The Evidence You Do Not Have to Provide
A high-value stock write-off here needs a second person, a specific permission, and a value test that cannot be gamed by splitting it. What it does not need is a photograph, a note from a supervisor, or any document at all. The file field is optional, everywhere, always.
Our stock adjustment has a genuinely good control on it, and one hole. Both are worth knowing precisely, because the good control is what will make you trust the module and the hole is what an auditor will find.
The position, immediately
Approval of a high-value adjustment is properly gated: a value threshold, a dedicated permission, and a separation of duties that no grant can override. Evidence is not gated at all. A file can always be attached and can never be required — so the strongest control in the module rests on a judgement made without any obligation to show what it was made on.
What the control actually does
Set a value threshold for your organisation. Below it, an adjustment approves as before. Above it, two independent things happen, and the second is the one worth paying for.
First, the approver needs a specific permission that exists only for this — not the general inventory permission, a dedicated one. Second, and separately, the approver cannot be the person who raised it, and that rule is not overridable by any grant at all. Somebody with every permission in the system still cannot approve their own high-value write-off.
The value test is also harder to game than it looks. It is computed as the absolute value of the movement, summed. An adjustment that takes stock out and puts stock back does not net to zero and slip under the line — a trick that works against a surprising number of systems.
Two people, a permission, and an absolute-value test. Then a free-text reason and no obligation to attach anything at all.
The hole
Every adjustment can carry a file. It always could — the receiving path has accepted an attachment since it was written, and the attachment module reaches across the product. The gap is not capability. It is obligation.
There is no rule anywhere that makes a document compulsory: not by value, not by reason code, not by warehouse, not for the organisation as a whole. The field is nullable and it stays nullable no matter how large the write-off or how sensitive the reason.
So the sequence that costs money is: somebody raises a large write-off with a reason of "damaged", nobody attaches anything, a second person with the right permission approves it because the reason is plausible, and the record is complete, compliant with every control in the system, and evidentially empty.
Why this bites harder in a distributed, low-margin operation
Nepal's trading and distribution businesses tend to be spread across terrain that makes supervision expensive. A branch is a day away. A godown is further. The person raising an adjustment and the person approving it are often not in the same district, and increasingly not in the same week.
When approval is remote, the attachment is not a nicety — it is the only thing the approver has. A photograph of the damaged carton is the difference between approving a fact and approving a sentence. The system will happily let them approve the sentence.
And it compounds with something else absent: nothing looks for patterns in write-offs. There is no detector flagging a user whose adjustments are unusual, or a reason code that has started spiking. So an evidentially empty adjustment is not merely unexamined at the time — it is not examined later either.
Compulsory evidence is a rule, not a feature
The attachment machinery exists and works across the product. What is missing is a condition in front of it. That makes this one of the smaller pieces of work we publish about, and it is worth stating the shape so you can judge it.
A required-attachment rule
Conditioned on the same things the approval threshold already knows about: value, and ideally reason code. The threshold service is the natural place for it.
Per-reason requirements
A tenant-maintained reason list already exists on every adjustment. Marking some reasons as evidence-requiring is a column and a check.
Write-off pattern detection
Genuinely larger than the other two — flagging an outlier user or a spiking reason code is analysis rather than validation, and worth scoping separately.
We do not publish dates. If evidence discipline is an audit requirement rather than a preference, describe it and we will come back with a written scope, timeline and cost before you commit.
Scope evidence rulesFour questions about stock write-offs, for any vendor
Can I make a document compulsory above a value?
A good answer sounds like
A setting, shown on screen, and a refusal demonstrated.
What it actually means
Ours is no. The distinction between "can attach" and "must attach" is the entire question and it is easy to blur in a demo.
Can the person who raised an adjustment approve it?
A good answer sounds like
No, and not even with full permissions.
What it actually means
This is the control that matters most and the one most often implemented as a permission rather than as a rule. A permission can be granted; a rule cannot.
What value does the threshold test?
A good answer sounds like
The absolute value of the movement.
What it actually means
If it tests the net, an adjustment out and back nets to nothing and never reaches an approver.
What notices an unusual pattern of write-offs?
A good answer sounds like
A named report or detector.
What it actually means
Ours is nobody. Most systems answer this with a report that exists and that nobody has ever opened, which is worth probing.
What AWRA OpsHub does today
- An organisation-wide approval threshold gating all four approve paths — the two on the web and the two on the API.
- A dedicated permission for approving above the threshold, separate from the general inventory grants.
- Separation of duties above the threshold that no permission can override: the raiser cannot approve.
- Value computed as absolute, so an in-and-out movement cannot net under the line.
- A tenant-maintained reason list on every adjustment, and a free-text note.
- File attachment available on adjustments and on the receiving path, and always has been.
- Full audit history — who raised, who approved, when, and what moved.
What it does not do
- Any way to make an attachment compulsory — by value, by reason, by warehouse or organisation-wide.
- Per-reason or per-warehouse thresholds, or a second approval tier. There is one threshold for the whole organisation.
- Anomaly detection on write-offs. Nothing flags an outlier user or a reason code that has started spiking.
- A required note. The reason is chosen from a list; the free-text explanation beside it is optional.
Not ours, by choice
- The approval control is the strongest thing in this module and we would put it against anybody's. The gap described here is narrow and specific, and it sits next to a genuinely good piece of work.
- Attachments are available everywhere. An earlier version of our own writing claimed they were not, which was false; the honest claim is about obligation, not capability.
- Nothing here is Nepali. It applies wherever the approver is not in the same room as the stock, which is most distributed operations.
Set the threshold with somebody who has seen it go wrong
The threshold number is the whole control, and setting it too high is the most common configuration mistake in this module. Bring a year of write-offs and we will find the line with you.
Set the thresholdFrequently asked questions
Can I enforce attachments through the workflow engine instead?
The workflow engine can act on events and route approvals, but the refusal you would need here is a validation at the point of saving the adjustment, and that is not something a rule can insert. Treat evidence discipline as a process requirement enforced by your approvers rather than by the system.
Does the threshold apply to check-ins as well as check-outs?
The value test is on the adjustment and is computed as an absolute figure, so a large movement in either direction is caught. That is deliberate: an unexplained receipt is as interesting as an unexplained issue, and often more so.
What is the practical mitigation today?
Make the attachment a condition of approval in your own policy, and have approvers refuse anything without one. That is weaker than a system rule because it depends on discipline, but it is enforceable by a person and it produces exactly the same record.
