Skip navigation
Like what you’re reading?

Who cares about peak download speeds in 5G?

Available in English 日本語
Think about the apps that you use on your smartphone all the time. Do you have any idea about what download speed is required for you to have a good experience when using them? Is it 1, 10, or 100 megabits per second, or even more? Is download speed expressed in bits per second a good measure of user experience in the first place? I don’t think so …

Strategic Product Manager, Business Area Networks

Who cares about peak download speeds in 5G?

Strategic Product Manager, Business Area Networks

Strategic Product Manager, Business Area Networks

47Mbps, so what?

I remember very well the day I got my first 4G smartphone. I was at home when I unwrapped the box, and one of the first things I did was a 4G speed test: 47 megabits per second (Mbps). Wow! I was excited because – like so many of my colleagues at Ericsson and across the whole industry – I felt that I had been part of the making of 4G. I showed off the 47Mbps to my then teenage kids, who had been smartphone power-users for a few years already. But all my son had to say was “So what?”. I realized that my kids could not relate to data speeds expressed in bits per second. Was 47Mbps good, bad, or just enough? Since then, I’m sure that most smartphone users cannot relate to data speeds expressed in bits per second.

Service providers should care about their subscribers’ 5G user experience

Subscriber acquisition costs and churn rates are key factors that determine the success of any communication service provider. Numerous studies have shown how churn is directly connected to user experience. This has been known since the early days of mobile telephony. We have all experienced the frustration of a dropped voice call. And when that drop occurs at an inconvenient point in the conversation, requiring you to redial, or even to re-enter a personal identification number (PIN) code to rejoin a conference call (but you can’t, because you are driving), that frustration can quickly become infuriating. No wonder that leading communication service providers target call drop rates of well under 0.5 percent.

For voice calls, measuring user experience is straight forward, using well-defined metrics such as dropped and blocked call rates and voice quality scores. But when it comes to Internet connectivity for smartphones (mobile broadband), the mobile industry still struggles to even agree what user experience means, let alone how to measure it.

It’s really simple, though …

Smartphone user experience is all about time-to-content

Time-to-content is what matters for smartphone users: after I click, how much longer do I have to wait before my video starts playing, or I can start scrolling on my news page? Like dropped voice calls I really don’t like that spinning wheel …

The damn spinning wheel

Figure 1: The damn spinning wheel1

Time-to-content is the key metric that the content provider industry has been using for decades to assess user experience. Entire industry branches, such as content distribution networks, have emerged to minimize time-to-content, and thereby maximize user experience. A consistently low time-to-content is crucial for any online business since it can impact conversion rates, i.e., the probability that a customer stays on the online site and eventually clicks “Buy”.

How come the mobile industry has not adopted time-to-content as a way to measure user experience?

20Mbps at every click is what I need to have a great 5G user experience

At Ericsson we have conducted an in-depth study of smartphones requesting popular content (YouTube, Instagram, Amazon, ebay, Uber, IKEA, and many more) to derive the relationship between time-to-content and what we refer to as throughput “at click”. The throughput “at click” is the throughput that is available to the device during the time-to-content phase. The study was based on tools and guidelines provided by Google on The relationship we derived is shown in Figure 2. The graph incorporates projections for the coming 5 years about the growth of the size of popular content, and improvements in device processing power.

Time-to-content depends largely on downlink throughput “at click”

Figure 2: Time-to-content depends largely on downlink throughput “at click”2

We have made a number of key findings:

  1. A downlink throughput “at click” of 20Mbps and an uplink throughput “at click” of 1Mbps provides great smartphone user experience for the vast majority of popular content. Beyond these speeds, user experience improves only marginally.
  2. User experience degrades sharply as the downlink throughput “at click” drops below 5Mbps or the uplink throughput “at click” drops below 300 kilobits per second (kbps).
  3. Streaming video is less demanding. Even at High-Definition (1080p) YouTube videos start playing in less than 2 seconds given a downlink throughput “at click” of 10Mbps. Once the video has started playing, the downlink throughput may even drop to 5Mbps without impacting the video playout quality.
  4. No matter how high the throughput “at click”, there is always the device processing delay that limits time-to-content. We project that in 2025 the device processing delay will be slightly less than 1 second for high-end smartphones.

Users expect a great experience everywhere and every time

Users not only expect great experience, but they also expect it everywhere and every time. Hence, 20Mbps in downlink and 1Mbps in uplink needs to be available to a user’s smartphone at every click, even in busy areas and during busy times when the network may be highly loaded. This is not always the case as shown in the chart below. It shows measurements of time-to-content collected from a smartphone in a live 4G network.

