Using loops (and running them in parallel)

A loop is one trip around the LEARN phases (L → E → A → R → N). Loops are built for learning quickly: you can run as many plans as you like inside a loop, so a loop might last a day, a week, a month — whatever you can realistically plan and resource.

Shorter loops are usually better: they make it easier to get and commit resources, and that keeps planning simple. Aim for the shortest loop that still produces a real piece of evidence.

Starting a new loop

There are two ways a new loop begins:

After Next or Not? — when you pivot or persevere at the end of a loop, Sandbox opens a fresh loop back at Look & Listen, carrying your learning forward.
On demand — you can start a new loop yourself, even while another loop is still running.

Running loops in parallel

Sometimes you need to begin new work before the current loop is finished — and you can't just add it to the running loop without putting its committed plans at risk. Common cases:

New resources free up. A person or budget becomes available mid-loop and can start something now, rather than waiting for the current loop to close.
A separate bet. You want to test a distinct idea or audience on its own cadence without disturbing the work already committed.
Different timeboxes. One stream is a quick week-long probe while another is a month-long build — so they each get their own loop.

In these cases, just create another loop (as many as you need) and run them in parallel. Each loop is an independent, self-contained unit of work with its own phases and plans.

What carries over (and what doesn't)

Manually starting a loop gives you a completely clean, empty loop — nothing is copied across.
Finishing a loop at Next or Not? (pivot or persevere) automatically clones the not-yet-begun work into the new loop — draft experiments, uncommitted sprint plans, and to-do work packages — so pending work isn't lost. Originals stay put in the loop they came from.

Moving work between loops

An admin can copy any artefact, experiment, or sprint plan into another active loop (the original is untouched), or move it when it hasn't begun yet. Look for the move/copy control on the item's row. This is handy for shuffling planned work onto the loop that will actually run it.

How loops relate to the project

Loops are autonomous, but they all roll up to the project. Every loop's work contributes to the project's overall reporting — so running several loops doesn't fragment your picture of progress; it enriches it. (Project-wide metrics and reports are coming in a later phase.)