Executive Summary
The team keeps asking for CSV exports because they don't quite trust the dashboards. This takes up a lot of the data team's time.
The expensive BI tool isn't used much, and people make decisions using out-of-date or separate spreadsheets.
Start treating your data like a product. Agree on a single source of truth, put some simple governance in place, and build dashboards that are fast and genuinely useful. This helps to build back that trust.
You’ve seen the email subject line a dozen times: "Quick Q: Can I just get a CSV export for this?"
It often comes from the marketing team, usually after they’ve spent a while looking at a Looker dashboard where the numbers, to them, just don't seem to add up. This isn't just a small bother; it's a quiet signal that they don't fully trust the data setup. And that has a real cost.
Every manual export takes time away from your data team and suggests the self-serve analytics platform isn't quite working as hoped. It seems the business would rather work with separate, static spreadsheets than trust the expensive BI tool you've put in place.
Why people ask for a CSV export
A request for a CSV usually isn't just about getting the data. It's often about trust. When someone asks for a manual export, what they're often saying is one of these things:
It's quite common for BI tools not to be used as much as you'd hope. When people come across numbers that don't match up, or find the tools a bit tricky, they tend to go back to the spreadsheets they know. This isn't their fault; it's a sign the system isn't meeting their needs. The dashboard ends up being a sort of 'read-only' library of data people aren't sure about, rather than a useful tool for making decisions.
"Zek is an absolute pleasure to work alongside. He possesses a remarkable ability to patiently explain complex data processes in a way that is easily digestible, replicable, and memorable." — Chandni, Global Head of Member Happiness (LinkedIn)
The underlying reasons for the lack of trust
In my experience, this pattern often shows up in direct-to-consumer companies I work with, usually around the Series B or C stage. The data setup that worked well for their early growth starts to show some strain as things get more complex. You start to see data stored in separate places, like the marketing platform, the CRM, and the e-commerce system, which makes it hard to get a complete picture of what customers are doing.
The problem usually isn't with the BI tool itself. It's more often about the data that's going into it. A few common issues I see are:
When people lose faith in the data, they naturally stop using the platform. It really is that simple. This lack of trust can be a real hurdle to getting the whole company to use data more effectively.
"Zek was invaluable during his time at Beauty Pie. He contributed so much expertise on data visualisation and self-serve strategy. He really embedded himself within the team and the wider company which resulted in efficient and exponential improvement of our self-serve platform." — Grace, Senior Analytics Engineer (LinkedIn)
How to start rebuilding trust in your data
You can't fix a problem with the foundations by just giving it a new coat of paint. In the same way, more training sessions on Looker probably won't solve the underlying trust issue. From what I've seen, the best way to win people back is to start treating your data like a product and fix how it gets from the source to the dashboard.
1. Agree on a single source of truth This is probably the most important step. It involves working with each department to define important metrics, like 'customer acquisition cost', just once. These definitions should then be built into your data transformation layer (using a tool like dbt, for example) so they flow consistently into your BI tool. When everyone is working from the same definition of a 'conversion' or an 'active user', the arguments over whose numbers are 'right' tend to go away. This consistency is what starts to build confidence.
2. Focus on data governance and quality It's a good idea to put a simple data governance plan in place. This doesn't have to be complicated. It just means having clear owners for different sets of data, running some automated checks for quality, and being open about how it all works. Your team should be able to feel confident that the data is accurate and reliable before it even gets to a dashboard.
"Over the course of a year at Beauty Pie, Zek has become a fantastic part of the data team. He is incredibly hard working, committed and really cares about delivering what's needed." — Sara, VP Data AI (LinkedIn)
3. Make sure dashboards are fast and easy to use A dashboard that takes ages to load will, quite reasonably, be abandoned. It's worth spending time optimising your data models for speed. And just as important, it helps to build with the person using the dashboard in mind. A dashboard for the marketing team should answer their specific questions, not try to show every metric possible. A good BI system gives people relevant information they can actually do something with.
"We put a number of vendors to a test and NorthStar Analytics clearly stood out from the crowd. Top marks for quality, speed, and accuracy... NorthStar Analytics was great at knowing where the issues were, resolving them quickly, and helping us move from basic dashboarding to more advanced analytics for our customers." — Jim, CEO (Clutch)
By fixing these foundational issues, you remove the reasons your team went back to using CSVs in the first place. What usually happens next is that the data team stops being a reactive helpdesk for data exports. Instead, it can become a proper partner to the rest of the business, helping everyone use self-service analytics that they can finally trust. The requests for CSVs will stop, not because of a new rule, but simply because they aren't needed anymore.