Why Only 1 in 4 Employees Uses Your BI Tools Frequently (And What to Do About It)
By Izzy Aly, Co-Founder, Systems and Architecture
Only 25% of employees use BI software in their daily work. The fix isn't another dashboard: it's an agent that pushes the answer into Slack or email, with a human approving the action.
What if I told you that for every four people on your team, three of them don't use the expensive business intelligence dashboards you've paid for in their daily work? Would that shock you? It shouldn't. For years, industry surveys have shown the same thing. BARC's BI & Analytics Survey 26, built on feedback from more than 1,000 professionals, found that on average only 25% of employees use BI software in their daily work (BARC, November 2025). In BARC's earlier study with Eckerson Group, that share had barely moved in seven years of tracking (BARC, 2022).
You've spent a fortune on platforms. You've hired consultants. You've built beautiful dashboards that display every KPI imaginable. And still, most of your team doesn't live in them.
I see this all the time. Business leaders come to me frustrated, convinced their team is the problem. They think their people are "data-averse" or just not trying hard enough. But the problem isn't your team.
The problem is that the dashboard itself was never the complete solution.
Your Dashboards Weren't Built for Quick Decisions
For years, we've been told that dashboards are the key to being "data-driven." Gather all the numbers in one place, and your team will make smarter, faster decisions.
Right?
Except it rarely works that way. Traditional business intelligence operates on a pull-based model.
This means your team has to:
- Remember to stop what they're doing.
- Log in to a separate BI portal.
- Find the right dashboard.
- Try to figure out what a chart is actually telling them.
- Translate that "insight" into a potential action.
- And plan and execute the actions they determined based on the insights.
You're asking them to be the detective, the analyst, and the strategist all at once. And you're making them do it in a tool that's completely disconnected from where they actually work.
It's like giving someone a library and expecting them to write a report on a deadline. The information is in there, somewhere, but it takes a ton of work to find what's relevant and make it worthwhile.
No wonder they stick with their spreadsheets.
Enter "Agentic Analytics" (It's Not as Scary as It Sounds)
So if the pull model is broken, what's the fix?
It's a shift from a reactive "pull" model to a proactive push model. We call this agentic analytics.
Forget the jargon for a second. Imagine having an AI analyst working for you 24/7. This agent doesn't just build charts; it helps complete the decision loop:
- It observes: It constantly monitors your key metrics.
- It explains: When something changes, it determines the reason and tells you in plain English.
- It recommends action: Suggesting specific, concrete next steps.
- It audits: It keeps a record of what happened, creating a trail of accountability.
Instead of you hunting for insights, the insights come hunting for you, delivered as timely alerts and digests right inside Slack, Teams, or your email.
That's the entire game.
Let's Be Honest: Where the "Pull Model" Really Fails
You can feel the difference, can't you? Let's break down exactly why this new approach is gaining ground where traditional dashboards hit a wall.
Problem #1: The Adoption Ceiling
We have already covered the 25% daily-use rate. It's a disaster. People don't check dashboards regularly because they're generic, complex, and often live outside their daily workflow. An agentic system, on the other hand, sends personalised, relevant alerts to the tools you already use.
A sales leader gets a Slack message about a pipeline risk. The marketing manager receives an email digest about campaign performance.
It's specific. It's timely. And it's frictionless.
Problem #2: The Decision Gap
Even with a live data connection, dashboards create a delay. The data might be fresh, but the insight is only as timely as the next person who reviews it. This gap between an event happening and a human noticing it can be costly.
Agentic systems close that gap. By pushing an alert the moment a threshold is crossed, they bring the insight directly to the decision-maker. It's the difference between installing a smoke detector and just hoping someone smells smoke in time.
Problem #3: They Show the "What" but Rarely the "Why"
A dashboard might display a large, alarming red arrow pointing downward. Great. Now what? You and your team have to drop everything to investigate. Was it a marketing campaign? A pricing change? A technical bug?
An agentic system conducts the initial root-cause analysis for you.
It doesn't just show you the red arrow. It tells you, "Hey, conversion rates dropped 5%, mainly in the Northeast, and it looks like it may be correlated with that recent pricing update."
One is a clue. The other is a starting point for an answer.
An Illustrative Example
To see how this works in practice, consider a common scenario for an e-commerce company.
An agentic system detects that the checkout conversion rate has dropped from a stable 2.3% to 1.8%. That might not sound like much, but it could translate into thousands in lost sales per hour.
A traditional dashboard might have eventually shown this, but someone would have had to notice the dip among dozens of other metrics.
The agentic system, however, doesn't wait. It could immediately send an alert to the e-commerce team in Slack. But it wouldn't stop there. It would also identify the likely culprit: a sudden increase in page load times on the checkout page, concentrated in specific geographic regions.
Within hours, the team could investigate the change that caused the slowdown and restore conversions. They didn't have to assemble a war room or spend days digging through logs.
The system observed, explained, and recommended where to look. That's the difference between being data-driven and being overwhelmed by data.
A Worked Example: A HubSpot-to-Slack Digest Agent
Take a made-up sales team at Acme Corp: one HubSpot pipeline, a dozen reps.
What it reads. Every weekday at 7:00, the agent queries HubSpot with read-only access. It asks four questions: which open deals sit past their close date, which have had no logged activity for 14 days, which changed amount since yesterday, and which moved stage.
What it posts, and when. At 8:00 it posts one digest to the sales leads' channel in Slack. The top is a short summary: how many deals slipped, how many went quiet, and their combined value. Below it, one line per flagged deal, with the owner, the reason it was flagged and a link to the HubSpot record. If a large deal slips during the day, it doesn't wait for the next digest: it sends that one alert straight away.
Where a human reviews it. The agent can't change HubSpot on its own. Each flagged deal comes with a suggested next step, such as "create a follow-up task for the owner" or "move the close date to next month". The sales lead approves or dismisses each one in Slack. Only an approved step goes back to HubSpot, written by a separate step that runs after the approval, and the log records who approved it. For the first two weeks, the RevOps lead checks each digest against HubSpot to catch a wrong filter before the team learns to trust it.
That's the observe, explain, recommend and audit loop from earlier, running on two tools the team already has open.
"Fine, But This Sounds Expensive and Complicated."
I get it. You hear "AI" and "autonomous systems," and your mind immediately goes to six-figure contracts, year-long implementation projects, and a whole lot of risk.
That was the old way.
- Traditional BI often required massive upfront investments in data warehouses, complex ETL pipelines (the plumbing that moves your data), and teams of specialised developers.
- Agentic Analytics platforms are typically designed for cloud deployment. They often use a SaaS, pay-as-you-go model. They can connect directly to your live data sources, reducing complexity and shortening time-to-value.
You can start small with a single, high-impact use case. Prove the ROI in 90 days, then expand. The days of betting the farm on a monolithic BI project are over. (Or at least, they should be.)
Your Data Isn't the Problem. Your Tools Are Incomplete.
For years, we've been trying to solve a human problem with a technology solution. We tried to force everyone to become an analyst.
It didn't work.
The future isn't about improving your team's use of dashboards. It's about building systems that bring insights to them. It's about delivering answers, not just data.
So, the next time you look at your beautiful, expensive, and mostly quiet BI portal, don't blame your team.
Ask yourself: are your tools creating more work, or are they delivering real answers? Are you asking your people to find insights, or are the insights finding them?
- Topics
- Share