A Shopify or DTC brand shipping 10,000 orders a year or more usually has two or three carriers and a feeling about which one is better. A carrier scorecard turns that feeling into a table: the same metrics for every carrier, cut by service and destination, refreshed monthly. This guide gives a template, an example, a way to read your carrier mix, a review routine and the tools that build one.
Short answer: A carrier scorecard is one row per carrier, service and destination, with seven to eight columns computed from your own tracking events, carrier invoices and claims log: on-time rate against the date you promised, handover time, first-attempt delivery rate, exception rate, damage and loss rate, cost per parcel including surcharges, and claim resolution time. Add each carrier's share of your volume and you have a carrier mix dashboard. A spreadsheet works if the events, the promise and the invoice sit on the same order; Sendcloud, ShippyPro and Perform.AI also sell carrier analytics.
What a carrier scorecard is
A scorecard is not a carrier's own performance report. A carrier measures itself from the moment it scans your parcel in. Your customer measures from checkout, against the date you showed. The scorecard uses your clock and your invoices, so carriers are compared on the same terms.
It is also not one number per carrier. A carrier can be strong to one country and weak to another, so the unit is the lane: carrier, service and destination. The metric definitions, and why each one matters, are set out in full on carrier performance metrics. This page is about turning them into a working scorecard.
Can tracking data measure carrier performance?
Yes, and for most of the scorecard it is the only data you need. Every tracking event has a timestamp and a status. Once each carrier's statuses are mapped onto one set of your own (label created, first scan, out for delivery, attempted, delivered, exception, returned), five of the eight metrics fall out of the events and the checkout promise: on-time rate, handover time, first-attempt delivery rate, exception rate and return-leg time.
Cost per parcel needs carrier invoices matched to shipments, and the two claims metrics need your claims log. The hard part is mapping: carriers do not define "attempted" or "exception" the same way. The checklist for getting events in cleanly is in carrier integration checklist.
The carrier scorecard template
Copy this into a spreadsheet or BI tool, computed per carrier, service and destination country for one calendar month.
| Metric | How to calculate it | Data you need | Why it matters |
|---|---|---|---|
| Volume share | Parcels on this lane divided by all parcels shipped | Shipments table | Tells you how much each result is worth acting on |
| On-time rate | Delivered on or before the date shown at checkout, divided by delivered | Checkout promise, delivered scan | The number your customer feels and your reviews reflect |
| Handover time | Median hours from label created to first carrier scan | Label timestamp, first scan | Shows warehouse and pickup delays the carrier report never shows |
| First-attempt delivery rate | Delivered on first attempt, divided by delivered | Attempted and delivered scans | Missed attempts become tickets and redeliveries |
| Exception rate | Parcels with any exception status, divided by shipped | Normalised exception statuses | Predicts "where is my order" contacts |
| Damage and loss rate | Parcels claimed damaged or never delivered, divided by shipped | Claims log, final status | Direct cost plus a lost customer |
| Cost per parcel | Invoiced amount including fuel, remote-area, oversize and redelivery surcharges, divided by parcels | Carrier invoice matched to shipments | The rate card is not the price you pay |
| Claim resolution time | Median days from claim filed to paid or rejected | Claims log | Shows how much a carrier's guarantee is worth in practice |
Two rules keep the template honest. Use medians for times, because one parcel stuck for a month will distort an average. And leave a cell empty, not zero, when a lane had too few parcels to judge. Agree a minimum (for example 50 parcels a month) before the first review, not after.
Carrier scorecard examples
Here is what one month of a filled scorecard can look like for a brand shipping home delivery to two countries. The numbers below are invented to show the layout and how to read it. They are not benchmarks for any carrier.
| Lane (carrier, service, country) | Volume share | On-time rate | Handover time (median) | First-attempt rate | Exception rate | Cost per parcel |
|---|---|---|---|---|---|---|
| Carrier A, home, UK | 48% | 94% | 6 h | 91% | 3.1% | 4.10 |
| Carrier B, home, UK | 22% | 89% | 18 h | 88% | 4.8% | 3.60 |
| Carrier A, home, Germany | 12% | 81% | 7 h | 86% | 6.9% | 7.40 |
| Carrier C, pickup point, Germany | 18% | 92% | 9 h | not applicable | 2.4% | 5.90 |
How to read it. Carrier B is cheaper in the UK, but its handover time is three times Carrier A's, which points at the pickup schedule: ask for a later collection before you move volume. In Germany, Carrier A is weak on every column and Carrier C's pickup points are faster and cheaper, so the question is whether to shift German volume there or test another home-delivery carrier. The table does not make the decision. It makes it a five-minute conversation.
