. Across the sample, one in four companies say no one is accountable. "Shared responsibility without explicit decision rights quickly becomes unowned responsibility," the report states.

The CMO figures sit awkwardly with other research. An Acquia study published in September found that 57% of CIOs and CTOs say they make CMS and DXP decisions, and that only 41% of CMOs are confident their teams can govern AI. The two studies used different samples and questions, but both point to marketing leaders deferring platform authority to technology chiefs at the moment that AI is reshaping customer-facing work.

Balazs of Adobe, quoted in the paper, frames the shift in marketing terms. "Agentic AI allows brands to create and deliver compelling customer experiences more efficiently and effectively," she said. "Leaders are rethinking fragmented, siloed activities in favor of a connected system to drive greater agility, accountability and value creation across the enterprise." She also warned against scaling "without losing the alchemy and craft at the heart of marketing."

Four costs the report assigns to the gap

The paper names four "connected vulnerabilities" it attributes to poor orchestration: stalled execution, disconnected customer experiences, a risk it calls "algorithmic extinction," and ungoverned agents.

Stalled execution

The first is deceleration and wasted capital. The two-thirds of companies that routinely uncover redundant investments are the evidence. The paper argues that "consolidation alone is not orchestration" and that surviving tools need "secure connections that enforce access permissions so content, data and context can move across the stack."

Disconnected customer experiences

Seventy-three percent of respondents say their customer-facing channels do not operate from a fully unified data model and brand architecture. Figure 4 breaks this down: 22% of all respondents manage channels independently through separate teams, 51% are partially integrated, and 27% run a unified data model and brand architecture across all channels.

The industry leaders figures cut both ways. A slightly higher share, 31%, report full unification, but 26% also report channels managed independently, compared with 22% for the full sample. The chart caption describes industry leaders as "more likely to report functions are siloed."

Confidence runs well ahead of practice. Eighty-three percent of CEOs say they are confident of delivering a consistent brand experience, as do 71% of industry leaders. The paper notes the mismatch and adds that "confidence built on the old map may not hold" as customers move toward conversational interfaces.

Algorithmic extinction

This is the section closest to search and performance marketing. Only 17% of respondents report a formal approach to what the paper calls "Agentic Engine Optimization (AEO)," defined in the survey as "ensuring that a brand is represented accurately when AI systems make decisions for a business or consumer."

That definition departs from common industry usage. In search marketing, AEO usually stands for answer engine optimization, the practice of appearing in AI-generated answers. Microsoft Advertising's retailer guidance, for example, treated AEO alongside generative engine optimization as a matter of product feeds and data completeness. The white paper's version is broader and extends to AI agents that act on behalf of buyers, not only those that answer questions.

Low confidence in accurate brand representation was reported by 39% of CEOs, 52% of CTOs and CIOs, 40% of CMOs and 48% of industry leaders. Technology chiefs, who claim ownership of orchestration most often, are also the most pessimistic about how AI systems portray their companies.

The paper rejects a marketing-only response. "This isn't something a new marketing approach can solve," it states, arguing that accurate representation requires structured data, consistent brand information and governance over "how automated systems interpret and use enterprise assets."

Prior research suggests the problem is measurable. Semrush's AI Visibility Index, released in June, found that only 36 brands out of more than 1,200 stayed in the top 100 on every AI platform in every month studied, and 45% of marketers it surveyed said they could not properly measure their AI answer visibility. An Adobe survey of more than 500 marketers found that 98% lacked a clear, documented AI optimization roadmap.

Ungoverned agents

Only 18% of organizations say AI agents making decisions are subject to accountability and reporting structures equivalent to those applied to human leaders. Among AI proliferators, with 100 or more tools, the figure is 17%.

Figure 5 shows the rest of the distribution. Across all respondents, 34% say "somewhat," 30% have not yet established clear structures, and 18% have not deployed AI agents at scale. Among proliferators, 28% say "somewhat," 35% have no clear structures and 20% have not deployed agents at scale. Companies with the most tools are, if anything, slightly further behind on governance.

The report's proposed standard is specific. Each agent making significant decisions "needs a defined role, a named human accountable for its decisions, boundaries on what it can do, and an auditable record of its actions and the authority under which it acted." That shifts the governance question, in the paper's words, "from asking whether a tool is secure and functional to asking whether an agent is acting within its assigned authority."

