Appearance
Audits (stocktakes) β
An audit is a stocktake you can actually finish. You pick a set of gear that should be somewhere, walk the shelves scanning what's really there, and Shelf tells you the difference β what it found, what it couldn't find, and what turned up that wasn't on the list.
Nothing else in Shelf does this. Locations tell you where a thing is recorded; an audit is the only place the recorded answer gets checked against the physical one.
When you'd run one β
- Before a big job, on the kit or location the job draws from, so you find the missing radio mic in the warehouse rather than at load-in.
- After a season or a tour, when a lot of gear has moved through a lot of hands and the records have drifted.
- Once a quarter on a location, as the routine that stops drift becoming a write-off nobody can date.
- On a person, when a freelancer finishes up and you want to confirm what came back.
Starting one β
There's a New Audit button on the Audits page, but it doesn't create an audit itself β it points you at the four places an audit can start from, because an audit always needs a set of assets to check, and that set comes from somewhere you're already looking at:
| Start from | Where | Good for |
|---|---|---|
| A location (or several) | Locations β select β Actions β Create audit | Room-by-room or bay-by-bay stocktakes. A location with children can include them. |
| A kit (or several) | Kits β select β Actions β Create audit | "Is this rig actually complete?" |
| A person's custody | A team member's page β Start an audit | Checking back what a freelancer or crew member is holding. |
| A hand-picked list of assets | Assets β select rows β Actions β Create audit | Targeted checks β the ten expensive things, or everything with a category you don't trust. |
Whichever you start from, the dialog asks for the same four things:
- A name. Make it findable later: "Store bay 3 β Q1" beats "Audit 4".
- A description (optional, up to 1000 characters) β worth using for "checking the returns from the Riverside job", so whoever picks it up knows the point of it.
- A due date (optional, must be in the future). This is what drives the reminder emails below.
- Assignees β the people who'll actually do the counting.
The set of assets is captured when the audit is created. Those become the audit's expected assets.
The lifecycle β
Two things worth remembering:
The first scan starts the audit. There's no separate start step, so an audit can't sit at ACTIVE with nothing scanned, and the "started at" timestamp is genuinely the moment counting began.
The expected list is only editable while the audit is still PENDING. Add or remove assets before anyone scans; once counting has begun, the list is fixed, so the counts at the end mean what they say. If the scope was wrong, cancel it and start a fresh one β that's cheaper than a completed audit nobody can interpret.
Scanning through a location β
Open the audit and hit Scan. You get the same camera scanner as the rest of Shelf, with a drawer that fills up as you go and a running count of expected / found / missing.
Practical notes from doing this on a real floor:
- Scan whatever's easiest to hit β the QR code, the Shelf code typed in by hand off a scuffed label, or an alternative barcode if the gear carries one. They all resolve to the same asset. See Barcodes & codes.
- Scanning the same thing twice is fine. The second scan is recognised as a duplicate and doesn't move the counts β you don't have to keep track of what you've already done.
- Two people can scan the same audit at once. Split a big store in half and work from both ends; the counts are kept straight server-side, not on either phone.
- Add a photo or a note to any row as you go β the camera button on each scanned row takes a picture straight from the phone and attaches it to that asset within this audit. This is what you want for "the case is here but the lid's cracked": photograph it while you're standing in front of it, not from memory afterwards.
If you scan something that isn't on the audit's expected list, it's added to the audit as unexpected rather than rejected. That's deliberate β gear in the wrong place is exactly what you're trying to discover.
What "missing" actually means β
Three words do the work here, and it's worth being precise about them because they get used loosely in conversation:
| Status | What it means | What it does NOT mean |
|---|---|---|
| Found | The asset was on the expected list and you scanned it. | β |
| Missing | The asset was on the expected list and, at the moment you completed the audit, nobody had scanned it. | It is not lost, not written off, and its own record is untouched β the asset's status, location and custody are exactly what they were before. |
| Unexpected | You scanned it, but it wasn't on the expected list. | It doesn't imply the asset is in the wrong place β an unexpected result can just mean the audit's scope was drawn too narrowly. |
Nothing is marked missing until you complete the audit. Up to that point an unscanned asset is simply pending, so an audit you walk away from halfway doesn't leave a trail of false "missing" flags behind it.
