Summary

Summary

  • ICTA’s KaliteNet initiative builds on earlier proposals for crowdsourced quality measurements and comparative reporting, with requirements designed to support meaningful fixed and mobile assessments.
  • Readers can judge regional charts and provider comparisons more confidently when reporting explains who and what was measured, how observations were handled, and where the evidence’s limits lie.

When a phone displays a download speed, a response time or a measure of streaming quality, the number describes a particular test under particular conditions. It does not, by itself, describe the service every customer receives. Mauritius’s KaliteNet brings that distinction into a public measurement initiative. ICTA launched the application at a workshop on 11 May 2026, and its current page describes a free service developed by nPerf SAS, available on the web and through mobile app stores.

A Cabinet note four days later outlined a wider purpose: users could examine download and upload speeds, response time, browsing and streaming quality, while ICTA would collect results for charts showing regional coverage and speeds. The policy question is how those observations become comparisons that the public can interpret reliably.

That question predates the launch. In January 2023, ICTA proposed crowdsourced assessments of quality of service and quality of experience, with comparative publications on providers. Its response the following May summarized four submissions, from Mauritius Telecom/Cellplus, Emtel, Ookla and Respondent A. The consultation exposed a central difficulty in interpreting voluntary measurements: observations may be unevenly distributed across places and times, shaped by user bias, affected by package caps or inflated by repeated tests. Respondents also raised privacy and questioned whether crowdsourcing alone could provide a sufficient assessment.

These were concerns raised during consultation, not established defects in KaliteNet. ICTA replied that crowdsourced data could be complemented by other methods and emphasized end-to-end experience, package limitations and comparable server locations.

The procurement record shows how those concerns translated into requirements, without establishing that every requirement was ultimately implemented. ICTA’s 2024 tender asked bidders to explain their statistical methods and support representative actual use, fair and non-discriminatory comparisons, fixed and mobile connections, manual and automated tests, throughput and latency, and browsing and video experience. It contemplated successive two-month national campaigns every six months, with timing still “to be determined,” a three-month pilot and a three-year operational phase followed by national reports.

ICTA awarded the implementation contract to nPerf SAS on 11 September 2024 for €347,700. These details define an intended measurement programme, not proof of completed testing, representativeness or achieved performance. Assessing the public service therefore requires separating contractual design from verified operation and published evidence.

A useful comparison must also keep four different questions apart. A test result records measured performance at a particular point under the conditions of that test; an advertised package describes what a provider offers, not necessarily what every session delivers. Geographic coverage concerns where service is available, while a regional chart of test results concerns where measurements were obtained and what they recorded. Neither alone establishes the outcome experienced by all users in an area.

For KaliteNet, an auditable measurement contract would make these distinctions visible: how samples are gathered, how repeated observations are handled, how locations and technologies are compared, and how package limits and server choice affect interpretation. The launch announcement establishes the platform’s stated purpose, not the answers to those methodological questions. Transparent reporting could let users judge what each comparison supports—and where its evidence stops.

Sources