19,257 users have contributed 344 million samples to our crowdsourcing analysis of the Singapore networks over two consecutive periods spanning three months each.
The evaluation area of our crowdsourcing represents
97.8 per cent of the built-up area of Singapore.
In order to validate and supplement the results of the drivetest, we have also performed an extensive crowdsourcing analysis of the mobile market in Singapore. This analysis is based on crowd data that has been gathered in two periods of three months each. P3 decided to follow this approach because our crowdsourcing metrics are based on three-month periods. As the drivetests were conducted in May 2018, we wanted to factor in crowdsourcing data from this period as well — resulting in including the measurement period April to June 2018.
On the other hand, we also wanted to present data that is as current as possible. This is why we also included a second period ranging from July to September 2018. Considering two consecutive three-month crowdsourcing periods allows us to also have a look at trends and developments between the two compared periods.
The results are based on usage data that have been collected from smartphone users who are utilising one or more of 800+ apps in which P3 has integrated background diagnosis processes determining relevant usage data 24/7, 365 days a year. Quarterhourly reports are generated daily and sent to P3‘s servers for a thorough analysis (see detailed description of the methodology on pages 9 and 10). This way, more 19,257 users have contributed a total of 344 million samples in the described six-month period. Based on the total population count of 4.6 million people, one of 308 inhabitants of Singapore has contributed to our crowd data. The evaluation area of our crowdsourcing represents 97.8 per cent of the built-up area of Singapore.
ALL OPERATORS SHOW HIGH PERFORMANCE, BUT THERE ARE STILL SOME VARIATIONS IN THEIR CROWD RESULTS
The overall results of our crowd tests confirm the drivetest results in so far as all three Singapore operators show respectable performance. In an overall assessment, M1 leads the field with a narrow margin in the April to June period. In the July to September period, Singtel and M1 are on par. StarHub follows at close distance in both cases. A direct comparison of both periods shows that Singtel‘s performance stayed the same, StarHub was able to improve by some percentage points, whereas M1 lost a couple of percentage points. However, these variances are minor all in all.
ALMOST PERFECT VOICE AND DATA COVERAGE FOR
ALL OPERATORS AND ALSO HIGH 4G COVERAGE
For all Singapore operators, we observed 100 per cent coverage within our test area for both voice and data services during both observation periods. Even the demanding 4G coverage is very high with values between 97 and 99 per cent. Also, the results for the so called Quality of Coverage (the percentage of actual availability of the mobile network services) were high for voice and data. An examination of the Quality of 4G Coverage shows that these values were a little lower in the first evaluation period with Singtel and StarHub improving in the second period, while M1 received a slightly better result in the first period than in the second one.
When it comes to data rates, M1 shows a slight advantage in the average as well as the top values (P90) of each Users‘ Best Throughput within the observation time. These averages were 11.2 Mbit/s in the first evaluation period and 10.5 Mbit/s in the second one.
StarHub and SingTel each follow at close distance with StarHub achieving 11.1 Mbit/s in the first and 10.3 Mbit/s in the second period, SingTel‘s results are 10.6 Mbit/s in the first and 10.1 Mbit/s in the second observation period. A possible explanation for the overall drop in these values between the April to June and the July to September period could be a higher number of users in the networks leading to a higher utilisation of the network cells and their capacities.
Stil, SingTel delivers the highest values for the top speeds (P90) observed per evaluation area (“EA“). This applies to both observation periods. The difference to the User‘s Best Throughput is that these top speeds have been achieved by varying users.
Occasional drops especially in some coverage KPIs between the first and the second three-month period can be explained by user fluctuations – the number of users within a particular observation area is typically not constant between evaluation periods.
SINGTEL SHOWED NO DEGRADATIONS IN BOTH CONSIDERED PERIODS
In the evaluation for Data Service Availability, also all three competitors shows generally pleasing results.
Our analysis identified one hour with limited service availabilty in the StarHub network in March and also a two hour degradation in the M1 network in September. Both incidents only had a very limited impact on the users‘ experience of network availability and stability.
THE CROWDSOURCED RESULTS CONFIRM THE HIGH PERFORMANCE OF THE SINGAPORE NETWORKS.
A look at the average values of the best throughputs achieved by each individual user participating in our crowdsourcing (see chart below) shows that the Singapore operators deliver considerably high performance also when compared to their global competitors.
HIGHEST OBSERVED DATA RATES IN SINGAPORE AND SYDNEY
M1, StarHub and Singtel provide high data rates to their customers, which in this comparison are only matched by the performances of the Australian operators Telstra, Optus and Vodafone in Sydney.
In London, only EE achieves similar results, while the other British operators as well as the mobile networks in Berlin, Cape Town and Jakarta overall deliver slower data rates. When looking at the P90 value (the threshold below which 90 per cent of the gathered values are ranging – see also the description of our methodology on pages 10 and 11), Sydney even shows slightly better overall results than Singapore.
However, when evaluating the actual values, it must be taken into account that areas such as Singapore are densely covered with buildings, whereas cities like Berlin or London contain a higher share of open spaces such as parks or fringe areas, which are provided with less mobile coverage and also comprise of less mobile users.
Furthermore, in order to ensure statistical relevance, we had to exclude Jakarta‘s smallest operator Smartfren from this comparison.
The analysis at hand concentrates on data throughputs. But an additional look at the Areas and Quality of Coverage with 4G services, 3G/2G data services as well as voice services in the considered cities also shows favourable results for Singapore and Sydney – however, some competitors such as London or Jakarta come close in one or the other of the analysed metrics.
Deeper insights can be made available to interested parties on request.