The Last Mile: Why Data Is Still Out of Reach for Decision-Makers

Last week I had coffee with the sales director of a retail chain. Partway through, he told me about something that had happened the day before.

He was in the company parking lot at 7 p.m., about to head home, and he needed a number: net margin by sales channel for the quarter. There was a board meeting the next day, and the CEO was going to ask.

He had messaged the BI analyst at 2 p.m. No reply, because the analyst was busy wrapping up a report for finance. He tried Power BI on his phone and couldn't find the right dashboard. He found a similar one, but the filters didn't match. He gave up.

In the end, he did what every sales director in Brazil does when they need a number and can't get it: he opened a side spreadsheet he keeps himself, with data from two weeks earlier, and took that into the board meeting.

He laughed as he told the story, as if it were normal. And it is. I hear some version of it all the time.

Twenty Years, Same Scene

I've been watching this scene since 2006. In twenty years, the only thing that has changed is the tool: Crystal Reports gave way to Power BI, and the spreadsheet on the shared drive gave way to SharePoint. The pattern stays the same. The person who needs the data can't get to it on their own.

Only 21% of professionals feel confident in their data skills (Qlik & Accenture, 2020). But the root problem is architecture more than literacy. We built an entire data infrastructure (ERP, data warehouse, ETL, governance, BI) and then put a locked door on the last mile. The key to that door is SQL. Or DAX. Or "knowing your way around 47 tabs in Power BI." If you don't have the key, you file a ticket and wait.

Conversational BI Is Breaking Down That Door

When a sales director can pull out a phone at 7 p.m., ask in Portuguese, "What was net margin by sales channel in Q1?" and get an answer in 30 seconds, with the SQL right there to check, that's more than an incremental improvement. It's a different model.

The technology matters less than who gets a seat at the table. When anyone can ask the data directly, questions surface that would never have turned into a ticket:

• How many customers bought only once in the last 90 days?
• Which salesperson's average order value is dropping month over month?
• Which product slips into negative margin when it ships free?

Questions like these can uncover millions in opportunity, or in hidden losses.

The director in the parking lot didn't need another dashboard. He needed an interface that understood his question and knew where to find the answer.

Among Brazilian executives, 72% already use AI to be more productive (Bain & Company, 2024). Yet almost none of them use it to query their own company's database. That's the real gap, and it has less to do with technology than with imagination about how to apply it.

At your company, what happens when someone needs a number that isn't on any existing dashboard?