As of November 22, 2026, all three parcel carriers have moved their APIs, and integrations built on the old platforms are either broken or on borrowed time. USPS retired its Web Tools APIs on January 25, 2026, and the old credentials do not work on the new USPS APIs. FedEx Web Services has been on maintenance-only support since July 1, 2026, its test environment closes on December 31, 2026, and the Web Services Track WSDL is disabled on February 1, 2027. UPS’s current API definitions authenticate with OAuth 2.0 tokens, but the UPS pages we could read give no cutoff date for older access methods, so check your own UPS developer account.
Most of the pages about carrier API migration were written by shipping platforms for their own users, and many were written before the changes took effect. This field note is for teams whose ERP, WMS or TMS calls the carriers directly. It gives a dated status taken from each carrier’s own pages, then the work: finding every touchpoint, deciding who owns the credentials, and testing labels, rates and tracking before and after the switch.
Where UPS, FedEx and USPS stand as of November 2026
USPS has finished its move, FedEx is retiring Web Services in phases with dated steps still ahead, and UPS has moved its published APIs to OAuth without a dated cutoff on the pages we could read. The table below is the status as checked on the carriers’ own sites on November 22, 2026; re-check it before you plan around it.
| Carrier | Old platform | Status on the carrier’s own pages | Next dated step | Replacement |
|---|---|---|---|---|
| USPS | Web Tools APIs | “Retired on January 25, 2026”; “availability may be degraded or interrupted at any time” | None; the retirement date has passed | USPS APIs at apis.usps.com, OAuth 2.0 |
| FedEx | FedEx Web Services (WSDL) | Maintenance-only support from July 1, 2026; existing integrations “may continue operating temporarily” | Test environment off December 31, 2026; WS Track WSDL off February 1, 2027 | FedEx APIs (REST), OAuth tokens |
| UPS | Older access methods for the UPS APIs | Current published API definitions use OAuth 2.0 bearer tokens; no cutoff date on the pages read | None published on the pages read | UPS APIs at onlinetools.ups.com, OAuth 2.0 |
A note on wording. Carriers use “retired”, “maintenance-only”, “disabled” and “decommissioned” for different stages. Retired means the platform is officially gone even if some calls still answer. Maintenance-only means it still runs but nobody will fix your integration problems. Disabled means calls fail. Plan against the stage that hurts you, which is usually the first one that removes testing or support.
USPS: Web Tools is retired and the new APIs need new credentials
USPS retired the Web Tools API platform on January 25, 2026, and its replacement, the USPS APIs, uses OAuth 2.0 with credentials that Web Tools users do not already have. Any integration still calling Web Tools is running on a platform USPS no longer supports.
The USPS Web Tools page says: “The USPS Web Tools API platform was retired on January 25, 2026 and service disruptions are underway.” It adds that “availability may be degraded or interrupted at any time.” If some of your calls still return data, treat that as luck, not as a grace period.
What the USPS FAQ says about credentials and testing
The USPS migration FAQ answers the questions integration teams ask first:
- Old credentials: “The Web Tools API credentials are not compatible with the new USPS API platform.” You generate an OAuth token as part of onboarding.
- Testing: the Test Environment for Mailers (TEM) at apis-tem.usps.com “is intended to mirror the production environment” and runs on production account information. You cannot test without fully onboarding and creating a production account.
- Limits: the APIs have usage limits; higher limits are requested through USPS API Support.
- Cost: “The APIs are free for valid USPS shipping/mailing use per USPS Terms and Conditions.”
- Address validation: the Addresses API “is only for validating shipping addresses during label generation. Bulk use, batch reporting, or mailing list generation is not permitted.”
That last point catches ERP teams. If your old integration ran nightly address cleanup on the customer master through Web Tools, that job has no direct replacement on the USPS APIs under these terms.
Onboarding, labels and payment
USPS’s developer portal describes onboarding through the USPS Customer Onboarding Portal: create or sign in to a business account, create an app, and use the app’s consumer key and consumer secret as the OAuth client ID and secret. USPS’s own API examples on GitHub add that “You will need a valid customer registration ID (CRID) and mailer ID (MID) in order to get an access token,” obtained from your Business Customer Gateway administrator.
