6% of Enterprises Are Making AI Work: Here's How

 Most enterprises have convinced themselves they are doing AI. They have pilots running. They have a vendor relationship. They have a slide deck with a roadmap on it. What the Reuters Insights and Scale AI Six Percent Report makes clear, after surveying 494 senior AI decision-makers across major enterprises in Q2 2026, is that running a pilot is not the same thing as making AI work. And the distance between those two states is growing faster than most organizations have admitted to themselves. 

Key Points

• The success rate is sobering: Of 494 large organizations surveyed by Reuters Insights and Scale AI in Q2 2026, only 32 — 6.5% — met all three criteria for genuine AI success: thoroughly integrated across most functions, more than half of pilots in production, and significantly ahead of their own targets.
• Speed is the most visible differentiator: Two-thirds of the Six Percent record a measurable business outcome within six months of starting a pilot. Fewer than one in three of the broader sample move that fast.
• The gap is not a technology gap: The Six Percent are not running different models or accessing tools others cannot reach. What separates them is a sequence of organizational decisions made before a line of code was written.
• Off-the-shelf tools dominate the failures: Standardized tools deployed with minimal customization are two and a half times more common among stalled deployments than successful ones.
• Data confidence is the sharpest dividing line: 81% of the Six Percent are extremely confident their data foundations can support AI. Only 34% of the broader sample say the same.
• The window is narrowing: The Six Percent are already deploying second-wave, sector-specific AI applications while the broader market is still building the foundations the leaders finished two years ago.

Who the Six Percent Are

The report's methodology is worth understanding before the findings. To qualify as part of the Six Percent, an organization had to meet three demanding criteria simultaneously: AI is thoroughly integrated across most business functions, more than half of their AI pilots have reached full production, and their AI programs are advancing significantly faster than their own internal targets. All three, at once. Thirty-two of 494 organizations cleared that bar.

The report is careful to note that these findings describe associations, not proven cause and effect. A cross-sectional survey can show what distinguishes the organizations getting AI to work, not prove that any single characteristic produced the result. The cohort is also small enough that each percentage point represents roughly one respondent. The findings are directional and substantial, not definitive.

What the data rules out is the easy explanation that the Six Percent are simply bigger or better-funded. When analyzed by revenue, sector, and region, the Six Percent look much like the full sample. Where they differ meaningfully is seniority: half of the Six Percent respondents are C-suite or board members, against a third of the full sample. The decisions that distinguish them were made at the top of the organization, before the technology was selected.

The Speed Advantage and What Produces It

The most visible gap between the Six Percent and the broader market is how quickly they move from pilot to measurable result. Among the Six Percent, 53% move pilots to full production within six months, nearly double the 30% of the broader cohort that manages the same pace. More striking: 22% of the Six Percent saw measurable business outcomes within three months of starting a pilot, compared to just 6% of the broader sample.

The report's framing of this is worth quoting directly: the Six Percent see business results from AI faster than most companies' budget cycles allow them to even check on progress. They have synchronized the pace of their AI work with the pace at which their boards review it.

What produces that speed is not better technology. It is what the Six Percent commit to before any return is expected. The most-cited resourcing investments ahead of deployment are training existing employees and change management, data preparation and labeling, and sustained executive sponsorship. The Six Percent report higher rates of all three than the broader sample, with senior leadership sponsorship showing the largest gap at 21 percentage points. The broader market is still retrofitting these commitments after pilots stall. The Six Percent front-loaded them before the first deployment began.

The acceleration question — what single change would most speed up the move from pilot to production — produces a revealing divergence. Across the full sample, governance frameworks, standardized processes, and employee training cluster at the top, each within half a point of the others. For the Six Percent, better tools and technology rises to 22%, more than double the broader sample's 11%. Executive leadership support, meanwhile, drops to zero for the Six Percent, against 9% for the full sample. They do not need to advocate for leadership buy-in because they already have it. The problem they are trying to solve is capability, not permission.

How They Build and Buy

Every enterprise AI program arrives eventually at the build-or-buy question. The Six Percent answer it differently in two specific ways.

First, they are far more likely to use a hybrid approach, combining internally built components with external platforms, rather than committing to a single procurement model. Hybrid builds account for 41% of the Six Percent's most successful deployments, nearly three times the 14% rate in the broader sample. The broader market's default is co-development with a single vendor, at 39%. The Six Percent co-develop less, not more. They build in-house where they have the capability, customize vendor products where an existing platform fits, and combine both where neither pure model works.

Second, when they do involve a vendor, they want either deep co-development with shared outcome ownership (47%) or a fully managed service the vendor runs end-to-end (22%). What they avoid is the middle ground, where the buyer sets the roadmap and the vendor configures. Implementation partnerships, the most common vendor relationship in the broader sample, are the least favored by the Six Percent.

