More and more companies claim to assess Internet networks performances for service providers. The differences between all of us come down to filtering methods and global methodologies when it comes to analyze crowdsourced data. One question then remains: how to ensure speed test metrics are reliable and accurate?
For over twelve years, we have developed telecom tools to help people assess their Internet connection. We continually refine our methodology to make sure our clients can trust the information they are receiving. Our well-proven technology is used by telecommunication operators, regulator bodies, journalists, and private companies.
Solutions
A strong server network for accurate metrics
Nearby servers have optimal results as they eliminate bottlenecks which reduce performance. The more server they are, the more accurate the results are by minimizing data links between devices and servers.
At nPerf, our speed test algorithm chooses the best server among more than 2000 across the world. The nPerf users can run a speed test thanks to a local server, reducing travel time that data need to make a round-trip from their device to our server.
A right way to measure an Internet connection
Our speed test measures the full capacity of an Internet connection. We flood the bandwidth to represent the full speed capacity with dedicated foreground tests. In this way, we can accurately measure download and upload speeds, latency, packet loss, jitter and so on.
Each user running a test captures a snapshot of a connection for that given time, location, device, and technology. All combined, these tests represent the real-world performance of a service provider based on its users’ experience. They embody the customer experience.
Our tests are consistent across devices. We have a mobile app for both Android and iOS, which goes further into Internet measurements thanks to streaming and browsing tests. But our users can also run a nPerf speed test on computers if wanted.
A filtering method for unbiased data
Quality first! We have worked for years on a filtering algorithm to make sure KPIs are the closest reflect of the real customer experience. For instance, we detect tests run on the same device, the location of devices, and the weight of each device. Any use that deviates from the normal use of the application is considered as irrelevant. Probes or measurements robots are ban from our results.
We evaluate each operator based on the samples provided by each of its users only. At nPerf, the quality over the quality is the key for accurate and reliable metrics.
For any Internet measurements data, make sure to check out where the metrics come from and how accurate they are. At nPerf, we only publish reliable data analysis done by our statisticians. We are an impartial and independent third party.
Want to know more about our data? Contact us!