Labels are the biggest structural change, and they are not in the default product: the USPS getting started page says the Labels APIs require “additional approval and configuration” and lists the APIs included by default, among them Addresses, Domestic and International Pricing, Locations, Service Standards and Tracking. The USPS Labels API needs an Enterprise Payment Account (or a permit account), and each label call carries a payment authorization token in an X-Payment-Authorization-Token header, obtained from a separate payments endpoint. The getting started page adds that the Labels APIs require you “to be enrolled in USPS Ship for both outbound and return labels and have an Enterprise Payment Account.” An ERP that used to send one request per label now makes at least three kinds of call: an OAuth token, a payment authorization token, and the label itself.
Tracking versions differ
USPS publishes two tracking versions. The GitHub documentation says revision 2 (v3r2) directs you to “the new, modernized version of the Product Tracking platform”, and that channels such as USPS.com and Informed Delivery that still use v3 “will have some different messaging, information and sometimes events than v3r2.” If customer service compares your portal with USPS.com, they may see different wording for the same package. Map events by code, not by text.
If a third-party platform calls USPS for you
Many ERPs and WMSs reach USPS through a shipping add-on or a multi-carrier platform rather than direct calls. USPS’s Web Tools page tells those users to ask the provider to confirm it is moving to the USPS APIs, to look for references to “modernized USPS APIs” or “USPS REST APIs”, and to check for endpoints beginning with https://apis.usps.com. Get the answer in writing, with the version your account runs, and confirm that the provider’s labels draw on your own payment account if that is how you are billed.
USPS also updated its rulebook. Postal Bulletin 22699 revised the International Mail Manual effective April 2, 2026, stating that “the Web Tools® Application Programming Interface (API) platform has been replaced with the USPS® API platform.”
FedEx: Web Services is being retired in phases
FedEx is retiring its legacy Web Services in stages announced on the FedEx Developer Portal. Support moved to maintenance-only on July 1, 2026, the Web Services test environment closes on December 31, 2026, and the Web Services Track WSDL is disabled on February 1, 2027, with further retirement dates still to be announced.
The four FedEx notices, in order
The FedEx Developer Portal announcements page carries the whole sequence:
- January 30, 2026, “Migrate to FedEx APIs: Act by Your Deadline.” “FedEx Web Services will retire soon and no longer be supported.” It set two deadlines: Compatible Providers to “Complete upgrades by March 31, 2026”, and customers to “Complete migration by June 1, 2026.”
- June 4, 2026, “Migrate to FedEx APIs: Act now.” “Starting July 1, 2026: FedEx Web Services moves to maintenance-only support.” “Existing Web Services integrations may continue operating temporarily for existing customers,” but “General Web Services troubleshooting and development support will end.”
- October 30, 2026, test environment retirement. “The FedEx Web Services (WS) test environment will be fully disabled on December 31, 2026. After that date, providers and customers will no longer be able to test Web Services integrations.”
- November 16, 2026, WS Track. FedEx will “disable the FedEx Web Services Track WSDL on February 1, 2027,” and “Additional Web Services retirement updates and schedules will be communicated separately.” (FedEx notice)
June 1 or July 1? Both, for different things
You will find both June 1 and July 1, 2026 quoted as “the” FedEx Web Services date. Both appear in FedEx’s own notices, for different things. June 1, 2026 was the deadline FedEx gave customers to complete migration. July 1, 2026 was the date support became maintenance-only. Neither was a published shutdown of all Web Services calls; the only production shutdown dated on FedEx’s portal so far is WS Track on February 1, 2027.
For planning, the date that matters most is December 31, 2026. After it, you cannot test a Web Services integration at all, so any change to one, even a certificate renewal or a new service code, goes to production untested.
What FedEx APIs change for an integration
FedEx’s authorization documentation lists three OAuth grant types: client_credentials for customers integrating directly, csp_credentials for “FedEx Internal, Compatible, and Integrator customers” (who add a child key and secret), and client_pc_credentials for proprietary parent-child setups. “The access token expires in an hour,” so your integration must request and cache tokens rather than authenticate on every call.
