Databricks Gave Excel Write Access, and the Top Comment Is "Who Still Uses Excel?"

Databricks released an Excel add-in.

Once installed, you can edit data directly in Excel and publish the changes to tables in Unity Catalog. Write permissions are managed by Unity Catalog, so not just anyone can put things into a table.

This feature fills a very specific gap. Business teams do most of their analysis in Excel: pulling data, pivoting, adding columns, making charts. But Excel and the data warehouse are disconnected. To get modified data back into the warehouse, you either export a CSV and have a data engineer upload it, or run it through ETL. Every extra person in between adds another layer of delay and chance for error. The add-in bridges that gap.

The official video is 4 minutes 58 seconds and demonstrates the flow from publishing in Excel back to Databricks. It's also available on YouTube: How to write data from Excel to Databricks

The comments are more interesting than the feature itself.

The first is from @toby or not toby, asking: Who is using excel?

This question is a bit of an outsider question. Anyone who works with data knows Excel is the most widely installed data analysis tool in the world. A large share of tables in data warehouses ultimately get exported into Excel for people to view. Databricks knows this too—that's why the add-in exists.

The second is from @KaD. He said he couldn't install the add-in by following the process on Databricks' page, but installing directly from the app store worked. He asked whether the long official process is really necessary.

This is a real-world pitfall. Installation documentation for enterprise software often describes the ideal path. In real environments, permissions, versions, and network policies create all kinds of blockers, and the shortest path often works better.

The third is from @YAIB and asks a real question: if two spreadsheets modify the same row at the same time, what happens before either one is published? He suggested adding a visible conflict check so spreadsheet users understand the boundaries of write-back.

This hits the nail on the head. Writing back isn't hard; concurrent writing back is hard. Excel's collaboration model and database concurrency control are two different things. One person changes cell B2 in table A, another changes the same row in table B, both click publish—who wins? Does the later publish overwrite the earlier one, or does it throw an error? The official post doesn't say.

When multiple people edit the same data, conflict handling determines whether this feature is usable in real teams. If it's just one person editing and publishing, it's an advanced upload button. To support multi-user collaboration, conflict prompts, version comparison, and rollback are all indispensable.

For years, data platforms have told business teams not to use Excel, and in the end they brought Excel in anyway. The first comment under the tweet is "Who still uses Excel?", but Databricks clearly knows the answer.