How to improve your DPMO score on the Amazon Cortex scorecard

DPMO is the Cortex metric that feels the least fair. A driver misses a scan, a parcel arrives damaged, a customer disputes a delivery — and suddenly your Defects Per Million Opportunities number has moved. Counted per million stops, even five or six defects on a 100-stop fleet can shift your weekly status from Fantastic to Great.

The good news is that DPMO is almost entirely preventable. Most defects trace back to the same small set of habits and the same small number of drivers. Here is how to find them and fix them before Monday's scorecard.

What DPMO measures and how it is calculated

DPMO in the Cortex context counts physical delivery defects: damaged or missing items, wrong-address attempts logged as exceptions, lost packages, and — depending on your Amazon programme — certain scan compliance failures. The figure is normalised to a per-million-stop scale so that a small DSP and a large one can be compared fairly. The practical effect for a small fleet is that a very small number of incidents can produce a very large DPMO number.

The threshold for each Cortex tier varies and can change between programme periods. What matters operationally is the trend: a rising DPMO mid-week is a signal to act immediately, not to wait for the final weekly number.

The defect types that move DPMO most

Most DPMO events fall into three categories. Understanding which category is driving your number tells you where to intervene.

• Package damage at delivery. A parcel arrives with visible damage and the customer records it. Often linked to van loading — parcels stacked badly or unprotected during the route. The van check before dispatch is the upstream fix, not the driver conversation after the fact.
• Address and access exceptions. A stop marked as undeliverable due to an address problem, gate code, or refused access can register as a defect depending on how it is logged. Drivers who log exceptions inconsistently, or who skip the contact attempt, create these at a higher rate than they realise.
• Missing and lost items. A package leaves the station but does not arrive. Could be a van loading error, a parcel left at the wrong address, or a scan that does not match the actual delivery. These are the hardest to investigate retrospectively, which is exactly why preventing them upstream matters.

Finding your highest-DPMO drivers before the scorecard lands

The weekly Cortex scorecard tells you what happened. It does not tell you which driver caused it. Waiting for the Monday number means you are always reacting to the previous week rather than preventing the current one.

Pull your defect events mid-week — by Wednesday at the latest — and sort them by driver. In most DSPs, two or three drivers account for the majority of DPMO events over any given four-week period. Those names are where your coaching time goes. The rest of the fleet is either consistent or the occasional outlier you address individually.

The van check connection

Most damage defects are preventable before the van leaves the station. A pre-departure walkaround that checks parcel condition, load security, and package counts will catch the majority of defect-producing events before they become customer complaints. A driver who does this every morning — not just when reminded — will almost always have a lower DPMO than one who skips it.

The return check matters equally. A van that comes back with unchecked condition, no end-of-shift photos, and no note of anything unusual cannot be properly cleared for the next morning. The condition report at end-of-shift is what closes the loop between the driver and the office and keeps damage events from being disputed weeks later.

Closing the feedback gap

Drivers cannot correct behaviour they do not know is causing problems. A DPMO event that surfaces in Monday's scorecard, for a delivery that happened the previous Tuesday, is seven days of missed coaching. The feedback gap is the main reason DPMO stays sticky even after you have identified the right drivers.

The fix is a short, daily loop: look at the previous day's defect events in the morning briefing, name the driver and the stop, and agree a concrete change. Done consistently over two weeks, the numbers move. Skip it for three days running and the pattern re-establishes itself.

The morning after a red DPMO week

When the scorecard lands red on DPMO, the first thirty minutes matter. Work through these in order:

• Pull the individual defect events for the week, sorted by driver. Identify the top two contributors.
• For each event, check the stop detail: what was the parcel type, what was the logged reason, was a contact attempt made. Look for patterns across the same driver across multiple days, not just their single worst event.
• Schedule a one-to-one with each high-DPMO driver within 24 hours — not a group briefing. The conversation is about the specific events, not general standards.
• Confirm the van check process was followed for the affected routes. If the pre-departure check was skipped, that is the root cause to address first.
• Set a mid-week checkpoint: look at the same drivers' defect events by Wednesday to confirm the pattern has changed. If it hasn't, move to a formal performance conversation.

Where DSPOps fits in

DSPOps pulls Amazon scorecard data into the platform and breaks it down per driver and per route — including DPMO. Instead of waiting for Monday's number, your OSM can see which drivers are accumulating defect events mid-week and run the morning briefing with the right data. Drivers see their own performance in the driver portal, so the drivers with the highest DPMO are aware of their standing without needing a separate conversation to prompt it. The van inspection tool records daily walkarounds with photos from any phone, giving you an audit trail of every pre-departure check.

If you want to see what your DPMO breakdown looks like inside DSPOps, you can book a 20-minute demo — we'll set up your data live during the call.