What FedEx asks of each kind of customer for WS Track
The November 16 notice splits the work by role. Customers connecting directly should work with their development team to move from WS Track to the FedEx Track API. Customers using a third-party product should “Contact your software provider immediately to confirm that your tracking functionality utilizes FedEx APIs” and apply any upgrades the provider specifies. Solution providers must move “any customers still relying on legacy WS Track” to their API-enabled version before February 1, 2027.
Tracking is often the forgotten call in a FedEx migration. Labels and rates get attention because a failure stops shipping, while a nightly tracking sweep or a customer service lookup can keep calling WS Track for months after the shipping station has moved. Search your code and job schedules for the Track WSDL specifically.
If you ship through a FedEx Compatible Provider, the June 4 notice says “your provider will manage the migration” and points you to them for credentials and multifactor authentication requirements. If your ERP calls FedEx directly, the work is yours.
UPS: OAuth tokens and versioned endpoints
UPS’s current API definitions require an OAuth 2.0 bearer token on every call, and they mark older endpoint versions as deprecated. The UPS pages we could read give no cutoff date for older access methods, so the safe assumption is that any integration not using OAuth and the current versions needs work now.
The UPS developer site refused automated access when we checked, so this section relies on two UPS-owned sources: the UPS Learning Center OAuth migration page and the API definitions UPS publishes on GitHub. Check developer.ups.com and your account’s app page for any notice that applies to you.
Two token flows for two kinds of integration
UPS’s definitions describe two flows. The Client Credentials flow is for when “the integration owner is also the UPS shipper”: your ERP calls UPS for your own accounts. The token endpoint is /security/v1/oauth/token, with a Customer Integration Environment at wwwcie.ups.com and production at onlinetools.ups.com, and an optional x-merchant-id header for the 6-digit UPS account number.
The Authorization Code flow is for software that serves other shippers: “your application will not have access to your customer’s UPS login credentials,” so each customer signs in with their own UPS login and your application receives an access token and a refresh token. A 3PL whose WMS ships on clients’ UPS accounts is usually in this second group.
Headers and versions
Each Track API call requires a transId header (“An identifier unique to the request”) and a transactionSrc header (“Identifies the client/source application that is calling”). Log both; they are what UPS support asks for when a call misbehaves.
The Shipping and Rating definitions list v2409 as the only valid version for their current paths and mark the paths that take older versions (v1 through v2205) as deprecated. A deprecated path still answers today, but it is the next thing to break, so move to the current version while you are changing authentication anyway.
UPS’s own applications need an administrator to opt in
For UPS Customer Solutions applications, the Learning Center page describes a one-time setup by the company administrator: a UPS.com user account with the shipping account numbers enabled, account associations established on UPS.com, and an opt-in “through new administration screens in each Customer Solutions product.” It also warns that “Changing the password for the user who generated the authorization code will invalidate” it, so the administrator must opt in again after a password change. That warning is the strongest argument for the credential ownership rules later in this note.
What changes inside an ERP or WMS shipping integration
A carrier API migration is not an endpoint swap. Authentication, message format, field names, label output, error codes and tracking events all change, and each one touches a different part of your ERP, WMS or shipping station. List the changes by touchpoint and test each one.
| Touchpoint | What usually changes | What to test |
|---|---|---|
| Authentication | Static keys or user IDs replaced by OAuth tokens that expire; new IDs such as USPS CRID and MID | Token requested, cached, refreshed before expiry, and failure alerting |
| Rating and rate shopping | JSON request shape, service codes, how surcharges are itemized | Same shipment rated old and new; differences explained line by line |
| Label creation | Payment authorization (USPS), new label format options, document references | Label prints on your actual printers and scans at the carrier |
| Void and edit | New endpoints; some edits now need payment authorization (USPS) | Void within the carrier’s window; edited weight re-rated |
| Address validation | Usage restrictions (USPS: label generation only) | Order entry and ship-time validation behave as before; no bulk jobs |
| Tracking | Event codes and wording, summary versus detail, version differences | Every event code you receive maps to a status in your model |
| Pickup and manifests | New pickup and SCAN form endpoints | End-of-day close produces the form your drivers need |
| Errors and limits | New error codes, HTTP 401, 403 and 429 responses, quotas | Each error lands in a queue with a readable message |
Two rows deserve emphasis. Rating changes surface weeks later, on the carrier invoice, if the new request sends different dimensions or service options. Compare a sample of shipments rated through the new API with what the carrier bills, and check billable weight with the dimensional weight calculator when a difference appears.
