Author: Thomas CORVAISIER

Thomas CORVAISIER is the CEO of GREENSPECTOR. After getting his degree in Engineering he worked for over 15 years in IT consulting alongside major companies, focusing on testing and software quality. He then worked on CSR topics (carbon accounting, environmental management) and eventually mixed these skills into GreenIT expertness and cofounding Greenspector.

Are we really having a positive impact?

Reading Time: 4 minutes

At Greenspector, our mission is to help our customers reduce the environmental impact of their digital services. We work on this every day, but is it enough to offset our own impacts?

To this fundamental question, we must admit that we do not have a simple answer at this time. We will propose to our Mission Committee to work on this subject in order to go further. In the meantime, here is what we can share with you.

To verify that we are having a positive impact, we need to know our own impacts, and estimate the impacts that our clients are avoiding thanks to us. For this exercise, we will focus on the most easily assessable impact data: greenhouse gas emissions.

Our impacts

This is the “easy” part of the calculation. We have been doing our carbon footprint every year since 2019, using the excellent software from our friends at Toovalu. So we know our greenhouse gas emissions. In 2021, they were 31 tCO2e.

Is this a lot? We can reduce this value to a monetary ratio commonly used in this type of approach: 31 tCO2e for 851 k€ of turnover in 2021, which represents 36.9 kgCO2e/k€ of turnover. By way of comparison, ADEME indicates an average value of 110 kgCO2e/k€ for the category “low material services”. We are therefore 66% below the sector average. This is good, but we can certainly do better.

Avoided impacts

This is where things get complicated. On the one hand, our activities are diverse (sales of licenses, audits of mobile applications or websites, help in choosing Android fleets, etc.). On the other hand, it is not so easy to know how much we have helped to avoid. We need to be able to measure “before” and “after” (which does not always happen) and ideally, between “before” and “after”, only our recommendations to reduce impacts should have been applied. However, if you develop applications or websites, you know that between two versions, it is rare that there are not multiple changes of all kinds.

Finally, there is a methodological question: let’s suppose an audit on a version 1.0, which leads to 30% less impact on a version 1.1. Can we consider that this 30% will be avoided ad vitam aeternam, in other words, can we consider that version 1.7, which will be released in 18 months, will have 30% less impact than 1.7 would have had if we had not intervened on version 1.0? For want of a better answer, we have chosen to consider that the avoided impacts apply for the next 12 months.

So we do not have the possibility of calculating avoided impacts on all our projects or for all our clients – which we regret. However, we can get an idea from a few cases for which this calculation was possible.

Let’s take three examples:

  • The SNCF Connect mobile application (Android version): we were able to determine the impact avoided on the TER ticket reservation process: -18.9% or -10 tCO2e over one year.
  • At the opposite extreme, the Orange Group’s Integrated Annual Report (IAR): the RAI 2021 website has a 55% lower impact than the RAI 2020, which for its audience – modest for this type of content – represents an avoidance of 0.024 tCO2e over one year.

Here we have 2 projects typical of our activity, extreme in their audiences, accumulating 10 tCO2e avoided, so an average of 5 tCO2e per project. However, in 2021 we worked on more than 70 application or website projects. By taking the average of these two projects, multiplied by 70 projects, we arrive at 356 tCO2e avoided. Another approach based on a categorisation of each service according to its intensity of use gives us 150 tCO2e. Let us retain this lowest value.

As a reminder, our own footprint that year was 31 tCO2e: even if our low value of 150t was still overestimated, it would be very surprising if it turned out to be less than 31t.

“And the third example?”, say those who follow

The third example is a lesser known service in our catalogue. This is our “Fleet Selection” offer, which consists, thanks to our laboratory measurements, of assisting our customers in choosing a smartphone for their business mobile fleet. The measurements and recommendations focus on autonomy criteria, of course, but also on the durability of the terminals (battery life, robustness, etc.) in order to check that they will be able to meet business needs for as long as possible.

In 2021, we helped a client choose the most sustainable device for a fleet of 35,000 smartphones and 5,000 tablets. Thanks to our measurements, he was able to ensure that he chose a model with an expected lifespan of 4 years instead of 3. The manufacturing impact of a smartphone is on average 54 kg CO2e, that of a tablet 108 kgCO2e (values extracted from the Greenspector impact model). Extending this fleet by one year before replacement therefore represents a saving of 810 tCO2e. It can be argued that we are not 100% responsible for this choice. If we attribute even 10% of these 810t to ourselves, this represents 81 tCO2e avoided – compared to the 31t we emit.

Conclusion

We don’t yet know how to quantify our positive impact precisely, and we may never know. But it is important for us to make sure that we do have a positive impact. This 2021 exercise has demonstrated that.

We will continue to work on this assessment of avoided impacts in order to arrive at more accurate results in the future. But even under conservative assumptions, we are confident that our actions are saving greenhouse gas emissions.

In addition, it should be remembered that improving the sobriety of apps has other positive impacts, particularly on the social aspect by enabling people with old and/or low-end terminals to access these digital services that are often indispensable in daily life.

All this makes us proud and strengthens us to continue our mission.