And missing is a finding, not an action. Shelf will not quietly delete an asset, change its location, or release its custody because an audit couldn't find it. That is always a human decision made afterwards, which is the next section.
Completing and reconciling the drift β
Complete audit does four things:
- Marks every expected-but-unscanned asset missing.
- Writes a completion note holding the four counts β expected, found, missing, unexpected β plus whatever you type into the completion box. Type something. "Bay 3 was mid-repack, three cases were on the truck" is the difference between a useful record and a number nobody can interpret in six months.
- Emails the audit's creator and its assignees the result.
- Adds a line to each asset's own activity, so someone opening that camera next year can see it was audited and how it went.
Then comes the part Shelf can't do for you. Work the two lists:
For each missing item, in rough order of likelihood:
- Is it out on a booking, or in someone's custody? Check the asset's page β if it's legitimately checked out, the audit's scope was wrong, not the gear.
- Is it in a kit that lives somewhere else? Kit members follow their kit's location.
- Is it in the next bay, mis-shelved? It'll probably show up as unexpected on the next audit of that bay.
- Genuinely gone? Update its location or status to say so, add a note explaining what you concluded and when, and β if it went missing at a venue β remember that anyone who finds it can report it through Lost & found months later.
For each unexpected item, decide which is wrong: the shelf or the record. Either move the asset back to where the record says it lives, or update its location to where you actually found it. Do it while the audit is still fresh; an unexpected list you come back to next week is just a list of questions.
Re-run the audit afterwards if you want a clean sheet β Duplicate on a completed audit gives you the same expected list under a new name, which is the fast way to do "and now let's check the bits we chased up".
Due dates, reminders and overdue β
If you gave the audit a due date, Shelf emails the assignees 24 hours, 4 hours and 1 hour before it's due, and again once it's overdue. An audit with no due date is silent β which is the right choice for a casual spot check and the wrong one for the quarterly stocktake somebody will otherwise forget.
Overdue is shown as a badge on the audit rather than a separate status: the audit is still ACTIVE (or still PENDING), it's just late.
Cancelling, archiving and the paper trail β
- Cancel an audit that was started by mistake or overtaken by events. The creator can always cancel their own; Owners and Admins can cancel any audit in the workspace, so a half-finished stocktake doesn't get stuck because the person who started it has gone on leave. Assignees are emailed. A cancelled audit can't be completed afterwards.
- Archive a completed or cancelled audit to take it off the working list. Archived audits are read-only β you can read every scan, note and photo, but nothing can be changed. Nothing is deleted.
- Print receipt opens a printable summary of the whole audit β the counts, every asset with its status, and the notes β which you can send to a printer or save as a PDF. For the jobs where someone needs a signed piece of paper.
Every audit also has its own Activity tab (exportable as CSV) recording who started it, who scanned what and when, assignee changes, due-date changes, and the completion.
Who can do what β
- Owners and Admins can create, edit, cancel, archive and delete any audit in the workspace.
- Base and Self-service users can see and work on audits they've been assigned to β scan, add notes, add photos β and nothing else. That's the right shape for a casual who's been asked to count bay 3 and shouldn't be browsing the rest of the workspace's stocktakes.
If the Audits page shows an unlock screen rather than a list, audits aren't switched on for your workspace yet β an Owner can turn them on from that screen.
Related β
- Assets, kits & labels β the gear you're counting
- Kits & custody β why a "missing" item is often just in someone's hands
- Barcodes & codes β what you can scan, and reading a code aloud off a bad label
- Reports β the standing numbers, as opposed to a point-in-time count
- Lost & found β for gear that left the building and didn't come back
- Troubleshooting β reading a missing or unexpected result without jumping to conclusions
- Glossary β audit vs report, and why "stocktake" means the same thing