Which tools deliver the top dashboards for carrier mix optimization?
A carrier mix dashboard is the scorecard plus two views: the share of volume each carrier carries per lane, and what the blended cost and on-time rate would have been under a different split. What follows is what each tool says about itself on its own site, read on 17 November 2026. We have not tested them, and no tool here is ranked as the "most reliable".
| Tool | What its own site says about carrier performance | Price on its own site |
|---|---|---|
| Sendcloud | Shipping Intelligence: "Benchmark carriers side by side on speed and reliability", "Compare delivery lanes and spot underperforming regions", "Break down spend by carrier, lane, and method". Shipping analytics is listed from the Premium plan | Premium "€175/mo + €0.08 per label", Pro "€639/mo + €0.07 per label" (monthly and yearly billing offered). The plan table lists "Delivery performance benchmarking" and "Courier SLA review" as "Enterprise / On demand" |
| ShippyPro | Optimizer: "breakdowns by carrier or service level to evaluate delivery timeliness against SLAs", "Average transit time by destination and Carrier". A separate Invoice Analysis module analyses "carrier invoices to identify errors, discrepancies, and extra charges" | Optimizer is listed as an "Add-on" on the Enterprise plan (Enterprise price "Custom"); the add-on price is not published |
| Perform.AI (formerly Parcel Perform) | AI Decision Intelligence: "Monitor every warehouse and carrier SLA in one view", reports from "90+ charts, including performance scorecards and issue heatmaps" | Not published (the site offers demos) |
| Pango | Carrier performance analytics on the same record as routing, labels and tracking (see below) | See pango.ai/pricing |
Two checks apply to all of them. Which plan includes the carrier views (on Sendcloud and ShippyPro they sit on higher tiers or add-ons)? And where do the promise date and the invoice come from? A tool that sees only labels and scans can compute handover time and exceptions, but not on-time rate or true cost per parcel. The wider buying guide is best multi-carrier shipping software, and the category itself is explained in what is multi-carrier management.
How to read your carrier mix
Start with volume share, because it tells you where a percentage point matters. Then read the mix three ways:
- By carrier. One carrier with most of your volume is a peak-season dependency, whatever its scores.
- By zone or country. Most mix decisions are made here, one country at a time.
- By service. Home, pickup point and express are different promises, and a carrier strong on one may be weak on another.
Then ask per lane: does the carrier with the most volume also have the best on-time rate at an acceptable cost? If not, test before you switch: move a share of the lane (say a fifth) for a month and compare. How to turn the results into routing logic is in carrier selection rules.
A monthly carrier review routine
- First working day: refresh the data. Pull tracking events, promise dates, matched invoices and the claims log.
- Compute the scorecard. One row per lane, the template columns, the minimum-volume rule applied.
- Flag movement, not levels. Mark lanes whose on-time or exception rate moved past your agreed threshold (for example two points) against the last three months.
- Check the invoices against the contracted rate per lane, surcharges listed.
- Hold a 30-minute review with operations, customer service and the contract owner: keep, test, move, or raise with the carrier.
- Send each carrier its own rows and the lanes you want improved.
- Update the routing rules and note the date, so next month shows the effect.
- Weekly in peak. From November to January, run steps 1 to 3 weekly, because a slipping lane costs more when volumes are highest.
Where Pango fits
Pango is an AI-native, agentic operating system for e-commerce logistics. It runs multi-carrier routing, rate shopping, label generation and shipping rules, and its branded tracking normalises the differing statuses carriers return onto one set. Its analytics cover warehouse handover time, warehouse and carrier performance, lost and delayed packages and carrier attribution, and a live customer uses them as a third-party source of truth to renegotiate carrier terms. Because the routing, labels and tracking run on the same order record, the scorecard and the routing rules it changes sit in one place.
To see your own carrier scorecard built from your orders, book a demo.
The bottom line
A carrier scorecard is a lane table graded monthly on your own data. Copy the template, agree thresholds first, read the mix by carrier, country and service, and make one decision per flagged lane. The tool matters less than getting promise dates, tracking events and invoices onto the same order.



