DASHBOARDS & GRIDS
Drilldowns and click to focus
A chart that answers "which region?" should let you click the region and ask the next question. Drilldown is how a dashboard holds a conversation.
- The chart shows regionsOrders by region: North leads at 395.
- Click NorthThe value becomes a filter on the whole dashboard.region = North
- Every widget refocusesKPIs, charts and grids now answer for North only.395Orders · NorthClothing212Footwear118Accessories65
The question after the chart
A chart of orders by region answers one question and raises the next. North leads at 395, ahead of West at 247, East at 168 and South at 142 - but which category? Which stores? Is it new customers or repeat orders? On a static report those questions start a new report. On a dashboard with drilldowns, they are a click. The value you clicked becomes a filter, and the whole dashboard answers the follow-up as if it had been built for it.
What a drilldown is
A drilldown is a navigation from a summary to the detail beneath it: region to store, category to product, month to week. Traditionally each level was its own report, linked by hand. On a dashboard over an indexed copy, a drilldown is just a filter plus a dimension: filter to North, break down by store, and the index returns the new counts beside the records in one request. There is nothing to pre-build, because every level is the same request with one more filter.
The levels do not have to be a fixed hierarchy. Region to store is natural, but so is region to category, or status to customer, or month to product. Because each step is a filter rather than a pre-built report, the path can go in whichever direction the question goes, and two people can drill the same chart two different ways and both get exact answers.
Click to focus
Click to focus is the drilldown without the navigation. Select a value directly from a chart - a bar, a slice, a row - and the dashboard refocuses around that selection: the KPI tiles now show North's 395 orders, the category breakdown shows Clothing 212 · Footwear 118 · Accessories 65, the grid lists North's orders. You have not left the dashboard; you have narrowed it. Clear the selection and everything widens again. It is the fastest way to answer "and within that?" because the question is asked by pointing.
- Clicka bar, a slice, a row
- Filterregion = North
- One requestfacets on the filtered set
- Every widgetrefreshed together
Why it is instant
Every widget on the dashboard is a facet over the same collection. When the filter changes, the dashboard sends one request with the new filter, and the index returns records and all the facets - counts per category, totals per store, the KPI values - computed on the filtered set, together. Nothing is re-queried widget by widget and nothing is pre-aggregated for a level nobody chose. That is also why the numbers agree: the 395 in the tile, the three categories that sum to it and the rows in the grid came from the same pass over the same records. What is faceted search? explains the mechanism.
Compare the older approach: a reporting tool with a pre-aggregated cube, where each drill path had to be anticipated and built, and a path nobody built was a change request. Facets over an indexed copy do not need the paths in advance. The dimension is a field; the filter is a value; the counts are computed when asked. Adding a new drill direction is adding a field to the collection, not rebuilding a cube.
It also means the drilldown is always consistent with the grid beneath it. Click North, and the orders grid lists exactly the records the tile counted, because they are the same request. A user who distrusts a number can open the rows behind it in one more click, which is the fastest way to build trust in a dashboard.
Designing for it
Drilldowns work when the dashboard's dimensions are the ones people actually think in: region, store, category, status, customer. Put those on the charts and in the filters, and a click on any of them is a meaningful question. Keep the KPIs at the top so the refocus is visible at a glance - the tile changing from 952 to 395 is the confirmation that the filter took. And give every filter a visible chip, so that a colleague looking over your shoulder can see what the dashboard is currently answering.
Two small conventions keep drilldowns honest. First, the refocus should be reversible in one click, so that exploring is cheap and nobody is afraid to click. Second, the selection should travel with a saved view when one is saved, so that Sales dashboard · North is a focus that was kept, not a copy that was made. Both are about making the click a question rather than a commitment.
Drilldowns also change what a dashboard needs to contain. When every chart can be narrowed by every other, a dashboard with five well-chosen charts answers more questions than one with twenty, because the questions are asked by combining, not by scrolling. The design work moves from adding widgets to choosing dimensions.
When the click runs out
A drilldown follows dimensions that exist on the dashboard. The moment the question needs a dimension nobody put there - "by hour of day" on a dashboard that has no time-of-day field - the click runs out, and that is the point to ask in chat, where a sentence can introduce the dimension. Dashboards vs chat is about that handoff. Most days, though, the click is enough, which is why it is the interaction people use most.