AOG Procurement: Where the Expedite Money Actually Goes

An AOG event compresses every normal control out of the process, and correctly so. Nobody should be running a three-quote comparison while an aircraft is out of service. The cost of the delay is larger than the cost of the part by a margin that makes the comparison pointless.

The problem is that the same reasoning gets applied retrospectively. Because no individual AOG purchase can be second-guessed in the moment, the total is rarely second-guessed at all, and a line that nobody examines is a line that only grows. The useful question is not whether any given expedite was justified. It is what proportion of the annual expedite bill came from events that were genuinely unforecastable.

#Three categories hide inside one number

Pull a year of expedite freight and premium-priced spot buys, tag each one against the event that caused it, and it separates into three groups with very different economics.

Genuinely unforecastable

A bird strike, a hard landing, an unexpected finding on a scheduled check. The part was not held because holding it was not rational. The expedite is the cost of operating an airline and there is nothing to recover.

Forecastable in aggregate, not individually

You cannot say which wheel assembly fails next month, but the annual consumption is stable and known. If these are being expedited, the failure is in min/max levels and reorder points, not in the AOG desk.

Not urgent at all

A part that was in stock at another base. A part ordered on an AOG ticket because the AOG ticket is the route that gets attention. A routine replenishment that was left until stock hit zero. This group is pure recoverable cost.

In most operations nobody has ever separated them, because separating them requires somebody to tag several hundred transactions against their cause — a week of unglamorous work that never wins against this week's operational problem. It is, however, the single piece of analysis that tells you whether your expedite bill is a cost of flying or a symptom of something else.

#The four leaks worth checking first

Stock you already own, in the wrong building

Multi-base operators buy parts they are holding somewhere else more often than anyone likes to admit. The cause is almost never a missing system — the inventory module usually shows the stock perfectly well. The cause is that at 02:00 with an aircraft down, the person on the phone searches the supplier's catalogue before searching their own network, because the supplier's catalogue is the one that answers instantly.

The fix is procedural rather than technical: an inter-base check as a mandatory, logged first step, with the answer required in minutes rather than at somebody's convenience. That only works if a specific person owns answering it.

Reorder points that were set once

Min/max levels tend to be set at implementation and then left, while consumption, fleet mix and utilisation all move. A consumable that now turns over three times faster than when its reorder point was written will hit zero and generate an AOG ticket, and the ticket will be recorded as an AOG event rather than as a planning failure — which is the reason it never gets fixed.

A quarterly review of the top consumables by movement against their current reorder points is a couple of days of work and it removes a category of event, rather than making the events cheaper.

Expedite freight bought on reflex

Next-flight-out is the default because it is the safe answer, and the safe answer is right when the aircraft is waiting. It is not right when the part is going into a check that starts on Thursday. The distinction is knowable at the moment of ordering — it just requires the person raising the order to have been told the required-on-site date rather than the word urgent.

Cores that never went back

An exchange unit carries a core return window. Miss it and the exchange converts to an outright purchase, usually at a price nobody agreed to and on an invoice that arrives long enough afterwards that no one connects it to the original event. This is the quietest line in the whole category: it is not an overspend anyone decided on, it is one that happened because a date passed.

It is also the most mechanically fixable. Every exchange has a known window. A list, checked daily, with somebody chasing the ones inside a week, closes it entirely.

#What an AOG desk should actually be measured on

Response time is the obvious metric and it is not enough on its own, because it can be optimised by simply buying the first thing offered at the first price quoted. A desk worth having is measured on a small set together.

1. Time from request to a confirmed order with a confirmed delivery commitment — not to a first response.
2. Proportion of events sourced from own network stock before external purchase.
3. Expedite premium as a percentage of part value, tracked as a trend rather than judged per event.
4. Core returns closed within their window, as a percentage.
5. Events tagged to a root cause, as a percentage — because untagged events cannot be prevented.

The last one is the one that compounds. Everything else measures how well you handle the events; that one is what reduces how many there are.

#Why this is hard to staff internally

An AOG desk has to be responsive at unsociable hours and diligent during ordinary ones, and those are different temperaments doing different work. Cover the first with your own team and the second — the tagging, the core chasing, the reorder-point review — is what gets dropped, because it is never the most urgent thing on anyone's desk.

It is also, conveniently, the half that requires no airworthiness judgement at all. Someone chasing a core return window or auditing an expedite premium against contract is not making a technical decision. They are working a list, every day, which is exactly the kind of work that benefits from being somebody's whole job rather than the tail end of everybody else's.