Roaming testing: types, methods, KPIs and how to automate it

Roaming testing checks that a subscriber can register, use data, send SMS and make calls on a partner network abroad, and that every event is billed correctly between the two operators. It covers formal GSMA test stages run when a roaming relationship launches: IREG functional tests and TADIG billing tests. It also covers the ongoing checks that keep roaming working afterward: end-to-end service tests, steering of roaming verification and quality benchmarking. This guide explains each type, how operators traditionally run them, and how to automate roaming testing with forced network selection, a roaming matrix and clear KPIs.

Who needs roaming testing

• Mobile network operators launching or maintaining bilateral roaming agreements, or connecting through a roaming hub.
• MVNOs and travel eSIM brands whose service depends on a host operator’s or aggregator’s partner list. They can’t fix a partner’s configuration, but they need to know when it breaks (see travel eSIM quality monitoring).
• IoT connectivity providers selling one global SIM across dozens of countries, often on LTE-M or NB-IoT, where roaming support differs by partner.
• Automotive and device OEMs whose products ship with connectivity that has to work in every market the product is sold.

Types of roaming testing

Type	What it proves	When it runs	Reference
IREG end-to-end functional	Registration, calls, SMS and data work between the two networks	Before commercial launch; after major changes	GSMA IR.23 and the IR.24/IR.35/IR.38/IR.25/NG.143 test specs
TADIG billing	Test usage produces correct roaming billing records	After IREG tests, before launch	GSMA TADIG/IDS documents (TAP, RAP)
End-to-end service	Real customer journeys work with real profiles and devices	Continuously after launch	Your own test plan
Steering of roaming	Devices land on preferred partners; steering doesn’t strand anyone	After steering changes; periodically	GSMA IR.73; 3GPP TS 23.122
QoS benchmarking	Latency, throughput and call setup are acceptable on each partner	Periodically; when complaints rise	Your service targets

IREG functional testing

GSMA’s IR.23 (Organisation of GSM International Roaming Tests, v8.0, January 2025) recommends a three-stage approach:

1. Interface self-certification for MAP, CAP, IP, Diameter, GTP and HTTP/2.
2. Exchange of numbering, addressing and operations data.
3. End-to-end functional and capability tests with each partner.

The stage 3 tests are specified per technology: IR.24 for basic inter-PLMN roaming, IR.35 for GPRS, IR.38 for LTE and EPC, IR.25 for VoLTE and NG.143 for 5G SA. The tests are written in one direction (subscriber of network A visiting network B) and repeated with the roles swapped for a bilateral agreement. IR.38 describes the classic split of work: the home network issues pre-programmed test SIMs and provisions its HSS, the visited network runs the tests, and the two discuss results.

TADIG billing testing

Roaming usage is billed between operators through Transferred Account Procedure (TAP) files, with the Returned Account Procedure (RAP) for rejecting bad records. TADIG testing checks that the test calls, SMS and data sessions from IREG testing appear in the visited network’s TAP files with the right identifiers, timestamps and charges, and that the home network accepts them. The differences are covered in IREG vs TADIG testing.

End-to-end service testing

Formal tests prove that two networks interwork once. End-to-end service tests prove that today’s customers on today’s devices actually get service. They use the same journeys a subscriber would: attach, open a data session, resolve DNS, load a page, send an SMS, make a call. They run continuously, because partner configurations change without notice.

Steering of roaming verification

Home operators steer roamers toward preferred partners. In 2G, 3G and 4G this is commonly done by rejecting registration attempts on non-preferred networks, or by updating the preferred network list on the SIM over the air. GSMA’s IR.73 sets out implementation guidelines. It notes that a handset in automatic mode makes up to four location update attempts (five for GMM procedures), 15 seconds apart, so reject-based steering can delay service by about a minute. It also limits how many attempts the home network may reject, so a customer isn’t left without service. In 5G, 3GPP TS 23.122 defines a control-plane mechanism in which the home network sends a protected preferred-network list to the device through the visited network during registration.

Steering tests check that devices end up on the preferred partner, how long that takes, and that no device gets stuck when the preferred network has no coverage.

QoS benchmarking

Two partners that both “work” can deliver very different experiences. Benchmarking compares latency, throughput, DNS time, call setup time and signal quality across partners in the same market. Data roaming is commonly home-routed, with traffic tunneled back to the home network’s gateway, so round-trip time to the home country is a key metric.

