Measuring the network is the easy part.
Owning the evidence is harder.
Most regulators already have data — operator returns, drive-test campaigns, perhaps a consumer speed-test app. What is harder to assemble is a source that is independent of the parties being regulated, deep enough to locate a fault rather than just flag one, and yours to publish. These are the gaps we are usually asked to close.
What a measurement programme often cannot answer yet
Operator returns are difficult to verify independently
Not because operators are acting in bad faith, but because there is rarely a neutral dataset to check them against. Drive tests fill part of the gap, though they are expensive, infrequent, and measure a car on a road rather than a household at eight in the evening.
Knowing a line is slow is not knowing why
Is the loss in the last mile, the ISP core, or international transit? Is it congestion or a physical fault? Without transport-layer isolation, a complaint is hard to turn into a finding an operator will accept.
Throughput and experience are not the same thing
Citizens do not experience megabits. They experience a video that takes four seconds to start and stalls twice, or a government portal that is slow to become usable. Measuring on the real services closes the distance between the figure and what people report.
Data residency is often settled late
Many commercial measurement tools hold citizen data abroad, with limited control over retention, resale or cross-border transfer. It is a straightforward question to answer up front, and an awkward one to answer after a programme is live.
What tends to slow a programme once it is running
Investigations move at the vendor's pace
A targeted audit, a spam sweep or a new threshold typically queues behind a release cycle. When the configuration sits with the supplier rather than the regulator, the programme responds more slowly than the market does.
Complaints without a closed loop
Without an official channel there is no reference ID, no operator deadline and no audit trail — so a citizen has no way to see that anything happened. Closing that loop is one of the most visible things a regulator can do.
One measurement engine.
Six ways to collect. One portal.
Everything below runs the same engine, the same test scripts, the same KPI definitions and the same analytics portal. What changes is only how the data gets collected — and you can start with one collection agent and switch on the others later, on the same contract, without re-procuring anything.
Collection changes. The engine and the portal never do.
This is the whole architecture in one line — and the reason a fixed-broadband programme and a mobile crowdsource programme are the same procurement.
One measurement engine
- Identical KPI definitions across every agent
- ITU-T aligned methodology, published
- Your own test scripts, no device lock-in
- Network and application testing
One analytics portal
- Every source in a single view
- Role-based views per team
- Scheduled reports & open API
- On-premise or region-pinned
Six offerings — each one a complete story
Every section below carries its own dashboards, its own demo video and its own live customer. Read the one that matches your programme.
The argument that runs through every section
Hardware probes give you compliance-grade evidence
Wired to the router at a known address with a verified plan, running the identical script every minute. Narrow coverage, high certainty — the layer a finding is usually built on.
Crowdsource gives you the whole country
Every district and building, not only the panel addresses. Broad coverage, contextual certainty. This is what tells you where to look, and corroborates a probe result when an operator pushes back.
A national picture on day one — before anything is deployed.
We already run 5GMARK, a consumer measurement app with its own global user base. In most markets that means hundreds of millions of QoE records have already been collected — across every operator, every region, indoor and outdoor. There is nothing to install and nobody to recruit. It is the fastest thing on this site to switch on, and it is often a sensible place to start.
Someone else already did the fieldwork
The 5GMARK app has been in consumer hands since the 4GMARK days in 2012. Every test it has ever run is geo-tagged, operator-identified and classified indoor or outdoor through a 17-step model that is 95% accurate. When a regulator engages us, that history becomes their baseline immediately — the same KPI definitions, the same engine, and the same portal they will use for everything else.
Reach nothing else can match
- District and wilayat benchmarking before a probe ships
- Shows where a fixed panel or a drive-test route should go
- An independent second source when an operator disputes a finding
- Wi-Fi tests reveal home broadband nationwide, not only panel addresses
And we will say so in the room
- It is not wired to the router — home Wi-Fi sits in the path
- Plan and operator are inferred, not verified from a bill
- Sampling follows where people are, not a quota matrix
- Use it to find the problem; use the probe to cite it
The data, in the portal, on day one
The proof for this offering
ARCEP — crowdsourced data underwriting a public national portal
The challenge
- Operators self-reported compliance with rural coverage targets; ARCEP had no field-validated data to challenge them.
