Two supporting surfaces. One holds what exists, the other teaches what any of it means.
Library
Every widget your organisation has built, in one place, whether or not it is currently on a board.
What it is for
Reusing rather than rebuilding. The most common waste in any analytics tool is somebody building a chart that already exists because they could not find it. The library is the answer, and it only works if people look before they build.
Putting an existing widget on a new board. Most boards should be assembled from things that exist rather than built from scratch.
Finding the orphans. Widgets on no board are either unfinished, superseded, or genuinely useful and forgotten. All three are worth knowing about.
Keeping it usable
Search before you build. Ten seconds, and it is the whole point of the surface.
Name widgets so they can be found. A library is only as good as its names, and "Chart 3" is a widget nobody will ever reuse.
Prune once a quarter. A library with three versions of the same chart is a library where nobody knows which is right, which is worse than having none of them.
Pruning
A library grows and never shrinks on its own.
Once a quarter, look for three things: two widgets with nearly the same name, widgets on no board, and anything whose scope refers to something that no longer exists. Delete rather than rename; a widget kept "just in case" is one somebody will eventually put on a board.
Explore
The gallery: board templates, report types, and the vocabulary the rest of Analytics assumes you already have.
It is deliberately curated
This is the one surface here that is not your data, and that is intentional. It is product education, so it shows worked examples rather than an empty state waiting for you to have built something.
Everywhere else in the platform an empty surface is honest. Here it would defeat the purpose: a person who does not yet know what a measure is cannot be taught by a blank page.
What it teaches
Board types, and what each is shaped for.
Report types.
Measures, cohorts and scope, which are the four words the rest of the workbench uses constantly and never defines in place.
Templates, which are a fast start rather than a finished answer.
Finding something in the library
Two habits make a library searchable, and both are about the moment of saving rather than the moment of looking.
Put the subject first in the name. "Pipeline by owner" sorts and searches better than "By owner, pipeline".
Say the scope. Two widgets differing only by period need names that say which is which, or somebody will put the wrong one on a board and never know.
Using a template
A template gives you a board with sensible widgets already on it, against your own data.
Treat it as a draft. Rename what does not match your vocabulary, delete what your business does not care about, and change the scopes. A template accepted unchanged is somebody else's opinion about your business.
Use one when starting cold. It is considerably easier to edit a board than to face an empty one, and the templates encode reasonable choices about what belongs together.
Do not use one to look busy. A templated board nobody asked for is another board nobody owns.
Orphaned widgets
The library shows widgets that are on no board, and that list repays a look once a quarter.
Three things end up there. Something abandoned half-built, which should be deleted. Something superseded by a better version, which should also be deleted. And something genuinely useful that nobody put anywhere, which is the reason to look.
The third case is common enough to justify the habit, and it is free to check.
Where to start if this is all new
Explore first, for fifteen minutes, to learn the four words.
Then Measures, to see what is already defined for you.
Then one widget in the Studio, against something you already know the answer to. Building a chart whose result you can verify is the fastest way to trust the engine.
Then a board, with three widgets, for a question you actually have.
That is about an hour, and it is the difference between using this surface and avoiding it.
Templates are opinions
A board template encodes somebody's view of what matters for a given area, and that view is a reasonable default and not your business.
Take one, then do three things: rename anything that does not match your vocabulary, delete anything nobody will look at, and check every scope. A template accepted unchanged is a board nobody owns, which is the thing this whole section warns about.
What belongs where
A short allocation, since the two surfaces overlap in people's heads.
Library is your organisation's own work. It grows, it needs pruning, and it is the first place to look before building.
Explore is fixed product education. It does not grow with your data and it is the first place to send somebody new.
Neither is a place to spend time daily. Library is a lookup; Explore is a one-off hour that saves several later.
When to add to the library rather than a board
Save to the library when the widget answers a question more than one board would ask, and add straight to a board when it only makes sense in that board's context.
The test is whether you can name a second place it belongs. If you can, it is a library widget. If you cannot, putting it in the library just adds something for the next person to scroll past.
The habit that matters most
Look before you build. Library first, then Explore for a template, then the Studio.
Most organisations end up with far more charts than questions. The surfaces here are arranged to make reuse the easy path, and the only thing that makes it work is the ten seconds of looking.
Did this answer your question?
No, ask a person