Maintenance spend leaks through the approval step, not the purchase

A maintenance tech needs a $180 circulator pump to finish a job today. The supply house is eight minutes away and closes at 4.

Whatever your purchase order policy says, the tech buys the pump. That is the correct operational decision and it is the point where the spend record breaks, because the purchase now exists and the paperwork does not.

The problem is timing, not compliance

Most maintenance spend controls are written as though the approval can happen before the purchase. For planned work it can. For the majority of maintenance purchases it cannot, because the need is discovered inside the unit with the resident standing there.

So a policy requiring pre-approval on every purchase produces one of two outcomes, and both are bad. Either techs stop buying parts and jobs take three days longer, or techs buy parts and the PO gets backfilled from memory a week later, which is a document that describes an approval that did not happen.

The workable version accepts the sequence. Small purchases get a spend limit and after-the-fact attachment. Larger ones get real pre-approval, and the threshold between them is a deliberate decision about how much control is worth how much delay.

Set the threshold where the delay stops being worth it

A defensible structure, with the reasoning rather than the numbers being the point:

Under roughly $250, the tech buys it on a card against an open work order, and the record is the receipt photographed at the register and attached to that work order before the day closes. No pre-approval. The control is the work order link and the spend limit, not a signature.

Roughly $250 to $2,500, the supervisor approves, and it can be a two-minute mobile approval rather than a form. Most of this bracket is same-day work, so an approval path that takes an hour is a path that gets bypassed.

Above roughly $2,500, a real PO with a scope, a vendor quote, and property manager or regional approval. At this level the delay is affordable and the exposure justifies it.

Pick your own numbers against your average work order cost. The structure is what matters: one tier with no approval and a hard limit, one tier with fast approval, one tier with real review.

An unlinked purchase is an uncodeable one

Here is why the work order link matters more than the approval.

A purchase attached to a work order carries the property, the unit, and the category automatically. That is everything AP needs to code it: the GL account follows from the category, and the property allocation follows from the unit.

A purchase with no work order arrives at AP as a line on a card statement reading "ABC SUPPLY 04/14 $180.42." Somebody has to email the tech, wait, and get an answer from memory. That reconstruction is where maintenance coding errors come from, and it is also where the three-week close delay comes from.

So the highest-return control in maintenance procurement is not approval at all. It is the requirement that every purchase name a work order.

Where GetDone stops and BillRoute starts

Being precise, since this touches two products.

GetDone holds the work order, the parts and cost fields on it, receipt photos, and the approval step. The maintenance side of the record.

The vendor invoice arriving later, the coding, the approval routing, and the export to the ERP is accounts payable, which is BillRoute's job. GetDone does not process vendor invoices, does not code to a chart of accounts, and does not pay anybody.

What GetDone also does not do, stated because it is the most requested gap in this area: parts inventory. No stock levels, no reorder points, no truck stock counts. Cost and receipt capture per work order, yes. Knowing what is on the shelf, no. If inventory control is what you are shopping for, we are not that, and a vendor who says yes to inventory should be asked how the count stays accurate when a tech takes a part at 6am.

Watch three numbers rather than reading POs

Purchase order approval workflows produce a lot of documents and very little insight. What tells you whether spend control is working:

Percentage of maintenance purchases linked to a work order. This is the one. Below 80% and your maintenance GL is partly fiction.

Average days from purchase to receipt capture. If it is above one, receipts are getting lost and you will find out at close.

Purchases just under the approval threshold, by tech. A cluster of $2,400 purchases against a $2,500 threshold is a finding, and it is one you can only see if the data is clean enough to look at.

That third one is the reason to bother with any of this. Spend control that produces no visible pattern is paperwork.

Pull one card statement

Take last month's maintenance card statement for one property. Count the lines you can attribute to a specific work order without asking anybody.

That percentage is your real control, whatever the policy document says.

What did it come out to?