5 keys to Application success

Reading Time: 3 minutes

You work hard on the design, ergonomics and UI which are prerequisites to the success of your application. But that’s not enough: more than 7 out of 10 users are ready to uninstall an application if they feel it to be too slow. Battery consumption is also a leading criterion for them. But how can you master these “technical” aspects, when you are a Product owner or Business owner? Now there is a solution…

Continue reading “5 keys to Application success”

GREENSPECTOR and BEWIZYU, a partnership for quality mobile applications

Reading Time: 2 minutes

GREENSPECTOR and BEWIZYU have announced an unprecedented partnership today. BEWIZYU consultants, trained to use of the GREENSPECTOR tool, are now certified to intervene with the editor’s clients to provide their technical and methodological expertise.

Continue reading “GREENSPECTOR and BEWIZYU, a partnership for quality mobile applications”

Will upgrading to Android 6 improve your battery life?

Reading Time: 4 minutes

There has been a lot of noise recently about Android’s newest version 8 “Oreo”. However, a lot of people out there are still using older versions, and may be wondering if the new versions are worth the upgrade. I’ve been confronted with this choice on my Fairphone2 running Android 5.1. Should I upgrade it to Android 6 Marshmallow? Google claims that Android 6 will get you more battery life. Let’s check that, shall we!

Never install an update without first checking that you actually need it

You know how people can heavily rely on their smartphone nowadays. I’m no exception to that. I’ve been intensively using my Fairphone2 for over a year now, and I must say that I’m pretty happy with it. Except for the very few occasional crashes (but no more than on my previous device), it has become my trusted everyday companion.

When I got it, it shipped with Android 5.1 (Lollipop). The Fairphone team is doing its job properly and a few months later, they proposed an upgrade to Android 6 (Marshmallow).

But Hey, come on… I’ve been in the IT business for quite a while now, and an IT user for even longer (my first OS had a blue background). As much as I love getting the latest version of pretty much anything, I’ve developed a healthy habit: never install an update without first checking that I really need it.

What about this battery consumption claim?

Part of this habit is based upon the «if it ain’t broke, don’t fix it» approach, and the other part is defiance. OK, this update will bring nice features, and maybe I will actually need some of them. But what would be the cost of it? Will this update increase the footprint of the software (bigger memory? disk size? CPU needs? And what about battery consumption?…).

I had already checked the novelties included in Android 6 and decided that none of those were interesting enough for me to jump in. However there was still one interesting claim: Android 6 was supposed to lower the battery consumption of the OS, thus allowing longer battery life.
So, prior to upgrading, I needed to know if this claim was true.

Let’s look into this

This is where working for a company which specialty is metering battery consumption of mobile devices comes in handy. Folks in the office next door spontaneously developed an energy probe for the Fairphone2 after I threatened to join the company choir, and they even let me play with a crude but simple version of a battery monitoring app. (Before you ask: no, this is not available for sale (yet), it’s only an internal lab thing. For now.)

So, during a few days, I was able to monitor the battery consumption of my Fairphone2 when making Phone Calls, when using the apps I use on a daily basis, and also some others just for the experiment: K9-Mail, Twitter, Meteo France (France’s official weather app), and I tossed in a few games on Hearthstone (“work hard, play hard” they say) and listened to the FM Radio through the built-in antenna and app, and through some randomly picked internet Radio App.

I started by measuring under Android 5.1. Then I upgraded the OS to Android 6, and did a second run of measures. Please note that the applications have not been upgraded in the process: we’re talking OS only here.

These measures have led me to some interesting findings. I’ve been able to check whether Marshmallow really lowers battery consumption or not, and I also got some insights about individual applications behavior.

Rankings my everyday apps

The first set of measures I ran under Android 5.1 led to these findings.

All values are given in µAh/s. These are “microAmpere.hour per second”. As the Fairphone2 has a battery capacity of 2,400 mAh, this means an app with a discharge rate of 300 µah/s will drain the battery in a little more than 2 hours.

• The discharge speed for my use cases vary in range 146 to 305 µAh/s
• Some notable values:

  • Hearthstone : 305
  • Twitter : 247
  • Phone call : 191
  • Radio (built-in FM): 160
  • Idle (doing nothing but not in sleep mode, screen is on): around 150

The full list of tests cases is as seen on the graph.

Casual findings: about Twitter and FM radio

By looking at these values, we get some unexpected insights:

  • Twitter with the Night Mode on doesn’t save any energy over Twitter with the default white background: OK, this is consistent with the Fairphone2 screen technology – read more about this)
  • Listening to the radio through built-in FM consumes 20% less battery than a random internet Radio app. Of course you only get your local stations, but depending on what you want to listen to, you are better off using this built-in FM.

Is Android 6 improving my battery life?

Well, it’s time to answer this question. Let’s upgrade the phone and run these measures again. Now here is what we get:
• Average discharge speed varies in range 102 to 313 µah/s
• Some notable values:

  • Hearthstone : 313
  • Twitter : 208
  • Phone call : 163
  • Radio (built-in FM) : 143
  • Idle : 102 to 134

I had a look at the other metrics: it seems like the processor is better managed. We get around 9% less CPU used by the system when the phone is idle. CPU-intensive apps also have better performances: I get a test score of 1,058 at 3DMark Slingshot with Android 6, vs 912 with Android 5.1) but this happens at the cost of an increase in energy consumption: +10% discharge rate for 3DMark.

Conclusion

In most cases, Android 6 brings a significant energy gain: around 5% to 15%. Beware though when using CPU-intensive apps, you’ll get better performances but at the cost of more battery consumption from these apps.

So, will I keep Marshmallow on my device? Yes.