Tracking changes are quieter and spread further. Every event code the new API returns has to map to a status in whatever model your customer portal, exception rules and ERP order status use. If you also take status from EDI 214 messages or other carriers, the mapping belongs in one place; our comparison of EDI 214 and carrier APIs for shipment status covers how to build that single status model.
Token handling that survives a busy shipping day
OAuth adds a moving part that static keys did not have: the token expires. FedEx says its access token “expires in an hour”, and the USPS and UPS token responses include an expiry value. Token handling that works in a test with one user can fail at 4:30 p.m. with twelve pack stations printing at once.
- Request tokens in one service, not at every pack station or in every ERP session. A shared integration service requests, caches and hands out the token.
- Cache the token with its expiry and refresh it a few minutes before it lapses, using the expiry the carrier returns rather than a hard-coded hour.
- Keep tokens separate per carrier account and app where the carrier scopes them that way, so a problem with one account does not stop every shipment.
- On a 401 response, refresh once and retry once. If the retry fails, stop and alert. Endless retries with a bad secret fill the logs and can run into the carrier’s rate limits.
- Never log the token or the secret. Log the issue time, the expiry and the account it belongs to.
Service codes, reference fields and label output
Your ERP’s ship-via codes map to carrier service codes, and the codes, names and allowed combinations can differ between the old and new API. Keep the mapping as a table owned by transportation, with the ERP code, the carrier code and the test shipment that proved it.
Reference fields need the same care. Purchase order numbers, invoice numbers and department codes that print on labels or appear on carrier invoices often move to differently named fields with their own length limits. Some retail customers’ routing guides require specific references on the parcel label, so a truncated PO number can turn into a compliance chargeback. Test with your longest real values, not short samples.
Label output is the last mile of the change. Confirm that the new API returns the format each station actually prints (a thermal printer language or an image or PDF), at the size and resolution your printers use, and that the printed barcode scans with a handheld. A label that looks right on screen proves nothing about the dock.
How to inventory every carrier touchpoint before you migrate
Before changing code, list every system, job and person that calls a carrier, with the account numbers and credentials each one uses. Migrations fail on the call nobody knew about: a returns portal, a customer service screen, a script on a shared drive.
Places to look:
- ERP shipping module or add-on: rating at order entry, label printing at shipment, tracking on the order screen.
- WMS pack and ship stations: labels, end-of-day manifests, pickup requests.
- TMS or rate shopping service: parcel rates beside LTL quotes.
- Ecommerce platform and returns portal: checkout rates, return labels.
- Customer service tools: tracking lookups, address checks.
- Scheduled jobs and scripts: nightly tracking sweeps, address cleanup, proof-of-delivery downloads.
- Third-party software and Compatible Providers: confirm in writing which API each one now uses.
For each touchpoint, record the carrier, the old API, the new API, the account numbers, the credential holder, the call volume on a peak day and the business owner. The integration inventory worksheet has the columns for this.
Example (fictional). A distributor’s inventory of carrier touchpoints, taken in March 2026.
| Touchpoint | Carrier and old API | New API | Credential holder | Peak-day calls | Owner |
|---|---|---|---|---|---|
| ERP order entry rate quote | UPS, older access method | UPS Rating v2409, client credentials | IT service account | 1,400 | Customer service |
| WMS label print, station 1 to 6 | FedEx Web Services Ship | FedEx Ship API | IT service account | 2,100 | Shipping lead |
| WMS end-of-day close | FedEx Web Services | FedEx API equivalent | IT service account | 1 | Shipping lead |
| Returns portal labels | USPS Web Tools | USPS Labels API with payment token | Unknown (former employee) | 90 | Ecommerce |
| Nightly tracking sweep | All three carriers | Carrier tracking APIs | Script on a shared server | 6,500 | Transportation |
| Customer master address cleanup | USPS Web Tools | None permitted for bulk use | IT | 40,000 monthly | Sales operations |
Two rows changed the plan. The returns portal ran on a former employee’s Web Tools account and had to be re-onboarded under a company account. The address cleanup job had no permitted replacement on the USPS APIs, so the team moved address checks to order entry and label time.