- Orange, SFR, Bouygues and Free each used different methodologies — no comparable dataset existed.
- The open-data mandate for monreseaumobile.com required real-user data ARCEP could control and publish.
What we did
- The free 5GMARK app collecting continuous QoE on file sharing, web and video — all four operators, iOS and Android simultaneously.
- Anti-bias controls stripping intentional manipulation and device/operator skew before reporting.
- A public data pipeline exporting CSV and shapefile straight into ARCEP's own portal.
The outcome
- More than 80% of ARCEP's public QoE portal runs on Mozark-collected data.
- An annual 360° national mobile benchmarking report, trusted by government and public alike.
- Local governments, enterprises and citizens can access the quality metrics directly.
Data specifications
Your app, your name, on both stores — and a complaint loop that closes.
A white-label speed-test app published under your developer account, carrying your branding, in your languages. Citizens test, see their own history, and — where you want it — report a problem that gets a reference ID, an operator deadline and a status they can follow. It is the offering that gives a regulator its own channel to citizens, rather than only a view of them.
The whole citizen journey, three minutes
Language selection, testing, coverage map, history — and the result arriving in the national dashboard.
The app as a citizen sees it: multilingual onboarding, speed and video testing, coverage mapping, test history, and the feedback path into the regulator's reporting flow.
Four things that make it regulator-grade
One SDK behind every surface
The app, the browser tester and the probe all run the same measurement SDK. A result is a result, wherever it came from.
Anti-fraud before it counts
Device integrity, geo-verification, IP and operator validation, and a deduplication engine that filters bot and spoofed submissions.
Reconfigure without resubmitting
Test servers, protocols, thresholds, geo-fencing and branding all change from a no-code console. No app store review, no waiting.
Accessible and multilingual
Screen-reader support and any language set — a hard requirement in most public procurements, and usually the thing that eliminates competitors.
The citizen side feeding the regulator side
Two regulators, two different problems
TRAI — independent QoS measurement at billion-subscriber scale
The challenge
- Every quality figure came from the operators themselves, leaving TRAI without a neutral baseline to enforce against.
- 22 circles and four major operators meant millions of ITU-T compliant tests monthly, surfaced as something actionable.
- Under Digital India, measurement had to be in citizens' hands with transparent real-time reporting.
What we did
- MySpeed, MyCall and a web tester, all on a single measurement SDK.
- Bhashini integration for 22+ Indian languages at national scale.
- Cloud-native backend for millions of monthly tests with real-time GIS dashboards and anti-fraud controls.
The outcome
- Millions of ITU-compliant tests monthly with zero third-party data dependency.
- India's first browser-based national speed test, deployed at scale.
- Six years live, and Mozark now sits in ITU-T Study Group 12 helping write the standards it implements.
FCC — the first federally accredited consumer broadband app
The challenge
- One subscriber in a block meant the whole block counted as covered — billions in subsidies allocated on maps the FCC could not verify.
- Citizens had no official app: no reference ID, no operator deadline, no audit trail.
- Every commercial measurement tool failed federal security accreditation.
What we did
- One-tap complaint that runs automated diagnostics at the same time, with full traceability.
- Geo-locked, federally accredited security built into the measurement engine itself.
- Remote reconfiguration of servers, protocols and parameters with no app resubmission.
The outcome
- The first federally-compliant consumer broadband complaint app at US national scale.
- A time-bound operator validation loop closing complaint → diagnosis → resolution.
- The FCC can verify network quality without relying on operator self-reporting.
Data specifications
One click in a browser. No app, no plugin, no download.
The citizen app reaches people with smartphones who are willing to install something. The web tester reaches everyone else — desktop users, laptop users, and above all the fixed-broadband households you cannot measure any other way without shipping hardware. It runs the identical engine, so the results sit in the same dataset.
What a citizen actually does
The browser test end to end, and the geo-tagged result landing in the regulator's national dashboard alongside app and probe results.
Four reasons regulators keep asking for it
Reaches non-smartphone users
A meaningful share of any population will not install an app. The browser is the only route to them, and to anyone at a work machine.
The fixed-broadband proxy
A desktop on Ethernet or home Wi-Fi is a fixed-line measurement. It is the cheapest fixed-broadband signal available to a regulator.
