Do you believe a HK$368/month 5G ("five generation") subscription only gives you a 2Mbps data rate with 20% packet loss for most time of the day?

Do you know the service providers always try to over-sell the number of subscribers and use "2-year contract" to lock revenue while service quality is degrading?

Do you know there is no actual investigations regarding your physical location and the customer service's mere function is to keep you on the plan for as long as they can?

Do you know all the complaints finally (said to) reach OFCA because of "interference" but both service providers and OFCA refuse to disclose any information of the case, even if it is actually of public interest?

Following is my story regarding the business dishonesty of ThreeHK. Followed by that is the growing log of the process, organized in reverse chronological order.

I subscribed to ThreeHK's package in March 2021: 5G for 1st 50GB data and unlimited 4.5G afterwards. The data quality was acceptable in the first three months, ranging from 5Mbps to 25Mbps download rate for most of the time. The service quality degraded sharly since June 2021, ranging from K bps level to 2Mbps for most of the time.

Their first excuse was that 4.5G had a "fair use principle" but refused to comment on how they limit data rate during 4.5G range. I showed them measurement results demonstrating there were signal issues but only got response "The engineering team checked your location was Ok". The evidence is very straightforward. I measured the data rate at the same time: while TWST saw <3M rate, the same card / same router/ same phone worked well in YSW (near the pier) and HSY with a rate around 10M-20M.

I asked their engineering team to produce a proof of measurement at my place while I am in place. They refused to do so, but repeatedly responded that "The engineering team checked your location was Ok".

At that time, the 5G range was working properly (but only for 50GB data). So I proposed to top up the data volume. They had strict policies on the plan migration, e.g. only at month boundary/ only top-up but no tune-down/ etc. This path didn't go out.

Instead, I was upsold a +HK$98/month feature called "NetPlus", which was said to enhance the data rate by twice during the 4.5G range. The test started at the end of Sept. The measured data rate did not enhance much, but user experience of using web services did improve. I conjectured that the NetPlus is a backend treatment to prioritize the subscriber's packet, so it did helped some key parameters (e.g. jitter/ packet loss rate), while the overall data rate (as impacted by physical constraints) did not improve.

Three months went by and it came to Jan 2022. The service quality further declined. It is disappointing to see that even during the 5G band (first 50GB), the data rate is 2-3Mbps. The customer service responded with similar tricks to waste my time, i.e. blaming routers/ locations, and quoted the same "no problem" from the engineering team. I then conducted systematic measures in 10+ locations in north Lamma Island. It turned out download rate could reach 200Mbps when we have line-of-sight connection to HK Island side. Also, unlike June 2021, the 2-3Mbps data rate was observed in not only TWST but also YSW and HSY, i.e. whenever you connect to Lamma base station. This indicated an alarming issue: it is not just my own wireless connection quality problem (potential base station angle/ potential interference); it is kind of backbone congestion of the Lamma. The investigation went on with my own statistics taking from Sept to Dec. There was an apparent time-of-day and day-of-week pattern of congestion, meaning that backend is more likely the issue.

I sent all the statistics (See 20220118 section) to ThreeHK waiting for a thorough investigation.

COVID impacted every single person and every single business. We totally understand that WFH might also be a factor of the service quality degradation. We could accept a discount/ a refund, based on thorough investigation and transparent communication.

The most disappointing part dealing with ThreeHK is that they exhibited a very high level of dishonesty throughout the process.

Back to June 2021, when the data rate issue is only in TWST, I reiterated many times of my measurement results and asked them to check base station issues, e.g. angle/ power. Their engineering team only responded "no problem", without further details. We highly doubt if they ever bothered to investigate or not.

As to the Jan 2022 issue, my measurement results show solid evidence that the data rate is a backbone congestion issue, INSTEAD OF SIGNAL QUALITY. Funny thing is that, this time ThreeHK started to blame signal quality, quoting "an interference" around my place. They said a case was filed to OFCA for further investigation but refused to let me know the date when the case was submitted, despite asking them three times.

