Purchase Order Approval Workflow: Why Buyers Skip It

A purchase order approval workflow is the rule set that decides who can commit a company's money, at what dollar amount, and who has to sign off above it. Most writing about it treats it as the buying company's problem. It is also yours, because your customer's purchasing policy is real whether or not your ordering channel knows about it, and when the channel cannot reflect it, the buyer leaves the channel.

There are three moving parts, and they are worth naming plainly: the requester who needs the part, the self-approval limit that says how much they can commit alone, and the approver chain that handles anything above it. Everything else is plumbing. The frame for the rest of this post is simple. An approval workflow is only as real as the ordering channel that enforces it.

Why do buyers route around the approval workflow?#

Because the compliant path is slower than the workaround, and the part is needed today.

That is the whole answer, and it is worth resisting the urge to dress it up. A maintenance buyer with a line down does not sit in a queue on principle. They call the rep, or they email a PO, and the approval happens verbally or in an inbox.

The cost of that is measurable. A Sievo guide to maverick spend, updated in March 2026, cites an SDC Executive figure putting the loss at an estimated 10 to 20 percent of targeted savings, and a Hackett Group figure of up to 16 percent of negotiated savings when stakeholders purchase outside approved suppliers. The driver named in that same March 2026 material is approval delay, not intent. People are not trying to break policy. They are trying to get a bearing.

Here is the distributor's version of the problem, which is easy to miss because it does not look like a problem. The routed-around order still arrives. Revenue looks fine. What you actually lost is the record, the pricing accuracy on that line, and any chance of that buyer self-serving the next time. You retrained them to call you.

So say it plainly: this is a design failure, not a discipline failure. People follow the process that works.

What a workable approval chain has to handle#

Four things, and most channels get one or two of them.

Self-approval limits. A maintenance buyer can commit up to their threshold without waiting on anyone. This is the part that makes routine reordering survivable, and it is the single biggest lever on bypass volume.

Escalation above the limit. Anything over the threshold routes to whoever can approve that amount. The chain should be visible to the buyer before they submit, not discovered afterward when nothing happens for two days.

Multi-location structure. A branch's rules and a corporate policy are not the same rules. A policy set once at the parent should apply down the tree rather than being retyped for every account, and a chain should be able to escalate to an approver who sits above the branch.

A worked example. A plant maintenance buyer with a $2,500 self-approval limit reorders a case of 6205-2RS bearings. It clears instantly, no chain, on the record. The same buyer's $18,000 seasonal stocking order routes up one level to someone who can commit that amount. Same catalog, same contract pricing, two different paths. Both are documented, and neither required a phone call.

The step almost every workflow gets wrong#

The change request.

Here is the failure mode. An approver looks at an order, wants one line changed, and has only two buttons: approve or reject. So they reject it, or more often they pick up the phone. Either way the order restarts in email, and the careful workflow you built now covers roughly none of what actually happened.

The correct behavior is a third path. The approver sends the order back with a required comment explaining what needs to change. The buyer edits the lines and resubmits. Nobody leaves the system, and the reason for the change is captured in writing rather than in somebody's memory of a call.

There is a non-obvious correctness rule buried in this, and it is the one workflows most often get wrong: when the contents change, prior approvals do not carry over. An approver who approved the old order did not approve the revision. The chain has to be rebuilt on resubmission rather than reused, and the voided steps stay in the history rather than being deleted.

That sounds pedantic until an invoice gets disputed. Then it is the difference between an audit trail and a story someone tells later.

Where the record has to live#

If the approval happened in an email thread, the record is that thread, and it disappears the day the person leaves the company.

At minimum, keep five things: who requested it, what the price was at the moment of submission, who approved and when, what changed between versions, and what the final committed amount was. Price at submission matters more than people expect, because contract pricing moves and "what did we agree to" is the most common form of a disputed invoice.

The distributor's stake here is concrete. A clean approval record is what keeps a disputed invoice from becoming a credit memo, and it is what lets an inside sales rep answer "where is my order" without guessing or apologizing.

This is the part of the problem Copiara is built around, and it is worth being precise about why it needs building at all. Shopify B2B covers a lot on every paid plan now: companies, company locations, payment terms, PO numbers at checkout. It has no buyer-side approval step on any tier. There is nowhere for a spend limit or an approver to live, so the chain falls back to email by default, which is the failure this whole piece describes.

Copiara's approvals put the spend limit and the routing on the merchant's own Shopify store, so the approval happens in the same place the order does rather than in a parallel inbox, and it clears before the order reaches Shopify checkout.

It is worth saying what this is not. Heavy punchout and EDI integrations do solve approvals, and they solve them well, for the handful of accounts large enough to justify the integration project. Everyone below that threshold has historically been left ordering by email. That is where most distribution volume actually sits.

How to fix an approval process buyers avoid#

Five moves, in the order that actually works.

1. Measure the workaround, not the policy. Count how many orders arrive by email or phone from accounts that already have store access. That number is your real compliance rate. Then ask three of those buyers one question about why, and believe the answer.
2. Raise the self-approval limit for routine reorders. Most bypass volume is small, repetitive, and low risk. It is also what trains a buyer to leave the system, and once they are out for the small orders they are out for the large ones too.
3. Make the chain visible before submission. A buyer who can see that their order goes to one named person, right now, will wait. A buyer staring at an invisible queue will call you.
4. Give approvers a real request-change path. If the only options are approve and reject, your approvers will use the phone, and the phone is not a system of record.
5. Rebuild the chain when the order changes. Correctness beats convenience here, every time.

Common questions about purchase order approvals#

What is a good self-approval limit? There is no universal number, but the useful test is coverage rather than caution: set it high enough that the routine reorder traffic clears without a chain. If most of a buyer's monthly orders are hitting an approver, the limit is doing nothing but adding delay to purchases that get approved anyway.

Does an approval workflow slow ordering down? Only the orders that should be slowed. A workflow that routes everything is a workflow buyers escape. A workflow that routes exceptions is one they keep using, which is the only way it captures anything.

What happens if an approver is out? This is why chains need more than one qualified approver at a level, and why escalation to an ancestor in the account structure matters. A single-approver chain is an out-of-office away from being a phone call.

Who should own the approval rules at a distributor? The rules belong to the customer, but the channel that enforces them belongs to you. Practically, that means the admin side of your ordering system needs to let an account's structure and limits be configured without a support ticket or a developer.

Copiara is being built as a Shopify app and is not on the Shopify App Store yet. If you run a distribution business and approvals are currently happening in your reps' inboxes, talk to us about what your accounts would need.