Same engine, same numbers
DL, UL, latency, jitter and packet loss on the identical ITU-T compliant backend — not a lighter, different measurement.
Embeddable in your own portal
Drop it into the regulator's existing website. Every test geo-tagged and fed into the national dashboard and public reports.
The proof for this offering
TRAI MySpeed Web — a national speed test with nothing to install
The challenge
- The mobile apps could not reach desktop and laptop users, which is where fixed-broadband experience actually lives.
- App-install friction excluded a large part of the population from contributing evidence at all.
What we did
- A browser-based tester on the same ITU-T compliant backend as MySpeed — one click, no plugin.
- Full QoS metric set reported per operator and per circle, geo-tagged on every run.
The outcome
- India's first browser-based national speed test, deployed at scale.
- Fixed-line and desktop coverage feeding the same national dashboard as the app data.
The measurement engine, inside something you already ship.
Not every measurement should require a new app or a new box. EdgeSDK drops the same engine into an operator's existing app or into CPE router firmware, where it runs quietly in the background under the consent terms you set. Nobody has to open an app for a measurement to happen, so the sample reflects ordinary households rather than only the people motivated to run a test.
Three deployments that only work this way
The in-home Wi-Fi layer
Embedded in router firmware, it measures the hop between the router and the devices in the house — the layer a wired probe deliberately excludes. Run both and you can finally say whether a bad experience is the ISP or the household's own Wi-Fi.
An operator's own subscriber base
An ISP embeds the engine in its existing customer app and gets continuous QoE from its real subscribers, on the same KPI definitions the regulator uses. Fewer arguments about methodology later.
Devices shipped at scale
Handset and CPE manufacturers embed the engine before shipment. Long-term embedded monitoring with no per-device deployment cost and no user recruitment.
Four steps, then it disappears
Native Android, native iOS or ReactJS. A working integration typically takes weeks, not quarters.
Test cadence, endpoints, thresholds and consent behaviour all set from the same no-code console as everything else.
Passive or active sampling in the background, with no user action required and consent handled at onboarding.
Results land in the same portal, on the same KPI definitions, next to app, browser and probe data.
A known address, a verified plan, every single minute, all year.
This is the layer that turns an indication into something citable. A fanless probe wired to the router by Ethernet, running the identical test script every minute at a household whose operator and advertised speed have been verified from a bill. Narrow coverage, absolute certainty — and the only layer that can isolate whether a fault sits in the last mile, the ISP core or international transit.
Two tiers, one platform
Lite goes into homes by the hundred. Pro handles 5G, multi-operator sites and doubles as the drive-test unit.
Built for national fixed-broadband panels
Raspberry Pi 5 in a fanless case, powered by USB-C or PoE+. The panelist plugs in power and Ethernet and does nothing else — everything after that is remote.
- Compute
- BCM2712 2.4 GHz quad-core ARM Cortex-A76, fanless
- Connectivity
- Gigabit Ethernet with PoE+ · dual-band Wi-Fi 5
- Cellular
- Optional USB modem add-on for fixed-wireless lines
- Location
- USB GPS dongle for precise site tagging
- Power
- 5 V/5 A USB-C with Power Delivery, or PoE+ HAT
When one SIM and one radio is not enough
Intel Elkhart Lake with four M.2 slots and four SIMs, Wi-Fi 7 and 10G SFP+, rated −20 °C to +60 °C. It monitors every operator simultaneously from one device — and it is the same unit that goes in the drive-test vehicle.