OFCA side is also disappointing. First, according to ThreeHK, if they really submitted such a case, the investigation did not yield any results 1.5 months after that. I filed a data request using "Code of Access to Information", to get a list of complaints of signal interference but got refused due to concern of commercial secrecy. Since signal interference, if ever existed, is of public interest, I asked for an internal review, requesting the same data excluding operator's name.

To conclude, the general trick of an operator's customer service is to blame your device/ location/ usage pattern/ service package. They would intimate you the subscription is binded for two years so you can not cancel early. They would not reproduce a proof of capability to provide service, but instead ask you to produce a proof that they are incapable of doing so. The conversation eventually leads to taking a screenshot using OFCA's App. They would then quote a signal interference issue under investigation by OFCA, so the ball is on OFCA. After waiting 3-6 months, if you follow up the case, they would let you know the interference is lifted. However, you don't find things have ever improved. Then they would offer a cancellation. From the time of first reporting to the time of cancellation, they locked you in for another half a year or so.

In my case, the response I got followed the common script, but didn't answer my questions from real measurements. I highly doubt if they ever did an investigation throughout the process.

In the end, I shall add a bit of a status update for Feb. Although the data rate is not as good as prior June 2021 days, it did get improved somehow at the end of Feb. If OFCA's investigation did not have a resolution yet, that means ThreeHK did have some handles to playaround to improve. We ask for total transparency and set proper expectations of the ongoing service quality to make our subscription decisions. At the same time, ThreeHK offered a cancellation of my plan.

I am still holding the plan so that I can keep on measuring during the 5G and 4.5G band. I believed that many others had the same pain points and were tricked by the same processes. (Several friends shared a similar process with other providers). I will keep paying $368/month until there is a solid resolution with OFCA, upon which time, I shall have strong evidence to charge indemnity of the contract, as well as cost of my time, due to their business dishonesty.

All kinds of support are welcome, no matter if you are tech journalist, solicitor looking for pro bono case, Lamma resident, or just another sufferer.

My bill cycle ends at 7th every month. As one can see from the screenshots in previous section, I was using 4G (no 5G signals at all) in Mar 7. Then it comes to Mar 8 morning. I was shocked to find out that it said I already used 37.15GB during midnight to 9:19am.

Let's do a back of envelope calculation. Suppose the data rate is 10Mbps, which is of course far below according to my measurements. Converting from 10Mbps to Bytes, that is roughly 1MB/s, given all sorts of overheads. From midnight to 9:19am, that is 9 * 3600 + 19 * 60 seconds.

That is, even if I stayed all night on and utilized to the maximum, the estimate would be 33.54GB, which is less than 37.15GB as reported by the App.

Also, as measured by OFCA's App, ThreeHK consistently incurred around 20% packet loss. That is also a source of overcharging.

I seriously doubted my monthly bill and requested a breakdown of the data usage via PDPO (Cap. 486) at Sept 2021, but received no response so far.

Ask for Three: stop overcharging your user, find the bug in the accounting system, and refund the overcharged portion.

Summary of today: I tested using 4G band and reproduced the observation that when connecting to the HK Island side the data rate is much higher even when the signal is much weaker.

Ask for ThreeHK: Stop fooling around with "signal interference". My measurement in Jan and in Feb, for both 5G and 4G, showed that the issue is in the Lamma backbone. If it is an overselling issue, we ask for reasonable compensation before mutually agreed cancellation. If it is an intentional rate limit by user differentiation on the Lamma gateway, we ask for transparency of the limiting algorithm, as well as a sound explanation in public for the "unlimited 4.5G data" as advertised by the provider.

Another finding is that speedtest.net returned much faster "download rate", as compared to OFCA. We can see that the speedtest.net often selected a server hosted by Three.

Power Station Beach

Note that "5G Signal Strength" field is empty. So this test and the following one both use 4G.

I also tried to conduct an Internet conference via Google Meet and the quality is very poor at this location and at this time. Videos are off for most of the time.

Windmill mountain

Then I hiked to the top of Windmill mountain -- I don't know the actual name of the mountain but there is a wind power station.

Apparently the signal is weaker at this location compared with it at most locations in Yung Shue Wan residential area.