Who should own carrier credentials and tokens
Carrier credentials belong to the company, held in a named service account and a secrets store, with a business owner and an IT owner for each. When credentials sit in one person’s login, a resignation or a password change can stop shipping.
The carriers’ own documents show why. UPS warns that changing the password of the user who generated an authorization code invalidates it. USPS ties access tokens to a CRID and MID managed in the Business Customer Gateway. FedEx Compatible Provider customers may face multifactor authentication requirements set by the provider. Each is a point where one person’s account can stop labels.
| Item | Business owner | Technical owner | Where it lives |
|---|---|---|---|
| Carrier account numbers and contracts | Transportation or logistics manager | IT integration lead | Contract file and ERP carrier setup |
| Developer portal login and apps | IT integration lead | IT integration lead | Shared company identity, not a personal login |
| Client IDs and secrets | IT integration lead | Platform or security team | Secrets store, never in code or spreadsheets |
| USPS CRID, MID and payment account | Finance or shipping manager | IT integration lead | Business Customer Gateway, recorded in the runbook |
| Token caching and refresh | Not applicable | Integration developer | Integration service configuration |
| Renewal and rotation calendar | IT integration lead | Platform team | Shared calendar with reminders |
In our engagements we write these owners into the requirements before building anything, and we hand them over in the integration runbook. That keeps the credential map from living only in the head of whoever did the migration.
When a credential has to move from a person to the company, do it as a planned change, not during an outage:
- Create the company identity on the carrier’s developer portal and record who administers it.
- Create a new app and new secrets under that identity; do not copy the old person’s secrets.
- Store the secrets in the secrets store and point the integration at them in the test environment first.
- Switch production during a quiet window, watch authentication errors for a full day, then revoke the old credentials.
- Update the runbook and the renewal calendar, and tell the carrier account manager who the new contact is.
A regression test plan for labels, rates and tracking
Test the new carrier integration with real shipment profiles, compare results against the old integration while it still answers, and prove that labels scan at the carrier before cutover. Do the Web Services comparison for FedEx before December 31, 2026, when its test environment closes.
Build the test set from your own shipping history, not from the carrier’s sample requests:
- Pick shipment profiles that cover your real mix: each service level you use, residential and commercial, single and multi-piece, international if you ship it, hazardous materials if applicable, and your heaviest and most oversized items.
- Rate each profile through old and new APIs where both still work, and explain every difference in total and by surcharge.
- Print each label on every printer model in the building, in the format the station uses, and scan it with a handheld.
- Send labels to the carrier for an acceptance check where the carrier offers one, and keep the evidence.
- Void and reprint within the carrier’s window, and confirm the voided label does not bill.
- Track a set of live shipments through every event your portal shows, and check that each event code maps to your status model.
- Break things on purpose: an expired token, a wrong secret, a rate limit response, an invalid address, a carrier outage. Each should produce a clear message in the right queue.
- Run an end-of-day close with manifests and pickup requests, and confirm the driver can accept the paperwork.
| Test | Pass condition | Evidence kept |
|---|---|---|
| Token refresh | No failed calls across three token lifetimes under load | Log extract with token timestamps |
| Rate parity | Every difference explained by a documented carrier change | Comparison sheet signed by transportation |
| Label scan | Every label scans on site and at the carrier | Photos or scan logs per printer model |
| Tracking map | 100% of event codes received in the test window mapped | Mapping table with unmapped codes listed |
| Error handling | Every forced failure reaches its queue with a readable message | Screenshots of queue entries |
The UAT test plan template gives you the structure for scripts, results and sign-off, and the go-live readiness checklist covers the gates for cutover day.
How to cut over without stopping the dock
Switch one carrier and one touchpoint at a time behind a setting you can flip back, starting with calls that cannot stop freight. Keep a rollback path only where the carrier still supports the old platform: there is none for USPS Web Tools, and for FedEx Web Services each rollback path ends on the date its WSDL is disabled.
A sequence that limits the damage of a surprise:
- Put the API choice behind configuration, per carrier, per touchpoint and, where you can, per pack station. Flipping a setting is faster and safer than redeploying code at 5 p.m.