How roaming testing is done traditionally

The traditional workflow is manual and logistics-heavy:

• Physical test SIMs are programmed by the home operator and couriered to each partner or test vendor.
• Partner coordination happens by email: scheduling, provisioning test IMSIs, exchanging results in spreadsheets.
• Field engineers or local test vendors run calls and data sessions with test handsets, sometimes in drive tests.
• Probe-based systems with SIM arrays in many countries automate parts of the process, but each new SIM still has to be shipped or remotely connected.

It works for a launch, but it’s slow to repeat. Re-testing 200 partners after a core network change, or checking a single partner at 3 a.m. after a complaint, doesn’t fit a courier-based process.

Remote and automated roaming testing

Three changes make roaming testing faster and repeatable:

1. Devices already in the visited markets. Modems or phones stay installed in the countries you care about and are driven remotely.
2. eSIM instead of shipped SIMs. The home profile is downloaded over the air to an eUICC in the visited country. A new partner or a new product can be tested the same day.
3. Forced network selection. The device is told which partner PLMN to use, so one location can test every partner in that country in turn.

Add scheduling, pass criteria and a results store, and roaming testing becomes a monitoring system rather than a project. The per-partner checks themselves are listed in the eSIM testing checklist.

Forcing a PLMN and RAT during tests

Most cellular modems use the standard 3GPP TS 27.007 network selection command:

• AT+COPS=? scans and lists the networks the device can see, with their status (available, current, forbidden).
• AT+COPS=1,2,"<MCCMNC>" selects a specific network manually using its numeric MCC-MNC.
• AT+COPS=1,2,"<MCCMNC>",<AcT> also pins the access technology, for example E-UTRAN or NB-IoT.
• AT+COPS=0 returns to automatic selection.

Things to watch:

• Forbidden lists. A reject with cause #11 (PLMN not allowed) adds the network to the SIM’s forbidden list, which affects later automatic selection. Reset state between tests.
• Reject causes are the real result. A failed manual selection should be read from the NAS signaling. “Roaming not allowed in this tracking area” and “network failure” point to different owners. See attach reject causes.
• Forced vs automatic. Forced selection tests whether a partner works. Automatic selection tests steering. Run both.
• RAT locks. Many partners support LTE but not LTE-M, NB-IoT or 5G SA roaming. Test each technology you sell explicitly.

Building a roaming matrix

A roaming matrix puts home profiles or products on one axis and visited networks on the other. Each cell shows the latest verdict per service:

Visited network	Attach	Data	SMS	Voice	RAT seen	Last tested
Partner A (country 1)	Pass	Pass	Pass	Pass	LTE	Today
Partner B (country 1)	Pass	Fail (DNS)	Pass	n/a	LTE	Today
Partner C (country 2)	Reject #13	n/a	n/a	n/a	n/a	Yesterday

A matrix works when it’s fed automatically and each cell links to the underlying run, with logs and signaling, so a red cell leads straight to the cause.

Roaming KPIs to track

KPI	What it shows
Registration success rate	Whether the partner accepts your subscribers at all
Time to register	Steering, retries and signaling delays
Reject cause distribution	Agreement, provisioning or steering problems
Data session success rate	APN, DNS and IPX/GRX configuration
DNS resolution time and success	Breakout and resolver health
Round-trip latency to home	Home routing cost and IPX path quality
Downlink and uplink throughput	Partner capacity and any throttling
SMS delivery success and time	SMS interworking in both directions
Call setup success and time	Voice and VoLTE roaming readiness
Steering compliance	Share of attaches on the preferred partner
Charging accuracy	Test usage matched to billing records

Cadence: when to test

• At launch. IREG functional tests, then TADIG billing tests, then a set of end-to-end service tests with real products.
• After changes. IR.23 lists triggers including new services, numbering or addressing changes, a new IPX proxy (STP, DRA, SEPP), a new vendor for an existing node and major software releases.
• Periodically. IR.23 describes the visited operator or roaming hub placing test calls weekly with each partner’s test SIMs, per type of network node. For travel eSIM and IoT services, daily smoke tests on key partners are common.
• On demand. When complaints rise for a country, test immediately on the specific partner.

Where SimCheck.ai fits

SimCheck.ai runs roaming checks on real modems with real eSIM profiles, on the SimCheck network or on your own edge devices anywhere. Any Quick Test can force an operator by PLMN and force the radio technology, so one site can step through every partner in a market. PLMN Hunt keeps scanning for a target network and attaches when it appears. The Roaming matrix report rolls results into home operator × visited operator, with drill-down to each run’s logs and decoded signaling.

Automate your roaming tests

If your roaming coverage is still verified by courier and spreadsheet, start with the partners that matter most and put them on a schedule. See how roaming testing on SimCheck.ai works, or read IREG vs TADIG testing for the formal launch stages.