That framing matters for advertisers handing campaign management to agentic AI systems inside ad platforms. A survey of developers and product leaders by Nylas in March found that only 4% let agents act without any human approval, which suggests human sign-off remains the default control in practice, even where formal accountability structures are missing.

The prescriptions

The paper closes with three recommendations, which double as a description of Code and Theory's services.

The first is to make orchestration "a consistent enterprise mandate," run as a standing function in the way companies run finance and marketing, with a defined leader, dedicated resources and a budget. "Orchestration is not a project with a start and an end, but a function," the paper states. It lists "issuance of a clear brief, budget allocation and RFP consideration" as part of that mandate.

The second is to "compress the organization" at four levels: roles, so that one person can do work that previously required several specialists; departments, so that sequential functions operate as one system; decision-making, pushed to front-line staff; and insights, connected directly to product, operations and supply chain. "The AI-native organization isn't just more connected. It's more compressed, with fewer handoffs, fewer layers and less distance between insight, decision and action," according to the report. The paper does not quantify what compression means for headcount.

The third is to connect data and systems beneath the work, so that "context travels with the work" and each step "should inherit the customer need, the business objective, the decisions already made and the permissions to act." The suggested starting point is a single workflow, measured on whether it "cuts delays, reduces rework or improves the customer outcome."

Adashek of IBM, the second outside executive quoted, puts the case against further purchasing directly. "When 85% of organizations say their AI technology is disconnected or siloed, the answer cannot be another tool," he said. "Leaders and teams set the strategy, and AI helps execute it."

Why this matters for marketers

For advertisers and agencies, three findings stand out. The first is the ownership split. With only 7% of CMOs placing AI orchestration in their own function and 57% of technology chiefs claiming it, decisions about customer data, brand representation and agent permissions are, on these numbers, increasingly made outside marketing.

The second is the brand-representation gap. The white paper ties disconnected customer data directly to how AI assistants describe a company, which links internal data architecture to search and commerce visibility in a way that marketing teams have been tracking through separate tools. Vendors have started packaging this; Typeface in September made available a system that turns brand guidelines into structured rules for AI agents, including inside Claude and ChatGPT.

The third is return on investment. Eighty-five percent of the executives surveyed say AI must now deliver measurable ROI, consistent with a TransUnion survey in August in which only 53% of senior marketers reported meaningful ROI from AI and 65% measured AI mainly through cost or time savings.

The limits of the evidence also deserve weight. All figures are self-reported by executives. The "industry leader" label is self-assigned. The study does not report a margin of error, does not link orchestration practices to financial outcomes, and was produced by an advertising department in partnership with an agency that sells the remedy it describes. The paper itself states that its findings are not "proof of causation." What it does document, with a sample of 801 senior executives, is that the people who run large US companies broadly agree their AI spending is fragmented, and disagree about whose job it is to fix it.

Timeline

Summary

Who: WSJ Intelligence, a unit of The Wall Street Journal's advertising department, and Code and Theory, Stagwell's digital transformation agency network. The survey covered 801 CEOs, CTOs or CIOs, and CMOs at US companies with at least $500 million in annual revenue. Outside executives quoted are Jonathan Adashek of IBM and Lara Balazs of Adobe.

What: A 23-page white paper finding that companies run an average of 183 AI tools, that only 15% of AI investments operate as a unified system, that 85% describe their AI stacks as siloed, that one in four companies say no one is accountable for AI orchestration, that only 17% have a formal approach to accurate brand representation by AI systems, and that only 18% govern AI agents as they govern human leaders.

When: The survey was conducted in June 2026 and reported in July 2026. The PDF is dated October 5, 2026, and the findings were publicised on October 7, 2026.

Where: The sample is limited to large US companies, across technology, finance, manufacturing, consumer packaged goods and retail.

Why: The study argues that AI spending has outpaced the structures needed to coordinate it, and that the result shows up in slower decisions, redundant investments, inconsistent customer experiences and ungoverned agents. For marketers, it points to CMOs ceding orchestration authority to technology chiefs while AI systems increasingly decide how brands are represented to customers. The figures are self-reported, and the study's co-author sells the remedy it recommends.