A single STAC search can return millions of matching items. The usual reflex is to paginate: pull footprints, render thumbnails, scroll, and hope you spot what matters. But there's a more useful first question. Not "which scenes matched?" but "what does the coverage actually look like?" Where is there data, over what time span, how densely, and, before you download a single pixel, what do the metadata already tell you?

STAC's aggregation extension makes this a first-class way to explore a catalog, and it changes the whole interaction. Instead of a list of scenes, you get a coverage map: a temporal histogram built from every match (via datetime_frequency), an H3 hex density layer showing where results concentrate (via centroid_geohex_grid_frequency), grid-cell frequency, and period dissolves (solar-day, monthly, annual) that collapse thousands of overlapping footprints into a clear picture of coverage over time. Actual footprints and imagery load on demand, once the distribution has told you where to look.

The same machinery turns metadata into an analysis surface. Aggregate on a property like cloud cover and patterns appear that are invisible scene-by-scene: which regions and seasons are usable, where an archive is persistently obscured, when your window of clear observations actually falls. That's metadata analysis, not scene browsing, and it's a different and often faster way to understand what a catalog holds.

We'll demonstrate with stui, an open-source, map-first STAC explorer built on MapLibre and designed aggregate-first from the ground up, being open-sourced for this talk. But the tool is only the vehicle. The real point is a shift in how you see a catalog: it is best understood as coverage, distribution, and metadata patterns, not as a pile of individual scenes, and the aggregation extension is an underused key to seeing it that way. Whether you run your own catalog or query someone else's, you'll leave with a new way to look at data you already have.