Every procurement function sits somewhere on a spectrum between two bad options. At one end, everything requires approval, buyers spend their days raising routine orders, and urgent items wait behind fasteners in an approval queue. At the other, spending happens with no oversight and the first anyone knows is the invoice.
The useful middle isn’t a compromise between the two. It’s applying different levels of control to different classes of purchase.
Segment Before Automating
Automating everything uniformly is how control gets lost. Automating by segment is how it gets targeted. The segmentation itself is the work; the automation is comparatively easy once you’ve decided which purchases belong in which tier.
Low-value, repeat, contracted items with an agreed price and a known supplier need no human decision at all. Requirement triggers, order generates, order sends. Your buyers should not be involved.
Higher-value contracted items can auto-generate but hold for a quick review before release — a buyer confirming quantity and timing rather than re-deciding the source.
Uncontracted or new items need proper sourcing. No automation beyond routing the requisition to the right person with the relevant history attached.
Anything unusual — a quantity far outside the normal range, a new supplier, a price movement beyond a threshold — escalates regardless of value. This is the exception rule that makes the rest safe.
Controls That Survive Automation
Approval limits by value and by role, applied to the order rather than the requisition, so splitting an order can’t be used to stay under a threshold.
Price tolerance checks against the contracted or last-paid price, with anything outside tolerance held for review. This catches supplier price creep that would otherwise pass through unnoticed.
Supplier validation, so orders can only be placed against approved and current suppliers. Dormant or delisted suppliers should fail automatically.
Budget checking where relevant, at the point of commitment rather than at invoice.
Three-way matching on receipt, which remains the fundamental control regardless of how the order was raised.
Where The Time Actually Goes
Measure before you automate. In most procurement functions the volume is dominated by low-value repeat orders and the value is concentrated in a small number of strategic purchases. If your buyers are spending most of their time on the former, automation frees them for the latter, which is where the actual savings are.
That’s the real argument for automation. Not headcount reduction — attention reallocation. A buyer who spends two days a week on routine transactions is a buyer not developing suppliers or negotiating contracts.
Watch What Automation Hides
One caution: automated ordering removes the human who used to notice things. A buyer raising an order manually might spot that consumption looks odd, or that they’re ordering this item unusually often. Automation removes that check.
Compensate with monitoring — consumption anomalies, order frequency changes, spend by category trending against history. The oversight moves from transaction to pattern, and it has to be deliberately built rather than assumed.
ticktick.ai automates requisitions by segment with tolerance and exception rules, and monitors consumption and spend patterns for the anomalies manual review used to catch.