- Move read-only calls first. Tracking lookups and address checks at order entry fail softly; a wrong answer there does not leave a pallet on the dock.
- Run rating in parallel. For a week, call both APIs where both still answer, use the old result, and log every difference. Explain each one before switching.
- Move label printing one station at a time, early in the week and away from carrier cutoff times. Keep one station on the old path until the new one has printed a full day.
- Move end-of-day close, manifests and pickups last, after labels are stable, because they bundle the whole day’s shipments.
- Remove the old path deliberately. Once the carrier’s next dated step passes, delete the old code and credentials so nobody flips back to a dead platform.
Timing matters too. The FedEx Web Services test environment closes on December 31, 2026, in the middle of many shippers’ peak season. If your FedEx migration is not finished, plan the remaining Web Services testing for early December rather than discovering in January that it can no longer be done.
Example (fictional). An industrial distributor runs UPS rating in parallel for five days before switching its ERP order entry quotes. Of 500 quoted shipments, 482 match to the cent. Fourteen differ because the new request sends package dimensions from the item master while the old one sent none, so the new API applies dimensional weight the old quotes ignored. Four differ because the residential flag was not mapped.
The team fixes the residential mapping, confirms with transportation that the dimensional weight in the new quotes matches what the carrier invoices, and switches order entry the following Monday. The fourteen “differences” turn out to be the old integration under-quoting all along.
Monitoring carrier integrations after cutover
After cutover, watch token failures, error rates by code, label failures by reason, rate differences against invoices and tracking freshness, each with a named owner and an alert threshold. Carrier APIs keep changing after a migration, so the monitoring is permanent, not part of the project.
Watch these signals daily for the first weeks and weekly after that:
- Authentication: count of 401 and 403 responses per carrier and per account. A spike usually means an expired secret or a changed password.
- Throttling: count of 429 responses. Rising volume, or a sweep job scheduled at the same time as shipping, can hit a quota.
- Label failures: by reason code, with the order numbers, so shipping can work around them before the truck leaves.
- Rate variance: quoted rate against invoiced charge, by service and surcharge, reviewed when invoices arrive.
- Tracking freshness: the share of open shipments with no event in a set number of hours. A silent feed looks the same as a quiet day unless you measure it.
- Carrier notices: someone subscribes to each carrier’s developer announcements and reads them. The FedEx WS Track date of February 1, 2027 was announced on November 16, 2026, under three months ahead.
Example (fictional). The alert rules a distributor’s integration team wrote after its carrier cutover.
| Signal | Alert when | Goes to |
|---|---|---|
| 401 or 403 responses | More than 5 in 10 minutes for one carrier account | IT integration lead |
| 429 responses | Any during shipping hours | IT integration lead |
| Label failures | More than 2% of labels in an hour | Shipping lead |
| Quote to invoice variance | Weekly total above an agreed dollar amount | Transportation manager |
| Tracking freshness | Open shipments with no event for 24 hours above 3% | Customer service supervisor |
The thresholds are the team’s own starting values, reviewed after the first month against what actually happened.
If tracking data feeds a customer portal or exception workflow, the same monitoring belongs in your shipment tracking software, so a carrier problem shows up as a visible gap rather than a stale status.
What to do next
Start with the inventory, because it tells you how much work is left and which date applies to each touchpoint. Then fix credential ownership, build the test set from real shipments, and put monitoring in place before the next carrier notice arrives.
A short checklist:
- List every touchpoint that calls USPS, FedEx or UPS, with the API it uses today.
- Confirm nothing still calls USPS Web Tools.
- For FedEx Web Services, finish testing before December 31, 2026, and move WS Track calls before February 1, 2027.
- For UPS, confirm every call uses an OAuth token and the current endpoint versions; check your developer account for notices.
- Move all carrier credentials to company-owned accounts and a secrets store, with named owners.
- Build regression tests from your own shipment profiles, including forced failures.
- Map every tracking event code to your status model.
- Set up monitoring and assign someone to read carrier developer announcements.
If you want help with the inventory, the credential map or the interface work itself, our ERP, WMS and carrier integration service builds and documents carrier interfaces, and you can contact us to talk through your list.