- Compute
- Intel® Elkhart Lake ATOM X6413E embedded SoC
- Cellular
- 4 × M.2 · 4 × micro-SIM · 5G Sub-6 NR SA/NSA + 4G
- Wi-Fi
- Intel BE200 Wi-Fi 7 (MLO) + Bluetooth 5.4
- Ethernet
- 10G SFP+ · 10/100/1000/2.5GBase-T
- Location
- Integrated M.2 GNSS — no dongle
- Power
- Wide range 9–36 V DC, so it runs off a vehicle supply
| EdgeX Lite | EdgeX Pro | |
|---|---|---|
| Typical use | Fixed-line & Wi-Fi panels at scale | 5G NR, DAS/SMO venues, multi-operator, drive test |
| Compute | BCM2712 2.4 GHz quad-core ARM Cortex-A76, fanless | Intel Elkhart Lake ATOM X6413E embedded SoC |
| Cellular | Optional USB modem add-on | 4 × M.2 · 4 × micro-SIM · 5G Sub-6 NR SA/NSA + 4G |
| Wi-Fi | Dual-band 802.11ac (Wi-Fi 5) + BT 5.0 | Intel BE200 Wi-Fi 7 (MLO) + BT 5.4 |
| Ethernet | Gigabit with PoE+ support | 10G SFP+ · 10/100/1000/2.5GBase-T |
| Location | USB GPS dongle | Integrated M.2 GNSS |
| Power | 5 V/5 A USB-C PD · PoE+ HAT | Wide range 9–36 V DC · 60 W adapter |
| Environment | Indoor, fanless, silent | −20 °C to +60 °C · 95% RH · 150×160×70 mm |
Full component-level data sheets are downloadable at the bottom of this section.
Three layers a speed-test appliance cannot see
- Download and upload throughput, multi-thread
- Throughput consistency — % of tests ≥ subscribed plan speed
- Latency idle and under load (bufferbloat)
- Jitter, packet loss, traceroute
- DNS resolution and TCP connect time
- UDP throughput, IPv6 reachability, uptime
- Segment isolation — last mile vs ISP core vs peering
- One-way delay and directional packet loss
- DSCP / QoS marking preservation
- Loss-pattern classification — congestion vs physical fault
This is the layer that turns "the line is slow" into "the loss is in your core, between these two nodes" — the difference between a complaint and a finding.
Video streaming — MOS out of 5
- A single quality score per service and per operator
- How long the video takes to start playing
- Freezes — how often, and for how long
- Picture quality actually delivered: 480p, 720p, 1080p
- Run on YouTube, Netflix, TikTok, Instagram, Snapchat
Web browsing — score 0–100
- One score per website, per operator
- Time until the first content appears (FCP)
- Time until the page is genuinely usable (TTI)
- Visual stability while loading (CLS)
- Run on your government, banking and news sites
- RTT and packet loss to Google, Meta, Netflix, TikTok and major CDN nodes
- Domestic vs international split, including traffic exchanged at the national IX
- Hop-by-hop traceroute to locate degradation per node
- Unbiased server selection so results are not inflated by CDN caching
Probe monitoring — the fleet at a glance
A dedicated module inside the portal. Arrow through the seven screens; each carries the one thing that matters about it.
How many probes — and why
Panel size is a statistical decision, not a price point. Drag the slider for a market of roughly 600,000 fixed lines.
300 gives national and per-operator results at 95% confidence, the largest region reportable on its own, and the remainder reported as clusters.
Fully managed — no field effort from you
Recruit to quota, never to convenience
Recruitment sits with us rather than with the operators, so the panel's neutrality is straightforward to demonstrate if it is ever questioned. Candidates come from online panels, CATI outreach into under-represented regions, and snowball referral from panelists already onboarded. Every candidate gets a verification call and a bill check before any hardware ships.
- Quota matrix agreed with you before recruitment opens — region, operator, segment, plan tier and household profile
- Operator, plan and advertised speed verified from a recent bill, not self-declared
- Written informed consent, withdrawal at any time, no PII captured
- Rolling recruitment until every cell is closed; fill rate reported weekly
What you see
Zero site visits, anywhere in the country
The probe ships pre-configured in a welcome kit with a one-page guide. The panelist connects power and Ethernet — that is the entire installation — and a panel executive follows up on every new site. Remote commissioning and validation complete inside two hours, including remote regions no engineer would reach economically.
- Wired to the router by Ethernet, never Wi-Fi — the probe measures the ISP line, not the household's router
- Remote commissioning, firmware and test scripts updated over the air for the full term
- 10% spare units held in-country from day one
- A regulator endorsement letter materially lifts panelist trust and acceptance rates
What you see
The identical script, every minute, all year
Statistical power on a probe panel comes from probes multiplied by time, not from panel size alone. Three hundred probes testing every minute produce around 100,000 measurement cycles a day — against a few thousand from an entire drive-test campaign. Every result carries its probe ID, timestamp and confidence score.