It turns out that the data rate is much higher than that in YSW. Again, this shows that signal strength is not the core issue here. Any quote of signal interference is just an excuse for their Lamma backbone issues.

•Three already closed my case. When contacting “2126 6133”, it says no cases are being followed up.

•I look-up the case from 1033. Information obtained below.

•Main excuse from OFCA not to give such data: 1) The data belongs to MNO, i.e. 3rd party data; 2) MNOs all refused to do so; 3) the interference information is commercial secret.

•Asked to OFCA to conduct an internal review^[1]. Also trimmed down the information columns to avoid exposing MNOs.

•"Only customer service has the detailed record."

•[can find in 0128 note] "there is signal interference; will apply to the company to cancel subscription".

•"We don't have further notes except for the last one (i.e. Jan 28)".

•Claim the Internet speed is due to signal interference

•Inform OFCA is investigating the case

•Claim the signal interference is not their responsibility

•Offer "cancel the upcoming subscription"

•CS refused to share the case ID with OFCA

•I made an offer: refund the incurred fee since my first report of this case and mutually end the contract; I will save time from follow ups.

•Ask for the case ID.

•Interference is not the only cause. Refer to the test map on Jan 18 where the pier has good signal but still suffers from low data rate.

•The way Three HK communicates with customers is dishonest. The whole purpose of CS is to delay the contract and charge as much as possible.

•I reserve the right to ask for compensation related to this contract.

Map of service level

•My 5G ZTE router is perfect, as it can get a download rate of 100M-200M bps at good locations.

•It seems the north coast & east coast/ hill of Lamm North is covered by a base station from HK Island. See the big bubbles around Windmill Hill, Pak Kok Tsui (around the lighthouse), Pak Kok Pier and Pak Kok old village.

•Residential areas of Pak Kok Tsui and Pak Kok San Tsuen have poor signal. Note the two black holes on the map.

•YSW pier (around library) has best signal, indicating a line of sight from the base station

•YSW main street and PS beach have similar level of signal; the signal degrades to similar level at Tai Peng and HSY beach. This reconciles with the salesperson's description that a base station is installed in the power station.

•When connecting to base station on HK side, the download rate is higher but upload rate is lower (even lower than YSW main street) than that when connecting to Lamma

•It shows that the emitting signal strength from the Lamma base station is not enough

Furthermore, the packet loss rate from Lamma station is 20%-30%, but it is single digit from HK side. This is an alarming signal that we are likely overcharged due to packet loss. Before ThreeHK discloses the actual measurement point and algorithm, we can only assume the data volume is measured by data passed between the device and base station. If this conjecture is true, the overcharge ratio is not just 20%-30%. Instead, it will be much higher, as there is retransmission scheme built in connection oriented network protocol.

Data log since Sept 2021

Following is tested with the CLI tool from speedtest.net, running by a cron job on my workstation from home.

•The average and best daily performance are similar.

•The download speed is able to reach around 25Mbps at some time of day and some time of week.

•Many days have poor performance, with a max download rate of less than 2Mbps.

•The best performance hours are between 1am-7am HKT.

•The work hours, i.e. 9am-7am, observe poor performance.

•The dinner hours, i.e. 5pm-10pm, observe worst performance, with 9pm being the worst of all

Conjecture: few people play cell phone during sleep; many people play cell phone during dinner time and work time; slightly more people play cellphone during dinner than during work.

•The data rate is significantly correlated with people's work & life patterns, i.e. not pure random variance across time.

•This indicates congestion to the Internet, i.e. the backbone rate (from Lamma access point to the Internet) is also a bottleneck, except for transmitting weak signal from the base station.

•The question is how are residents on Lamma Island prioritized? By 5G quota? By NetPlus? By monthly usage?

Due to the variety of servers provided by speedtest.net's network of sponsors, we can check the average rate to each test point. It is funny that 3HK ranks 15th among other servers.

Since we can observe high speed when connecting to HK island side, it shows ThreeHK do have the backbone to deliver good data rate. So the further question is, how does ThreeHK route the data from/ to Lamma Island? How is Lamma Island prioritized with other regions?

