Hey everyone! I've been through a few migrations now (Tableau Server to Cloud, a PBI tenant consolidation, and even a Looker instance merge), and moving historical data is always the trickiest part. It's never just a simple "lift and shift" if you want to keep your team's trust in the data.
The biggest lesson I learned is to **treat historical data and new data as two separate migrations.** Your cutover for new reports can be quick, but history needs a careful, phased approach.
Here's what worked well for us:
* **Define "History":** First, agree on what timeframe matters. Do you need 3 years of daily snapshots, or just 13 months of month-end? This decision drastically reduces the effort.
* **Extract and Stage:** Pull the historical data from the old system into a neutral staging area (like a data warehouse or even CSVs). Clean it there, aligning dimensions and metrics to the new tool's data model.
* **Parallel Load:** Don't turn off the old system! Load the staged history into the new tool and run both systems in parallel for a reporting cycle (e.g., a month). This lets you validate numbers side-by-side and build confidence.
* **Communicate the "History Cutoff":** Be super clear with your users. "Reports in NewTool show data from Jan 2023. For anything prior, please refer to OldTool until [date]."
Our last major migration took about 6 weeks total, but only 2 of those were for the history load and validation. The key was taking that time to get it right—nothing erodes trust faster than mismatched numbers.
What specific tools are you folks migrating between? The approach can vary a bit depending on if you're moving between cloud platforms or bringing an on-prem history to the cloud.
—Mike
Measure twice, slice once.