Loop FAQ
Answers to common setup and operating questions about Loop.
For what the app is and how to set it up from scratch, see the
Loop Overview.
Setup
Do I have to turn Loop on before anything gets measured?
Yes. Loop needs a licence, and an admin enables it from the Apps Marketplace. It is off in every organization until someone opts in. Until then nothing is recorded and nothing is measured, including the measurements other apps would otherwise start for you.
Who can open Loop once it is enabled?
Business admins and anyone holding the App admin badge for Loop, always. Managers can also open the overview, but only once an admin turns on Allow managers to view the Loop overview in Settings. That switch is off by default. Everyone else has no Loop surface at all.
Is the Ask Loop advisor switched on already?
Yes — Ask Loop advisor in Ask AI is on by default. It answers “what should we do about X?” in the Ask AI sidebar, grounded in your own numbers. Every answer ends in a plan a named person still has to commit; the advisor never creates anything itself. It follows the same access rule as the app, so managers reach it only when overview access is on.
Permissions and access
Why does a manager only see part of the ledger?
By design — a manager sees only the measurements they are responsible for. That means plans they own or declared themselves, plans scoped to a location they manage, and suggestions they personally acted on. Admins see the whole organization. Managers also do not get the pipeline card, because it is an organization-wide shape.
Why can’t a Loop app admin open AI Outcomes, Suggestions or Corrections?
Those interior tabs need a business admin role, and the App admin badge for Loop is not one. Someone holding only the badge sees Overview, Portfolio, Quarterly Review and Settings. The other five tabs are hidden from them rather than shown and then refused. Give them the business admin role if they need those pages.
Who is allowed to record the decision on a plan?
The plan’s named owner, or a Loop admin acting on their behalf. Anyone else with overview access can read the plan and its receipt but cannot decide it. The same rule governs cancelling a plan, inviting a pilot, and saving a kept plan as a reusable pattern.
Making and measuring plans
How far back can I date a change we already made?
Up to 90 days. Loop takes its baseline from the day the change actually took effect, so a date further back would measure a different organization. Future dates are refused too — declare the change when it genuinely goes live.
How long does Loop watch a number before it reports a result?
30 days by default, and you can set anything from 1 to 365 days. The result is computed on the overnight scoring run, shortly after 03:30 UTC, once the window has fully elapsed. So a 30-day plan reports on the morning after day 30, not on day 30 itself.
Can I change a plan’s metric or window after it has started?
No — a plan cannot be edited once created. If you got it wrong, use Cancel plan on the ledger row and make a new one; that removes the plan and its measurements. A plan that already carries a decision cannot be cancelled, so cancel before you decide.
What are my choices when a result comes in?
Five, each recorded with your reasoning: Keep, Adjust, Expand, Retire and Roll back. Keep and Expand leave the change in force and start a recheck, so a change that quietly stops working reopens itself for review. Adjust immediately opens the next version of the plan, prefilled, with the expectation reset.
When something looks wrong
Why won’t Loop let me record a decision on my plan?
Decisions wait for evidence — a plan with no measured result yet offers only Cancel plan. This is enforced on the server, not just hidden in the page, so the button will not appear early for anyone. Wait for the window to close, or cancel the plan if it was a mistake.
Why did our result come back inconclusive?
The window closed with too little data to call honestly. Most metrics need at least 5 records in the measurement window; a few low-volume ones set a higher bar of their own. The receipt states the actual count and the bar it missed, so you can pick a longer window or a busier metric next time.
A measurement failed — can I re-run it without waiting for tonight?
Yes, if you are an admin: Retry now on the ledger row re-runs the scoring immediately. Failed measurements are also retried automatically on each overnight run, up to 10 attempts. After that the row stays visibly failed rather than being quietly abandoned. The reason for the failure is shown on the row itself.
Why can’t I pilot this change at another location?
Two conditions have to hold. The change must have been decided Keep or Expand. Its metric must also be one Loop can compute for a single location — an organization-wide-only metric cannot honestly measure one site’s pilot. Loop additionally skips any location with no manager to invite, and names the ones it skipped.
I’ve been invited to pilot a practice — what am I agreeing to?
You are agreeing to run the practice at your location and own its result. Nothing is measured until you accept and confirm the date it actually went live at your site. That date starts a measurement for your site alone; the original site’s result is never assumed to carry over. You can decline instead, with a reason the inviter will see.
Why are sample stories counting in our totals?
Because you have no real measurements yet. While that is true, the loaded sample stories fill the tiles and pipeline under a banner saying so. The moment your first real measurement lands, every total reverts to real data only. Remove the samples any time from Settings.
Why aren’t we seeing any improvement opportunity cards?
Quiet is a valid state — Loop shows cards only when your own data supports one, and never invents work to look busy. A trend card, for instance, needs at least 20 records in each 30-day window and a worsening of at least 15% before it appears. Dismissed cards also stay gone permanently unless you chose to snooze them instead.
Autonomy, sharing and controls
What can Loop do on its own if I turn on earned autonomy?
Only two routine, reversible things, and only for practice types that earned the right. You cannot set that level by hand; unlike other apps Loop has no AI-mode dial, and Settings shows the evidence behind each earned level. After 3 decided plans with 2 kept, Loop proposes the next turn as a notification you act on. At 5 decided plans, a 60% hit rate and a recheck that held, it may act. Acting means drafting a stalled follow-up plan, or pausing an automation whose plan regressed. Both are reported to the owner and undone in one click, and Earned autonomy is off by default.
What leaves our organization if I turn on “Share anonymized evidence”?
Only a count of organizations. It lets you be counted in cross-customer proof such as “proven at 5 organizations” on a pattern card. That count appears only once 3 or more organizations have opted in. Your name, your numbers and your dates never leave your tenant, and the switch is off by default.
Where do I pause Loop’s automations in a hurry?
In AI Operations in the admin sidebar, not inside Loop. That is where pause-all and resume-all, the approvals queue and action history live. It is available to every business admin whether or not Loop is enabled. Loop owns the learning half — the ledger, plans and decisions — so the emergency controls deliberately stay outside it.
More help
- Loop Overview
- Ask AI — the assistant answers Loop questions from these articles.