Figure 3: Congestion leads to poor user experience

Figure 3: Congestion leads to poor user experience

The smartphone was stationary, in the business district of a major US city, measuring time-to-content for once per minute. As can be seen from the chart, time-to-content is consistently low in the middle of the night when the network load is low. But as the load increases during the day also the time-to-content increases. User experience also becomes increasingly inconsistent as the range between best- and worst-case time-to-content grows. In the middle of the day this 4G coverage area is completely congested.

Obviously, frequent samples over time are required to get the full picture of the situation.

High network load moves with the mass of people throughout the day as illustrated in Figure 4. The animation shows how high load first occurs in the morning at the central commuter station. Then high load moves to the business area, then to an event arena, and later in the evening to a residential area.

Figure 4: High network load occurs in different areas at different times

Users don’t experience the average

It is important to measure the full range of user experience, not only the average. Just like the car in Figure 5, users don’t experience the average; they experience the whole range. They remember the extremes, particularly the negative ones. My favorite restaurants meet my expectations every time. I don’t like restaurants that exceed my expectations one time but disappoint me the next time, even if this adds up to a good average score.

Figure 5: Averages can be misleading

Figure 5: Averages can be misleading3

Measuring 4G and 5G user experience inside the radio base station

The latest Ericsson Radio System software includes an innovative feature to passively measure every throughput “at click” experienced across all live users, 24/7, in all 4G and 5G coverage areas. Each throughput “at click” sample can then be mapped to a time-to-content sample based on the relationship shown in Figure 2. This way the full range of user experiences is captured as shown for two European 4G networks in Figure 6. The vertical lines represent the average time-to-content in each coverage area, and the upper bounds of the shaded area represents the corresponding busy times worst-case. We selected the 90th percentile as the worst-case in these charts which means that 10 percent of all user activities had a time-to-content that was even worse.

Time-to-content per 4G coverage area based on more than 800 million user activities measured 24/7 in week 45, 2021

Figure 6: Time-to-content per 4G coverage area based on more than 800 million user activities measured 24/7 in week 45, 2021

So, who cares about peak download speeds in 5G?

Today's typical smartphone users don’t benefit much from download speeds of multiple 100Mbps. But they care a great deal about snappy access to content everywhere and every time, at every click. There should be no more spinning wheels in 5G!

Many 4G networks around the globe are congested in busy areas during busy times leading to spinning wheels for many smartphone users. These networks badly need a capacity booster shot into the busy areas. However, a good pair of glasses is needed to be able see where the booster shots should be placed.

Placing a capacity booster shot requires a good pair of glasses

Figure 7: Placing a capacity booster shot requires a good pair of glasses4

The ultimate judges of network performance are the users. Hence, one should look through the users’ pair of glasses to assess network performance. A good pair of glasses enables you to see the full range of user experiences, 24/7, across all user activities, and in all coverage areas. Don’t be seduced by rose-colored glasses, which hide poor user experiences behind some average, making network performance appear better than it is.

5G is not only about smartphones, though. Many communication service providers have already started offering 5G based Internet connectivity to entire households. These households are often not connected by a copper or fiber line and may not have had any Internet connectivity before. And now we’re talking download speeds of multiple 100Mbps.

5G might enable the “next big thing” that replaces or complements smartphones as the mainstream consumer device. Some of the virtual and augmented reality prototype apps I’ve seen are supercool. Let’s see what happens when my kids’ generation gets to try them out. These apps likely require higher data speeds than what the typical smartphone user requires to have a great user experience.

5G is also not only about consumers. I strongly believe that 5G will unlock great new opportunities for enterprises. Some of the robots in today’s smart factories require data speeds that exceed the data speeds that satisfy consumers by far. Not only that; they also require extremely stable and ultra-reliable data speeds.

And who cares about latency in 5G? I'll address this subject in an upcoming blog post

Read more:

Time-to-content: Benchmarking network performance

Ericsson Network IQ Statistics

Explore the Tech Unveiled blogs and videos


1 Source:

2 Source of the two emojis:

3 Source of the image:

4 Source of the two icons:


Reiner's daughter
Reiner's daughter

A picture from a family trip in November 2021. My kids are no longer teenagers, but they still allow me to peak over their shoulder every now and then to watch how they interact with their smartphone. I find it fascinating how quickly my daughter flicks through the pictures and videos of her Instagram feed. "Nothing special among my friends.", she says.

The Ericsson Blog

Like what you’re reading? Please sign up for email updates on your favorite topics.

Subscribe now

At the Ericsson Blog, we provide insight to make complex ideas on technology, innovation and business simple.