Every company already has a purchase approval process. The only question is whether it was designed or whether it accreted — a Slack thumbs-up here, a “check with finance first” there, a spreadsheet someone maintains out of self-defense. This post is the designed version: nine practices that make approvals fast for requesters, bearable for approvers, and enforceable for finance.
1. Put the approval before the commitment
The single most important property of an approval process is when it runs. An approval that happens after the vendor started work, or when the invoice arrives, is a notification with extra steps — the money is already committed. Everything else on this list assumes the sequence is: request → approve → purchase order → delivery → payment. If your current process approves anywhere later, fix that first.
2. Route by amount, category, and department — not the org chart alone
“Everything goes to the manager” is not routing, it’s a queue. The decision of who approves should reflect risk: bigger amounts warrant more or more senior approvers, some categories deserve a specialist eye (software goes past IT, marketing spend past whoever owns the budget), and departments shouldn’t approve each other’s spend by accident.
3. Set an auto-approve floor
Approving a $30 purchase costs more in attention than it protects in money — and worse, it debases the currency: an approver who rubber-stamps twelve trivial requests a day will rubber-stamp the thirteenth one that isn’t trivial. Pick a floor (many teams start around $100–250), auto-approve below it when the budget allows, and keep the record. Approval fatigue is the quiet killer of approval processes.
4. Make the request carry the decision
An approver should be able to say yes or no from the request alone: real line items with quantities and estimated prices, a suggested supplier, the department or project the cost lands on, and one sentence of why. If approvers routinely have to ask follow-up questions, the form is broken — every clarifying round-trip roughly doubles approval latency.
5. Check the budget at approval time
A budget that’s reviewed at month-end is a report; a budget that’s checked at the moment of approval is a control. The approver should see what this purchase does to the department’s remaining budget before deciding — and the process should decide what happens at the limit: warn, require an admin override, or hard-stop. Which one is a policy choice; having no answer is the failure.
6. Treat approval latency as a metric
The hidden cost of an approval process is waiting. When approvals take days, people learn to route around them — they buy first, batch requests to avoid the queue, or inflate this request to cover the next one. Nudge idle approvals, measure time-to-decision, and remember the goal is not maximal scrutiny but the right scrutiny, quickly.
7. Keep the audit trail automatic
Who requested, who approved, what changed, and when — captured as a side effect of the process, not as a document someone writes afterwards. This is what makes the difference eighteen months later when an auditor, a co-founder, or your own memory asks why a purchase happened. If reconstructing a decision requires searching three inboxes, there is no audit trail.
8. Build in the exceptions: delegation and override
Processes fail at the edges. An approver goes on vacation and their queue silently rots; a genuinely urgent purchase can’t wait for Monday. Plan both paths: approvals must be reassignable, and an admin override should exist — logged loudly, so the exception stays exceptional.
9. Review the thresholds as you grow
The matrix that fits a 12-person company suffocates a 40-person one. Amounts that deserved a founder’s eyes at $5k stop deserving them; new departments need their own routes. Put a quarterly reminder on it — threshold review is a ten-minute job that keeps the whole process legitimate.
A sample approval matrix to start from
Numbers are deliberately round — adjust to your spend profile. The structure is what matters: an auto-approve floor, one middle band, and escalation for the amounts that could hurt.
| Order value | Who approves | Why |
|---|---|---|
| Under $250 | Auto-approved if the budget allows | Attention is spent where the money is |
| $250 – $5,000 | Department manager | Owns the budget the purchase lands on |
| $5,000 – $20,000 | Department manager, then finance | Cross-checks the category and the cash impact |
| Over $20,000 | Manager → finance → founder/CEO | Big enough to warrant the top of the chain |
Add category rules on top where risk lives outside the amount — software subscriptions (renewals compound), anything with a contract term, and any new vendor’s first order are common candidates for an extra step.
Making it stick
None of this holds if it lives in a policy PDF. People follow the path of least resistance, so the compliant path has to be the easy path: a request form that’s faster than writing the email, routing that happens without anyone choosing recipients, budget context that appears without anyone looking it up. That is the actual job of purchase approval software — SpendCue routes by amount, category, and department, auto-approves below your threshold, checks budgets at approval time, and logs every decision, which is this entire post as configuration rather than discipline.