TLDR: even with sample 5G quota, the signal is weak and the data rate is way below 5G's expectation; the test is the same on iPhone 13 and ZTE router.

Pre-test validation

20220117 ZTE

20220117 ZTE 5G NSA

20220117 iPhone 13

Test result on iPhone with the 5G SIM card directly; The speed is way below 5G standard

Xx: between 0114 and 0117 (forgot exact date because I assumed good will and did not plan to write this running log by then)

The above are their usual responses like previous rounds. This is the direct trigger for me to measure the data on my own to show that their service level is below standard. To summarize, their usual responses are:

•It's your phone's problem

•It's your router's problem

•There is no signal interference

•If you insist there is service quality issue, use OFCA software to measure and give them the screenshot

•If you already run out of 5G quota, the lower data rate is meant by 4G.

•There is "fair use" principle

Signal lost for 30min, which drastically interrupted my work. I emailed to "[email protected]" with the content:

Signal loss for 30min while the supposed 5G plan should be in effect:

The message and screenshots were sent by CMHK on my phone. The photos were taken by another phone running SmarTone and emailed to the CMHK phone.

Your attachment has been received. We deeply understand your feeling and frankly want to settle your case promptly via phone. It would be highly appreciated to have the following information for verification and our follow up action. Thank you!

1.Mobile number

2. Registered name in full

3. Last 4 units of your HKID number (including the bracket content)

4. Your device model

5. Detailed location for the low speed situation （including flat number, floor, block）and, how about outdoor?

6. Any changes of setting or phone model recently?

7. Any APPs did you try? Any error box? download speed

8. Wifi is On or Off?

9. Your contact number and convenient time (10:00am -22:00pm):

You are most welcome to contact us through our 24 hours Customer Service Hotline 1033. We also suggest you can contact 3iChat by instant chatting via My3 App or our website https://3ichat.three.com.hk/michat/Client/Dashboard/login?bu=MOBILE&lang=en or email to [email protected] if further enquiry.

The information requested here was intimidating. I think the rational act is to let the engineering team check the base station around the report location first.

How could the data service provider ask for so much information via the Internet when the user reported signal loss?

It's like when you break your leg and phone the hospital, and they just reply: please come to our triage point to assess your status; furthermore, please prove that you are you; oh, almost forget to ask, did you try to walk with another leg or walk with hand?

•5G data quota (50GB) can not be topped up on demand. New monthly quota needs to wait until the next cycle to reflect.

•Paid extra HK$98/ month of "Net Plus" to "enhance the 4G data part". Pending further tests.

•Why did Three's 4G signal & data rate suddenly worsen around mid June? It previously worked well for three months, so the "fair use" can not explain the whole story.

•Why did Three's 5G signal & data rate suddenly worsen around late Dec/ early Jan? It previously worked for initial 7 days per month even with 4G signal/ data rate downgrade.

•[Till Sept 2021] Why did Three's 4G/5G service quality vary according to *angles* drastically in nearby locations?

•[New in Jan 2022] Why was Three's download rate via 5G limited to 2-3Mbps on Lamma Island even around the pier area where the signal is very strong?

•Regarding the process transparency, how did Three's engineer verify "there is no signal problem" each time upon first report of the case?

•Regarding the process transparency, how did an OFCA screenshot inform Three that there is signal interference? Why can not a speedtest.net result do the same?

•My location: Tai Wan San Tsuen.

•Base station: estimated to be 600m-800m away, 235 degree SW of my location, within Line of Sight.

•The data rate within "unlimited 4G" part was acceptable from Mar to early June

•Three HK is fully aware of my regular service location (home), so geolocation coverage shall not be an excuse.

•The ZTE 5G router was bought from Three HK's retail store where I purchased the SIM card + service plan, so 5G device shall not be an excuse.

•Even testing with iPhone 13, the results are the same as the ZTE router at the same time.

•When the data plan enters "unlimited 4G" region: 1) the data rate drops sharply to around 3-4Mbps at Tai Wan San Tsuen; 2) at the same time can still reach 20-30Mbps around the YSW pier; 3) at the same time can still reach 20-30Mbps around HSY beach