Period-over-Period Measures in Looker
Feb 17, 2026
Data
Tracking change over time—month-over-month, year-over-year—is essential for monitoring performance. Until recently, doing this in Looker meant relying on table calculations, which could be:
Hard to read and maintain
Prone to human error
Limited to visualization-only logic
Looker's new period-over-period measures help bring time-based comparisons into LookML itself - making it easier to build reliable, reusable metrics. (Looker docs: period-over-period)
🧱 Before: Table Calculations
Previously, getting last year’s value for a metric required a workaround like the following table calculation:
Limitations we encountered with this approach:
⚠️ Only works in visualizations—can’t be filtered or reused elsewhere
🧩 Logic cannot get duplicated across dashboards
🔍 Harder to QA and understand over time
📉 Fragile if the time grain or sort order changes
✅ Now: LookML Period-over-Period Measures
With the period_over_period measure type, we can now move this logic into our LookML layer.
Example: Last Year’s Value
Example: Year-over-Year % Change
Example: Looker Explore
Example: Looker Visualisation
🏢 How We Use It
At Super, we’ve started using these measures to track key metrics like:
📦 Sales – Year-over-year trends
🧲 Customer and merchant retention – Month-over-month comparisons
🚀 Feature impact – Week-over-week shifts after a product launch
Because these are built into our LookML layer, they can be reused across different dashboards, which helps ensure consistency and saves time.
🔍 Why This Helps
✅ Cleaner and centralized logic – Easier to review and test
📊 Works across dashboards – Filterable, sortable, and visualization-friendly
🔧 Adaptable – Supports different time periods (week, month, quarter, year)
💡 Final Thoughts
This feature might seem small, but it's been a noticeable improvement for our data team. It reduces repetitive work, helps standardise time-based comparisons, and makes dashboards easier to maintain and trust.
Clément Raul, Data
June 30, 2025
Super Thinking
