Amazon DSP Performance Data: Why Seeing It Is Not the Same as Acting on It

Every Wednesday, Amazon drops a set of files into your DSP portal that tells you exactly how your operation performed last week. Your scorecard PDF. Your CX Feedback report. Your POD compliance data. Your Netradyne safety spreadsheet. Your Engine Off Compliance file. Your Proper Parking Sequence report. Your delivery completion data, your DNR breakdown, your customer escalation log.

It is, by any measure, a comprehensive picture of your operation.

And for the majority of DSP owners, it sits in a folder — reviewed once, maybe twice, absorbed at a surface level, and then replaced by the same files the following Wednesday. The metrics that were moving in the wrong direction last week are still moving in the wrong direction this week. The driver whose seatbelt-off rate has been climbing for three weeks has not been coached. The route that keeps producing DNR flags is still built the same way. The EOC compliance issue that appeared two weeks ago has not been addressed because nobody connected the data to a specific action and made sure that action happened.

This is the central problem of DSP operations in 2026. It is not a data problem. Amazon gives you more performance data than almost any other contractor relationship in logistics. It is an action problem — the gap between seeing what the data says and executing the operational response that the data is calling for.

This guide is about closing that gap.

Understanding What Amazon Is Actually Sending You

Before you can act on your performance data, you need to understand what each file is measuring, how it feeds into your scorecard, and which files contain the signals that require the fastest response.

The Weekly Scorecard PDF

Your scorecard PDF is the master document — the composite view of how Amazon has scored your operation across the four primary categories that make up your Overall Standing: Safety and Compliance, Reliability, Team, and Quality. Each category is made up of specific metrics, each of which has its own threshold for each performance tier.

Your Overall Standing is an equally weighted average of these four category scores. This matters because a single category dragging is enough to pull your overall tier down even if the other three are strong. Most DSP owners who are stuck between Good and Fantastic have a problem in one specific category, not across the board — and fixing it requires identifying which category is the anchor and which specific metrics inside it are below threshold.

The scorecard PDF updates on a weekly cycle, but it is not a real-time document. The data it reflects is already days old by the time it appears in your portal. This lag is one of the reasons that weekly scorecard review alone is an inadequate monitoring strategy — by the time you see a metric problem in your PDF, the operational window to prevent further damage from that same week's performance has already closed.

Netradyne Safety Data

Your Netradyne spreadsheet is the most granular safety document in your data stack. It breaks down every safety event category by driver — seatbelt-off rate, speeding event rate, sign and signal violations, distracted driving events, following distance incidents, and harsh braking events — with the raw counts, rates per 100 trips, and the threshold benchmarks for each tier.

This is the document that drives your FICO score, which feeds into your Safety and Compliance category on the scorecard. The Netradyne data comes daily, not weekly, which means it is your earliest warning system for safety metric problems. A driver who starts accumulating sign violations on Monday is visible in Netradyne data on Tuesday. By Wednesday's scorecard PDF, those violations are already reflected in your weekly score. If you are waiting for the scorecard PDF to identify safety problems, you are responding to last week's data — not this week's.

The Netradyne portal at netradyne.com gives you video evidence for every flagged event. This is critical for two purposes: coaching conversations that need to show a driver specifically what they did, and dispute submissions for events that were incorrectly flagged by the sensor system.

Engine Off Compliance (EOC)

EOC measures whether your drivers are turning off their vehicle engines when they stop for deliveries. Amazon added this metric to address both safety concerns and fuel efficiency, and it now appears as a formal scorecard input. A driver who leaves their van running at every stop — a habit many drivers develop without realizing it is being tracked — will drag your EOC compliance down across hundreds of stops per route per day.

EOC is a coaching problem with a clear data signal. The EOC file shows you which drivers are non-compliant and at what rate. The operational response is straightforward: identify the drivers, coach them specifically on the behavior, document the coaching, and monitor for improvement. The problem is that this file arrives separately from the main scorecard PDF, which means it often gets reviewed separately — or not reviewed at all — by operations teams that are already overwhelmed with the primary scorecard data.

Proper Parking Sequence (PPS)

PPS tracks whether your drivers are following Amazon's required parking safety sequence before exiting the vehicle for a delivery — gear in park, hazard lights on, engine off, in that order. It is a relatively new metric, added to Amazon's performance tracking as part of their broader push on delivery safety, and it is one that catches many DSP owners off guard because their drivers were never explicitly trained on the sequence.

Non-compliance with PPS does not just affect your scorecard — it is a genuine safety issue. Vehicles that are not properly secured before a driver exits are a liability. But from a scorecard management perspective, PPS is a trainable metric. A driver who does not know the sequence and gets coached on it specifically will typically correct the behavior. The problem is catching it before it compounds into a meaningful scorecard impact.

CX Feedback and Customer Escalation Data

Your CX Feedback report captures customer sentiment across your deliveries — specifically the Positive Response Rate, which is the percentage of customer feedback that is positive. Amazon's threshold for Fantastic in this metric is a PRR of 98 percent or higher. Your customer escalation data shows the specific complaints that were filed, categorized by type.

These two documents together tell you where your customer experience is breaking down. High escalation rates in a specific category — behavioral complaints, delivery location issues, missed deliveries — point to specific operational or coaching interventions. But the intervention only happens if someone is reading these documents carefully, connecting them to specific drivers or routes, and following through with action before the same pattern produces the same results next week.

The Action Gap: Why Most DSPs Are Stuck

Most DSP owners are not stuck because they lack data. They are stuck because the process between receiving data and taking action is informal, inconsistent, and always competing with more immediate demands on their time.

