How AI Is Changing Who Gets to Do Business Analysis
For most of the history of business analytics, analysis was a specialist’s job. You needed to know how to query a database, build a report, or at minimum know which spreadsheet to open and how to read it. That created a natural ceiling on how many people in a company could actually participate in understanding how the business was doing.
That ceiling is coming down. Not because AI has made analysis trivially easy, but because it has removed enough of the technical friction that the people closest to the work — the managers, the team leads, the individual contributors — can now ask questions and get real answers without routing every request through a data analyst or an ops lead.
The implications for how growing companies make decisions are significant, and most organizations haven’t fully worked through what this means for them yet.
The Old Model and Its Limits
In most companies between 50 and 250 employees, business analysis runs through a small number of people. Sometimes it’s the founder, who built the business and understands it better than anyone. Sometimes it’s an operations lead or a financial analyst who became the de facto company analyst over time. In either case, the pattern is the same: analysis is scarce, it’s rationed, and it happens on a schedule that has more to do with when the analyst has bandwidth than when the business actually needs the insight.
Pete Caputa, CEO of Databox, describes what this looks like from the CEO’s seat. He used to run a monthly reporting process where managers reported up through directors and VPs to him, and he’d sit with all of that information and identify what needed attention. It was a significant time investment for everyone involved, it happened once a month, and by the time the conversation happened, some of what was being discussed was already a few weeks stale.
That model made sense when it was the only option. It no longer is.
What AI Actually Changes
The change AI enables isn’t that it does analysis better than a skilled analyst. It’s that it lowers the floor on who can do analysis well enough to make better decisions.
A product manager who wants to understand why a feature’s adoption rate dropped last week doesn’t need to file a request and wait. A marketing manager who wants to know whether a specific campaign drove meaningful traffic doesn’t need to ask someone to pull the numbers. A sales manager who wants to compare activity levels across their team on a Tuesday afternoon can get that answer without involving anyone else.
Caputa’s own workflow illustrates what this looks like at the CEO level. He now logs into Claude, which is connected to Databox’s MCP server, with skills defined that know how his business is built. He can pull real-time analysis across HubSpot, SEMrush, YouTube, LinkedIn, ad platforms, Mixpanel, and his CRM. He can also connect to Asana and Slack to pull context that might explain what the numbers are showing. What used to require a monthly cycle of upward reporting now happens on demand, at whatever level of detail the question requires.
The broader shift, though, isn’t just that the CEO can move faster. It’s that the same capability can be extended down through the org. When anyone can run meaningful analysis, the frequency of data-informed decisions goes up across every level of the company.
Two Things That Change When Analysis Is Distributed
The first thing that changes is frequency. When analysis requires a specialist, it happens on the specialist’s schedule. Once a month, once a quarter, when someone asks loudly enough. When anyone can run analysis, it happens when the question comes up, which is continuously. Problems get spotted earlier. Decisions get made with more current information. The lag between something happening in the business and someone acting on it shrinks.
The second thing that changes is the nature of the questions being asked. When analysis is scarce, people ask the questions they know they need to ask. The standard metrics, the expected reviews, the KPIs that are already on the dashboard. When analysis is abundant, people start asking the questions they’ve always been curious about but never had the bandwidth to pursue. Why does this customer segment behave differently from that one? What happened the last time we ran this type of campaign? Is there a correlation between these two metrics that we’ve never looked at?
Those exploratory questions are where a lot of the real insight lives. And they tend to only get asked when there’s no cost to asking them.
What Has to Be True for This to Work
Distributing analysis across an org requires more than access to an AI tool. A few things need to be in place before the capability is actually useful.
The data has to be clean and connected. If your sales data lives in one system, your marketing data in another, and your financial data in a spreadsheet that someone updates manually, the AI has nothing coherent to work with. The foundation is integrated, well-captured data that reflects what’s actually happening in the business.
The metrics have to be defined correctly. AI is good at sequencing analytical steps and explaining what data means in plain language. It is not reliable at calculating metrics from raw data on its own. Engagement rates, conversion rates, pipeline velocity — these require specific formulas, and an AI working from raw tables will sometimes get the math wrong in ways that aren’t obvious. The solution is a semantic layer, a system that sits between the data and the AI and handles the metric calculations correctly. Without it, you can get analysis that looks authoritative and leads you in the wrong direction.
And the culture has to be ready for it. Distributing analytical capability means distributing visibility into how the business is performing. That’s a real shift for organizations where data has historically been tightly held. It requires trusting that your team will use that visibility to make better decisions, not to second-guess everything or draw conclusions from incomplete context.
Where Most Companies Are Right Now
Caputa is direct about the current state: some organizations are seeing massive uptake of AI-powered analysis, and others aren’t sure how to approach it yet. The distribution is wide.
But the direction is clear. The companies at the leading edge are already changing two things about how they work: the frequency of analysis and the number of people participating in it. Those two changes, compounded over time, produce organizations that are faster, better informed, and more capable of self-correction at every level.
The companies that get there first won’t just have better dashboards. They’ll have built a fundamentally different kind of operating capability, one where insight isn’t rationed and the distance between a question and an answer is measured in seconds rather than weeks.
Listen to the full podcast episode here.