Common questions about AI Shift Review rules. You manage these under Settings → Rules. For the full guide on writing rules, see Writing AI Shift Review Rules. For how Looper actually evaluates shifts, see AI Shift Review Structure.
How rules work
Why do my completed shifts need manual approval before they're invoiced or paid?
If a completed shift isn't being added to an invoice (or approved for payroll) automatically, it's almost always because a Blocking AI Shift Review rule flagged it — not a glitch. A blocking exception holds the shift until someone resolves it.
Every new workspace starts with a set of default rules, and most are set to Block billing — for example Missing Required Care Documentation, Empty Care Documentation, Missing Client Signature, Missing Employee Signature, and Billable Time Outside Regular Clock-out Window (plus Payable Time Outside Regular Clock-out Window, which blocks payroll). They exist to keep incomplete shifts from being billed or paid by mistake.
Treat those defaults as a starting point: review them under Settings → Rules and adjust them to fit how your agency works — switch a rule to Non-Blocking so it flags without holding the shift, tighten its Definition, or remove it. If shifts are getting held more often than you'd like, that's the first place to look.
What do "Blocking" and "Non-Blocking" mean — and how do I set them?
When you add a rule (Settings → Rules → Add Rule), you give it a Name and a Definition (what Looper should look for), then choose where it applies — Applicable to Billing, Applicable to Payroll, or both. For each area you pick a level:
Blocking (Billing) — a flagged shift won't be automatically added to an invoice until the exception is resolved.
Blocking (Payroll) — a flagged shift won't be automatically approved for payroll until the exception is resolved.
Non-Blocking — the shift is flagged for awareness, but it still flows through to invoicing and payroll; nothing is held up.
A rule can apply to billing, payroll, or both — with a different level for each — and every rule must apply to at least one. Use Blocking for issues that must be fixed before money moves (missing care documentation, missing signatures); use Non-Blocking when you just want visibility.
Why didn't Looper flag a shift I expected it to flag?
A few common reasons:
The Definition is too vague. Specific criteria ("travel time exceeding 30 minutes") work; vague ones ("travel time looks high") don't.
The data isn't in the system. Looper can only check what Careswitch records. If a field isn't captured, no rule can fire on it.
Wrong terminology. Use the words billable or payable to point Looper at the right data set. "Recorded" should only be used for Care Documentation.
AI margin of error. Looper is an AI — it will occasionally miss. Adjust the Definition and try again.
Why did Looper flag a shift it shouldn't have?
False positives usually trace back to a Definition being too broad. Tighten the criteria (specific thresholds, narrower scope) and the flag rate drops. If a single rule is generating too much noise, consider switching it to Non-Blocking or removing it.
Writing better rules
How specific should my rules be?
Specific. "Excessive travel time" doesn't tell Looper what to look for; "Payable travel time exceeding 30 minutes" does. Numbers, units, and exact field names ("billable hours", "scheduled end time", "care task responses") give Looper something concrete to evaluate.
What's the difference between "billable", "payable", and "recorded"?
Billable: what you charge the client/payer for. Use this when your rule should evaluate revenue-side data.
Payable: what you pay the caregiver. Use this when your rule should evaluate cost-side data.
Recorded: reserved for Care Documentation (care tasks, summaries) — not for time or mileage.
If you want a rule to catch both billable and payable values, say "billable or payable" explicitly in the Definition.
How many rules should I have?
Quality beats quantity. Most agencies do best with 10–20 well-tuned rules covering their biggest risk areas. Too many rules means almost every shift has exceptions, which trains your team to ignore them — defeating the purpose. Start with the defaults and add a few of your own for issues you actually run into.
Can I test a rule before turning it on for real?
Yes — add it as Non-Blocking. It'll surface flags without holding up billing or payroll. Watch what it catches over a week or two, refine the Definition, and switch it to Blocking once you're confident it's catching the right shifts.
Managing rules over time
Will Looper review old shifts when I add a new rule?
New rules apply going forward — existing shifts aren't re-checked automatically. A shift is re-checked whenever it's edited or re-processed (from the shift's page, or via billing/payroll), so editing a shift will re-run the current set of rules against it.
What happens to existing exceptions if I delete a rule?
No new shifts will be flagged by it. Flags it already created clear themselves automatically — they're marked resolved the next time each affected shift is re-checked — and resolved flags stay on the shift for history.
Can I have different rules for different services or recipients?
Rules are workspace-wide — they apply to every shift in your workspace. If you need different behavior for different clients or services, write the rule's Definition to scope it (e.g., "for shifts with service code X, payable travel time exceeding 30 minutes").
How often should I review my rules?
A quarterly look is a good rhythm. Check what's flagging too much, what's not flagging enough, and what new issues your team has run into that aren't covered yet. Refer back to Writing AI Shift Review Rules for guidelines.
If your question isn't here, see Writing AI Shift Review Rules and AI Shift Review Structure for the deeper guides.