- Core broadband, TWAMP isolation, web vitals and video MOS on the same cadence
- Results cached locally if connectivity drops, then replayed — an outage is measured, not lost
- Targets derived from that household's own subscribed plan, not a national average
- Test servers sit outside the operator networks, so no route is optimised for us
What you see
Keeping the panel alive and honest
A panel drifts quietly unless someone is accountable for keeping it healthy. Heartbeat monitoring raises an offline alert after 24 hours, faulty units are swapped by courier without an engineer visit, and roughly a tenth of the panel rotates every quarter so fatigue bias never accumulates. Operator and plan are re-verified quarterly, because subscribers change tariff without telling anyone.
- Day-after recall validation on every probe's output before it enters an aggregate
- Outlier and anomaly filtering, plus anti-tamper controls at the edge
- Panel availability held above 95% monthly, with the figure published alongside each report
- Every raw record auditable per probe, per minute, for the full term
What you see
Standing reports, plus the one-off question
Monthly national reporting, quarterly operator benchmarks by region and an annual state-of-broadband report all run on a schedule you set. Between them, the portal answers the questions a report was never designed for: validate one complaint from one probe's raw history, isolate the failing segment, or run a before-and-after study on a network upgrade.
- Advertised versus measured compliance, against plans verified from a bill
- Complaint validated down to one address, one minute, with the failing segment named
- Scheduled PDF and Excel delivery, plus an open API into your own systems
- A public reporting feed whenever you decide to publish
What you see
Quota, eligibility, hosting and risk
The parts a procurement team reads and a pitch audience skips.
- By region — subscriber weighted, with a floor for remote regions
- By operator — proportional to subscriber share; no operator below 15% of any reportable region
- By segment & plan — residential and business, across every advertised speed tier
- By household profile — urban and rural, villas and apartments, national and expatriate mix
The matrix is agreed before recruitment opens, and quota fill rate is reported weekly.
- Active fixed broadband subscriber at their main home or business premises
- Free LAN port on the router — the probe connects by Ethernet, never Wi-Fi
- Always-on power and a 12-month commitment to keep the probe installed
- No household member employed by an ISP, the regulator, or a network vendor
- Operator, plan and advertised speed verified from a recent bill
- Candidate fills an open cell in the agreed quota matrix
On-premise, in country
- Full platform in your own or a local data centre
- All measurement data stays inside the country
- Your access control, your IT policy
- Same features and update cadence
Mozark cloud, region-pinned
- Hosted in the region of your choice — live in weeks
- Nightly raw-data export to storage you control
- SOC 2 certified · GDPR compliant
- Migration path to on-premise at any time
| Risk | Type | Mitigation |
|---|---|---|
| Recruitment slow in remote regions | Operational | CATI and snowball channels plus incentive uplift; over-recruit dense regions, then rebalance. |
| Panel attrition erodes validity | Operational | 10% spare units in-country, quarterly rotation, 12-month tenure, replacement inside five days. |
| Type approval or customs delay | Regulatory | Approval filed on the day of contract signature; the regulator is usually the type-approval authority. |
| Operators dispute the results | Stakeholder | ITU-T aligned methodology published up front; neutral servers; raw records auditable per probe. |
| Data residency or privacy objection | Regulatory | Full on-premise option; no PII captured; SOC 2 and GDPR aligned; you own every record. |
| Home Wi-Fi masks ISP performance | Technical | Probe wired to the router by Ethernet; in-home Wi-Fi measured and reported separately. |
Four programmes running on this hardware today
A Gulf regulator — national fixed-broadband probe panel
The challenge
- An existing home-measurement device programme proved speed and nothing else — no way to isolate whether a fault sat in the last mile, the core or international transit.
- Throughput figures said nothing about whether video and government services actually worked for citizens.
- A voluntary device programme only measured the households that had asked for one, leaving 500,000+ lines unrepresented.
- The national data protection law required a defensible answer on where measurement data would live.
What we did
- A quota-controlled panel of 300 wired probes across all eleven governorates, sized statistically rather than to a budget, with the calculation published alongside every report.
- TWAMP segment isolation, ITU-T P.1203 video MOS and Core Web Vitals on top of the standard broadband KPI set.
- 200M+ crowdsourced records already held for that market, giving a national baseline from day one and showing where the probes should sit.
- Full on-premise deployment option, no PII captured, and every raw record owned by the regulator.