Wednesday morning, you download your scorecard files. You open the PDF, scan the numbers, feel relief or frustration depending on what you see. Maybe you flag one or two things to address. Then a driver calls with a route problem, a station manager needs something, there are messages in the group chat, and the scorecard files go into the folder. Thursday arrives with new Netradyne data. By Monday, the coaching conversations you intended to have based on Wednesday's data have not happened. By next Wednesday, you have a new set of files and the same backlog of unaddressed issues from last week.

This is not a failure of intention. It is a structural problem. The data review function and the operational response function are both sitting with the same person who is also managing dispatch, HR issues, vendor relationships, and everything else that comes with running a DSP. Attention that is divided across all of those functions is not enough to close the action gap consistently.

The result is a scorecard that reflects whatever your operation naturally produces — not the scorecard that systematic, weekly data-driven intervention would produce.

What a Real Data-to-Action Process Looks Like

The DSPs that consistently hold Fantastic Plus have operationalized their response to performance data. They do not hope that reviewing the data will lead to action — they have built a specific process that connects every metric signal to a specific responsible person and a specific deadline.

Wednesday: Download, Review, Flag, Assign

The moment the scorecard files are available, someone on the operations team downloads all of them — not just the scorecard PDF, but the Netradyne data, the EOC file, the PPS data, the CX Feedback report, and the customer escalation log. Every file, every week, without exception.

Each file gets reviewed against the previous week's numbers, not just against Amazon's thresholds. A metric that is above threshold but moving in the wrong direction is a flag. A metric that is at threshold is a flag. A metric that is below threshold is an urgent flag. The distinction matters because acting on a metric before it crosses a threshold prevents a scorecard impact. Acting after is damage control.

Every flag gets assigned: which driver, which route, which behavior, which metric, what the current number is, what it needs to be, and who is responsible for the coaching conversation or operational fix. Not a general note — a specific assignment with a deadline.

Thursday Through Tuesday: Execute, Document, Verify

Coaching conversations happen with documented records. The Netradyne portal gets used to pull video evidence for coaching sessions that involve specific driving events. EOC and PPS conversations happen with the specific drivers whose data shows non-compliance, with timestamps and signatures on the coaching record.

Route-level problems identified in the delivery data get reviewed with dispatch — are there structural issues with a specific route that are producing consistent DNR or DCR problems? Is a specific zone consistently generating delivery exceptions? These questions get answered before the same route goes out the same way next week.

Every coaching conversation, every operational adjustment, every dispute submission gets documented in a format that is retrievable when you need it — not in a mental note, not in a text message thread, but in a written record with dates and specifics.

The Daily Layer: Netradyne Is Not a Weekly Document

The single most important habit shift for DSPs that want to hold Fantastic Plus is treating Netradyne data as a daily document, not a weekly one. Netradyne events accumulate daily. Coaching that happens on Tuesday for Monday's events can prevent the same behavior from producing the same events Wednesday through Friday. Coaching that happens on Wednesday after the scorecard PDF is reviewed is responding to events that already happened across an entire week.

This means someone needs to be in the Netradyne portal every day — not for hours, but for enough time to flag new safety events, identify which drivers are accumulating events at a rate that will affect the weekly score, and make sure coaching is happening before the week's data is locked.

EOC and PPS: The Metrics Most DSPs Are Leaving Unattended

Because EOC and PPS arrive as separate files from the main scorecard PDF, they are consistently the most under-managed metrics in the DSP space. Many owners are reviewing their scorecard PDF every week without ever looking at their EOC file. Their EOC compliance is dragging their Safety and Compliance score down, and they do not know it because nobody is reviewing the file.

The fix is simple but requires discipline: EOC and PPS review is a non-negotiable part of Wednesday's data review, not an optional add-on. Both files get opened, both get reviewed against thresholds, and any non-compliant drivers get flagged for coaching before Thursday.

The Difference Between Watching Data and Managing Performance

There is a version of data management that feels like performance management but is not. It looks like this: you open your scorecard every week, you know which metrics are above and below threshold, you have a general sense of which drivers are struggling. You are watching your data.

Performance management is different. It means that every metric signal produces a specific action, every coaching conversation is documented, every operational adjustment is tracked for effect, and the performance of your operation next week is systematically different from this week because of what happened between Wednesday and Wednesday.

The gap between watching and managing is where most DSP owners are losing scorecard standing and bonus money they do not realize they are losing.

What This Requires That Most Owners Cannot Provide

Running a true data-to-action process requires consistent attention from someone who is not managing fifteen other things simultaneously. The data review takes time. The coaching documentation takes time. The daily Netradyne check takes time. The EOC and PPS review takes time. The route analysis takes time.

None of these tasks is difficult. All of them are consistent — and consistency is exactly what breaks down when the person doing them is also running dispatch, handling driver call-outs, managing the station relationship, and trying to have a conversation with their family before 10pm.

This is the structural problem that outsourcing solves — not because a remote team is smarter than you, but because they have the dedicated attention that the function requires and you do not have the bandwidth to provide it while running everything else.

At Nizod, our DSP operations team runs this process for every client we work with. We download every performance file on Wednesday. We review every metric against thresholds and against the previous week's numbers. We flag every driver who needs coaching, prepare the evidence from the Netradyne portal, and document every conversation. We check Netradyne data daily. We review EOC and PPS files as a standard part of the weekly review — not as an afterthought. And we connect every data signal to a specific operational action with a deadline.

The DSPs we work with do not have a gap between their data and their operations. They have a team whose entire job is closing that gap, every week, without letting anything slip.

If your scorecard is reflecting what your operation naturally produces rather than what systematic data-driven management would produce, that gap is costing you. In Fantastic Plus bonuses you are not hitting. In dispute windows you are missing. In coaching conversations that are not happening in time to prevent the same metrics from moving the same direction next week.

Reach out through our contact page or visit our Amazon DSP Operations service page to learn more about how we run performance data management for DSP owners.