The off-the-shelf finding is the most damning in the report for organizations that defaulted to standard tooling. Among successful deployments, just 6% were built off-the-shelf. Among organizations whose pilots stalled, off-the-shelf was the dominant approach at 15%. The report states the contrast clearly: off-the-shelf tools are conspicuously rare among the successes and conspicuously present among the stalls.

Data Is the Actual Differentiator

Asked what made their most successful AI deployment work, the Six Percent point first to data. High-quality proprietary data used for AI model training appears in 56% of the Six Percent's top-three success factors, compared to 48% across the broader sample. The data confidence gap is the single largest divide in the entire study: 81% of the Six Percent are extremely confident their data foundations can support AI, against 34% of the broader sample.

How the Six Percent source that data is also distinct. When they go external for data, they are more likely to buy a finished data product from a single vendor — 25% versus 10% of the broader sample — and significantly less likely to outsource labeling to third-party services — 9% versus 23%. They avoid assembling the data layer from piecemeal labeling contracts. They source it as an integrated product from a single strategic partner.

The sequencing implication the report draws from this is direct: the data layer is a decision to make early, before the model and the use case are locked, because it shapes everything that follows. Most organizations make it late, after the model is selected and the use case is committed, and pay for that sequence in integration failures and stalled deployments.

What the Six Percent Are Doing Next

The investment trajectories reveal how far ahead the leaders actually are. Among the broader sample, 43% expect AI investment to increase significantly over the next 12 months. Among the Six Percent, 88% expect a significant increase, with fewer than 10% expecting only a slight one.

More telling than the size of the investment is where it is going. The broader market's heaviest forward bets are AI-integrated productivity suites and workflow automation and agentic platforms, in a near-tie at 19% each. These are the foundational investments the Six Percent have largely already made. The Six Percent's heaviest bets are productivity suite extensions at 31% and specialist vertical AI tools at 22% — industry-specific applications in legal, finance, and healthcare that require the data foundations and deployment confidence the broader market is still building.

The report describes this plainly: the broader sample is making the agentic and workflow-automation platform investment the Six Percent have already made. The Six Percent are making second-wave bets on the vertical and domain-specific applications that require a functioning foundation to run. Neither group is wrong about its forward bet. They are simply at different points on the same path, and the distance between them is not shrinking.

The Eight Moves the Report Recommends

The report closes with eight specific actions drawn from the Six Percent's behaviors. They are not a guarantee, but they describe what the most successful enterprises have in common.

Secure executive ownership before funding the pilot. Pilots without a C-suite or board sponsor whose attention extends past procurement are more likely to stall. Plan for a multi-quarter pilot-to-production cycle, since programs expecting a return within a single quarter are systematically under-resourcing the build phase. Start the agentic shift now, because the Six Percent have already deployed agentic platforms and organizations treating this as a future problem risk arriving late.

Buy a partner, not just a product, and be skeptical of off-the-shelf defaults. Spend on people training at least as heavily as on software licenses, since a larger deal will not unblock an unprepared organization. Redesign the workflow before go-live, not after. Get data in order before scaling the model, despite the fact that only 17% of organizations name data infrastructure as a top investment priority for the next 12 months — the lowest of the ten use-case categories surveyed. Test, evaluate, and keep improving, measuring against business KPIs rather than adoption metrics or strategic narratives.

What This Means for Marketing and Growth Leaders

The Six Percent Report is not abstract. The organizations it profiles are hospitals, financial institutions, logistics networks, and professional services firms — the enterprises that shape how entire industries operate. When AI delivers value in those environments, the effects compound. When it stalls, the cost is measured in wasted budgets, regulatory exposure, eroded trust, and competitive disadvantage that accrues quietly until it becomes visible all at once.

For marketing and growth leaders specifically, the report's findings about what separates successful deployments from stalled ones apply directly. The AI tools being deployed in sales and marketing functions — content generation, campaign analytics, customer intelligence, lead scoring — are subject to the same dynamics. Off-the-shelf tools deployed without customization and without data foundations to support them will produce the same stall patterns documented in the report across every other function.

The organizations winning with AI in their marketing programs are doing the same things the Six Percent did in their enterprise programs: front-loading organizational decisions, treating data as the strategic asset it is, building hybrid capabilities rather than defaulting to a single-vendor relationship, and measuring against business outcomes rather than activity metrics.

Understanding what that looks like in practice for a marketing organization is exactly what our AI growth strategy work at Winsome Marketing is built around. The gap between the Six Percent and everyone else is closeable, but the report is explicit: the window is narrowing, and the decisions need to be made now. If you want to understand where your organization sits on that curve and what the path forward looks like, our team can help you figure that out.