The outcome
- National and per-operator reporting at 95% confidence, with the largest governorate reportable on its own.
- Advertised-versus-measured compliance baselined against plans verified from a bill, not self-declared.
- Complaint investigation down to one household's minute-level history, with the failing segment named.
- Monthly, quarterly and annual regulatory reporting scheduled without a vendor ticket.
Marina Bay Sands
An 80-location fixed-probe network monitoring StarHub and Singtel simultaneously across the entire hotel and casino complex — 24/7 KPI dashboards and SLA validation for Sands Group.
SNCF & ARCEP — railway network
Fixed probes across SNCF's high-speed rail infrastructure measuring in-train and trackside connectivity, alongside the ARCEP QoS campaigns that supply 80%+ of France's public regulatory portal.
Bouygues Telecom
Probe-based fixed broadband and indoor coverage monitoring — residential QoS at scale, supporting SLA reporting and network planning with operator-grade data confidence.
Hardware data sheets
The other half of a regulator's job — and the one measurement tools rarely cover.
Measurement tells a regulator how the network performs. It says nothing about the thirty unsolicited calls a citizen received this week, or the impersonation scam that cost them money. This offering is the consumer-facing enforcement side: national do-not-disturb preferences, spam and fraud reporting, and complaints that carry a reference number and a resolution. It runs on the same platform and it is live today for 1.4 billion subscribers.
Same platform, same citizen, different problem
A regulator that already has a measurement app in a citizen's hand has the hardest part done: distribution and trust. Adding preference management and fraud reporting to that same app — or shipping it as a companion — costs a fraction of building it standalone, and it is usually the feature that drives adoption of the measurement side. In India both live under the same programme, on the same backend, with the same anti-fraud layer.
From a spam call to a tracked complaint
Registering a SIM, setting preferences, reporting a scam, and following the complaint through to its outcome — the loop that makes enforcement real rather than declarative.
DND preference management
Block all unsolicited communication, promotions only, or a custom preference set — configured per SIM card independently, so a dual-SIM household is not forced into one policy.
Fraud & scam reporting
Report suspected fraud, scams and impersonation. Financial fraud cases route straight to the official cybercrime channel rather than sitting in a telecom queue.
Complaint tracking & status
Every complaint gets a unique reference ID. The citizen sees status and outcome inside the app — which is the difference between a regulator that appears to act and one that visibly does.
Multi-SIM support
Each SIM registered and managed independently for accurate operator routing and enforcement attribution.
Five steps, and the citizen sees four of them
The citizen registers each SIM and sets a preference — all promotions, none, or a specific category set.
An unwanted call or SMS is reported in two taps, with the originating number and category captured automatically.
The complaint is attributed to the originating operator; suspected financial fraud is routed to the cybercrime authority.
The operator is given a time-bound window to investigate and act, with the regulator holding the audit trail.
The citizen tracks the reference ID to a visible outcome. Nothing disappears.
The proof for this offering
TRAI MyDND — national spam and fraud reporting at 1.4 billion subscribers
The challenge
- An epidemic of unsolicited commercial calls and SMS with no scalable, subscriber-facing way to report violations.
- Spam reporting scattered across operator channels, with no aggregation and no regulatory weight.
- Financial impersonation scams needed a route to cybercrime authorities, not a telecom complaint queue.
What we did
- A national DND application with per-SIM preference management and two-tap reporting.
- Unique reference ID on every complaint, with status visible to the citizen throughout.
- Automatic routing of suspected financial fraud to the official cybercrime channel.
- The same anti-fraud and device-integrity layer used by the measurement apps.
The outcome
- A unified DND platform for 1.4 billion subscribers, replacing fragmented spam-reporting channels.
- Enforcement backed by an audit trail the regulator owns rather than the operators.
- A consumer-facing reason to install the regulator's app — which lifts the measurement dataset too.
The KPI set does not change when the collection method does.
This is probably the most consequential architectural choice behind the platform. A download measured by a citizen's phone, a browser, an embedded SDK and a wired probe are all computed the same way, against the same definitions, on the same engine. That is why a crowdsourced finding and a probe finding can be put on the same chart — and why adding an agent later never means re-baselining anything.
