Your AI dashboards can probably tell you a lot: who has a license, who logged in, how many prompts they submitted, how many tokens they consumed, and what each tool costs. The catch? You probably have a different dashboard for every AI platform and even when you bring the data together, it only tells part of the story.
As organizations add ChatGPT, Microsoft Copilot, Gemini, Claude, and other AI tools, tech leaders have to piece together usage and spend across a growing collection of platforms. In One Model’s research, organizations used three to four different methods, on average, to measure AI impact and spending across their environments.
And even if you manage to pull all of that data together, you’re still left with a harder question: What are people actually doing with AI?
That’s where the bigger visibility gap starts. Two employees can use the same AI tool every day and look nearly identical on an adoption dashboard. One might use it to rewrite emails and summarize documents. Another might use it throughout a complex workflow to research a problem, analyze information, challenge assumptions, refine an approach, and produce a finished piece of work.
Same tool. Similar activity. Very different work happening underneath it. As organizations pour more money into AI, that distinction matters.
AI has moved well beyond the experimentation stage. In One Model’s research of 50 senior technology buyers, 56% said their organizations already spend more than $1 million annually on AI, and 94% expect that spending to increase over the next 12 months. Nearly a third are already running 11 or more AI tools.
The pressure to explain what all that investment is producing is nearly universal. Ninety-eight percent of respondents said they’ve already faced pressure to justify AI spending to the CFO, CEO, or board, or expect to soon.
The problem isn’t a lack of data. It’s that most organizations have plenty of data that answers the wrong questions. They can see total AI spend. They can usually see the cost by vendor. They have usage dashboards from individual AI providers. They may have internal BI dashboards, finance records, or spreadsheets pulling some of it together. They’ve built a pretty good picture of AI activity. What’s missing is the work behind it and what happens as a result.
Consider a software engineer, salesperson, finance analyst, and marketer. Should you expect them to use AI the same way? Of course not. Yet traditional adoption metrics flatten those differences. Logins, tokens, prompts, and active users tell you how much activity occurred. They can’t tell you whether AI helped complete meaningful work, whether the interaction required repeated rework, or whether a team has developed a repeatable way of working worth sharing.
That makes some of the most important questions about enterprise AI surprisingly difficult to answer. Which teams have embedded AI into meaningful workflows? Where are employees developing sophisticated AI skills? Which use cases are spreading organically? Where are employees struggling and likely to benefit from training? Are expensive AI tools being used for the work they’re best suited to perform? And ultimately, where does stronger AI use begin to show up in productivity and business outcomes? You can’t answer those questions by counting prompts. You need context.
This is the visibility gap One Model AI Impact is designed to close. AI Impact brings together usage data across your AI tools and connects it with workforce context so leaders can understand how AI is being used across roles, teams, functions, and the organization.
There’s another layer of context that can reveal something traditional AI dashboards can’t: what people are asking AI to help them do.
With prompt-level intelligence enabled, AI Impact can analyze patterns in AI interactions to provide deeper insight into how employees are putting AI to work. A prompt contains clues about the task itself: whether someone is researching, analyzing, coding, drafting, planning, or solving a problem.
But a prompt is only one part of the picture. A series of prompts can form a session around a single piece of work, giving organizations more context about what was attempted, whether something useful was produced, and where friction occurred.
Looking at those patterns can help an organization move from measuring AI activity to understanding the work behind it, how that work gets done, and where it succeeds or breaks down.
That distinction is critical, but it’s also sensitive. Prompt-level intelligence raises legitimate questions about privacy and employee trust, which is why organizations need control over whether they use that level of analysis. That’s why AI Impact gives customers the ability to turn prompt-level visibility on or off based on their own policies and priorities. This could be company wide or reflect specifically with role-based permissions. This level of visibility gives leaders enough context to make better decisions about their AI strategy.
For years, SaaS adoption followed a fairly simple logic: buy the software, drive logins, increase feature adoption, and reduce shelfware. AI changes that equation because AI tools aren’t used for one predefined workflow. The same general-purpose AI platform might help one employee draft a routine email and another solve a problem that previously took hours of specialized work.
That makes raw adoption a particularly weak proxy for value. And technology leaders already recognize the problem. In our research, productivity impact ranked as the biggest AI measurement challenge, followed by demonstrating AI ROI. Meanwhile, showing leadership the impact of AI on team and business outcomes was the highest-rated capability buyers wanted from an AI effectiveness solution. Only 20% of respondents said their organization has a formal AI ROI framework applied across tools.
That leaves a lot of organizations facing an uncomfortable conversation: We know AI usage is going up. We know spending is going up. But can we prove the organization is getting better because of it?
The next generation of AI measurement can’t stop at licenses, logins, tokens, and spend. Those metrics still matter, but they’re only part of the picture.
Leaders need to see where AI has become part of real work, how usage differs across the workforce, and where stronger practices are emerging. From there, they can begin to model the value of that work against the cost of AI activity and see where investment is creating value. That context can inform decisions about training, technology investments, governance, and where to scale successful AI practices across the business.
Because the question facing CIOs, CTOs, and AI leaders is changing. It’s no longer simply, “Are our people using AI?” It’s “What are they doing with it, and is it making us better?” AI Impact is built to help answer that question.