Sustainability

Pivot Table Hierarchy Filters - Drill down into your data like never before

Vanesa García avatar Vanesa García · · 2 min read
Pivot Table Hierarchy Filters - Drill down into your data like never before

You can now filter pivot table data by any hierarchy value, directly from the dashboard, no reconfiguration needed.

How does this improves your life?

If you want track very specific data, now you just need to configure the filter once and keep an eye on it!

Examples:

  • Dangerous waste from 1 specific facility
  • Energy consumed from GdO or Solar Panels
  • Scope 1+2 Carbon footprint emissions from 5 specific facilities

What changed?

  • Inline filter button on every pivot widget: click the filter icon to narrow data by facility, scope, country, fuel type, or any other hierarchy in your table
  • Filter section in the config panel, set default filters when configuring a widget so it always opens with the right view
  • Active filter badge, a count badge on the filter icon shows how many hierarchies have partial selections, so you always know when filters are active
  • Smart filter persistence, filters are saved with the widget configuration and survive page reloads
  • Automatic cleanup, removing a hierarchy from rows/columns automatically cleans up any filters that referenced it

For current clients

This is already live, no action needed. Open any pivot table widget on your dashboard and look for the new filter icon in the top-right toolbar. Click it, select the values you want to see, and the table updates instantly.

For everyone else

Want to see how Dcycle’s advanced pivot tables can help you analyze your sustainability data? Reach out to your Customer Success contact and we’ll walk you through it.

Product UpdateVisualization
30-minute demo

See it with your own data.

Thirty minutes with our team. We show you how your data would come in, and what you would get out.

See Dcycle in action
  • No commitment
  • With your data
  • In your language

What your session looks like

  1. 01 You tell us what you’re being asked forCustomers, auditors or banks: what data and by when. 0–5 min
  2. 02 We show how your data comes inInvoices, ERP, suppliers: on an example close to yours. 5–20 min
  3. 03 You leave with a planWhat to automate first and how much time you save. 20–30 min
With someone from the team who then supports you