Which agent collects what
Where a cell is empty it is a platform limitation we state openly, not a gap we hide. Nothing here is redefined per agent.
| KPI family | Crowdsource | QuickApp | Web Tester | EdgeSDK | EdgeX Probe |
|---|---|---|---|---|---|
| Throughput & consistency | ✓ | ✓ | ✓ | ✓ | ✓ |
| Latency, jitter & loss | ✓ | ✓ | ✓ | ✓ | ✓ |
| Latency under load (bufferbloat) | ~ | ✓ | ~ | ✓ | ✓ |
| DNS & TCP connect | ✓ | ✓ | ✓ | ✓ | ✓ |
| Web browsing / Core Web Vitals | ✓ | ✓ | ✓ | ✓ | ✓ |
| Video streaming QoE | ✓ | ✓ | ~ | ✓ | ✓ |
| App video MOS (ITU-T P.1203) | ✓ | ✓ | ✗ | ✓ | ✓ |
| RF / cellular radio | ✓ | ✓ | ✗ | ✓ | ✓ |
| Wi-Fi link metrics | ✓ | ✓ | ✗ | ✓ | ✓ |
| TWAMP segment isolation | ✗ | ~ | ✗ | ~ | ✓ |
| Traceroute & path analysis | ~ | ✓ | ✗ | ✓ | ✓ |
| International CDN round-trip | ~ | ✓ | ~ | ✓ | ✓ |
| Packet capture (PCAP) | ✗ | ~ | ✗ | ✗ | ✓ |
| Voice & call quality | ~ | ~ | ✗ | ~ | ~ |
| Availability & uptime | ✗ | ✗ | ✗ | ~ | ✓ |
✓ available today · ~ partial or on roadmap · ✗ not possible on that platform
Every KPI, filterable
Filter by where it runs and which layer it sits in.
15 families shown.
Netflix · YouTube · Instagram · TikTok · Snapchat
Rather than screenshots, open the product.
Pick a role, then click through the modules the way that person would. Every module carries the question it exists to answer and the one thing worth knowing about it. Twelve modules, one login, one master filter driving all of them.
Configured, not coded.
The same deployment serves a regulator, an event and a fixed-broadband probe fleet at once. Colour, logo, geography and operator labels are values in a table — not a fork of the codebase.
Multi-tenant & white-label
One deployment, many tenants. Branding and geography are configuration values, so a new tenant is onboarded in hours.
Config-driven
Every page, widget and KPI is individually toggleable, and the result saves as a reusable template.
One master filter
Source · Operator · Tech · OS · Geography · Period drive every widget on every page, all the way down.
Management + engineering
Executive answer-at-a-glance and analyst drill-down live in the same product, not in two tools.
Thirty-plus regulators. These are the ones happy to take a call.
Every organisation below runs a live national or enterprise-scale deployment on this platform — not a pilot, not a proof of concept. Reference calls are available before an RFP is drafted.
Where the platform is the official measurement system
White-label consumer broadband app, FedRAMP accredited, live on both US app stores and integrated with the Broadband Data programme.
MySpeed, MyCall, MyDND and the web tester across 22 circles, in 22+ languages, six years live at billion-subscriber scale.
Crowdsourced QoE plus expert drive test supplying 80%+ of the data on France's public regulatory portal, continuously since 2015.
Central QoS/QoE aggregation platform bringing drive test, crowdsource, operator and network data into one regulatory view, plus the national drive-test campaign.
Multi-year crowdsourcing deployment for the Belgian telecoms regulator, covering fixed and mobile network measurement.
QoS/QoE training and platform across the Francophone regulator network — Senegal, Mali, Ivory Coast and more.
TADURÉZO citizen app plus 12 measurement kits producing 1.2M expert test points across 48,000 km² for New Deal Mobile subsidy claims.
The same platform outside the regulator world
An 80-location fixed-probe network monitoring StarHub and Singtel simultaneously across the hotel and casino complex, with 24/7 KPI dashboards and SLA validation.
Fixed probes across high-speed rail infrastructure measuring in-train and trackside connectivity, alongside the ARCEP QoS campaigns.
Probe-based fixed broadband and indoor coverage monitoring — residential QoS at scale supporting SLA reporting and network planning.
App Testing Kit probes with the operator and the Department of ICT, correlating application failures directly with network root causes.
Marks are used solely to identify live Mozark deployments and remain the property of their owners.
Standards, certifications and compliance
ITU-T Study Group 12
Member since 2021 — we help shape the QoS/QoE measurement standards we implement, rather than only complying with them.
SOC 2 Type 2 · ISO 27001
Independently audited security controls, renewed annually, covering the platform, the pipeline and the operations team.
GDPR & data sovereignty
No PII captured by the measurement engine. On-premise option, regulator-controlled retention, nightly raw export.
Federal-grade delivery
FedRAMP-accredited work for the FCC — the security bar every commercial measurement tool failed. US Patent #11243866.
Built for regulatory programmes, rather than adapted from a commercial benchmark.
Everything here can be verified with a reference call before an RFP is drafted. Where a competitor is stronger we say so; where the assessment is category-level rather than vendor-specific, we label it.
What actually separates this from the alternatives
A national picture on day one
Hundreds of millions of crowdsourced records already collected in most markets — every region, every operator — before a probe ships. Few suppliers can begin from a dataset they already hold for your market.
Layers a whitebox cannot see
TWAMP segment isolation, ITU-T P.1203 video MOS and Core Web Vitals. A speed-test appliance measures the pipe; this measures whether the service works.
One vendor across everything
Fixed, mobile, drive test, apps and consumer protection on the same engine and portal — no integration project, no second procurement, no methodology argument between datasets.
You own every record
Full on-premise deployment option, SOC 2 and GDPR aligned, nightly raw export to your own storage. We do not resell measurement data to third parties, so the dataset stays yours alone.
Live national deployments
FCC, TRAI, ARCEP, TDRA and IBPT run on this engine today, some for a decade. One reference call away, in every region.
Thirteen years of engine
Started as 4GMARK in 2012 and evolved through every generation to 5G SA. Field-hardened across 30+ countries rather than newly written for a tender.
The three ways a regulator can do this
Assessment reflects publicly documented capabilities of each category, not a named vendor.
| Capability | Crowdsource only | Fixed panel specialist | Mozark |
|---|---|---|---|
| Wired probe measuring the ISP line at the premises | ✗ | ✓ | ✓ |
| National crowdsourced baseline from day one | ✓ | ✗ | ✓ |
| TWAMP segment isolation — last mile, core, transit | ✗ | ✗ | ✓ |
| Video MOS (ITU-T P.1203) and web QoE scoring | ~ | ~ | ✓ |
| Local apps and government service journeys | ✗ | ✗ | ✓ |
| Localised dashboards and citizen-facing outputs | ✗ | ✗ | ✓ |
| Consumer protection — DND, spam and fraud reporting | ✗ | ✗ | ✓ |
| On-premise hosting inside the country | ✗ | ~ | ✓ |
| Regulator owns raw data; no resale to third parties | ✗ | ✓ | ✓ |
| Mobile, drive test and fixed on one contract | ~ | ✗ | ✓ |
Every specification, downloadable from this page.
These files are embedded inside this document — no server, no login, no link that expires. Click one and it saves to your machine, including when this page is opened offline or forwarded as an attachment.
KPI and data specifications
Applies to every collection agent — the definitions do not change per product.
EdgeX probe data sheets
Not embedded here
Corporate & compliance pack
SOC 2 Type 2 report, ISO 27001 certificate, ITU-T membership, certificate of incorporation and reference letters from ARCEP, IBPT and the European Commission.
Sample RFP responses
Complete technical responses submitted to national regulator tenders, including compliance matrices and methodology annexes.
Reference calls
Direct introductions to the regulators listed under Customers, arranged before an RFP is drafted rather than after.
Let's build the future of
network transparency together.
Mobile, fixed broadband, consumer protection — or all three on one contract. We are ready to deploy a proven, white-label platform tailored to your regulatory or smart-city programme.
Sanket Jain · Connectivity Business · Mozark
Three things worth doing first
Take a reference call
FCC, TRAI, ARCEP, TDRA or IBPT — pick whichever is closest to your own programme, and ask them directly rather than reading our version of it.
See your own country's baseline
We already hold crowdsourced records in most markets. We can show you the current picture for your country before anything is procured.
Run the portal on live data
A working tenant configured to your regions and operators is the fastest way to judge whether the answers are the ones you actually need.