This document outlines the steps we’ve taken in 2020 to reach that carbon-neutral goal.

Our approach borrows from the work of a number of other companies, notably Wildbit, Basecamp, Who Gives A Crap, and Shopify — we thank those companies for their work in publicising their efforts and hope that we can do the same.

Our approach for 2020 focuses on:

• Reliably estimating our annual carbon footprint, to serve as a baseline for reduction efforts and to serve as an input to our offsetting efforts;
• Offsetting all of our estimated emissions for 2020; and
• Identifying concrete steps we can take in 2021 to reduce our emissions.

As you’ll see, we’ve tried to tackle the big-ticket items while acknowledging there’s much more we can do in future years. We welcome feedback from anyone on how we can improve in 2021 and beyond, and conversely would be more than happy to answer any questions to help other small companies like ours take similar steps.

Estimating our carbon footprint

We use tCO2e (metric tonnes of carbon dioxide equivalent) to measure our footprint. This keeps our calculations simple and allows us to use a single number for offsets and future comparison.

Wherever possible, we’ve leaned conservative in our estimates (for example, we calculate home office emissions based on full usage during 2020, despite the fact that we weren’t 100% WFH for the entire year).

Offices (8.68 tCO2e)

We have a single co-located office, at The Commons in Collingwood. Since 2019, they have been 100% carbon neutral (one of many reasons we’re located there), and they do a fantastic job of encouraging sustainability within the office.

While we’ve always had some remote work and work from home at Disco, 2020 has obviously seen a huge uptick in that due to the COVID-19 pandemic. The electricity used to power employee workstations, and the energy used to heat their home at times when they would otherwise be in the office, are attributable to Disco.

We conducted a survey to identify employees whose homes were not powered by renewable energy, and how their heat was supplied. We then worked on a basis of a 40 hour work week, for 45 work weeks per year, to get total electricity and heating emission estimates.

Our number crunching here came out with emissions of 1.64 tCO2e per year for electricity, and 7.04 tCO2e per year for heating, for a total of 8.68 tCO2e per year for our offices.

Equipment lifecycle (8.96 tCO2e)

In addition to the ongoing emission cost of running our workstations, there’s a a lifecycle emission cost associated with the sourcing, manufacture, and shipment of our hardware. These emissions are often much more significant than those generated by ongoing use, so it’s important to factor them into our overall footprint.

We source all of our hardware from Apple, who very handily provide Environmental Reports for each product they manufacture (example).

For the purposes of our 2020 estimate, we’re counting the lifecycle cost of all hardware in use at the company, regardless of the year it was purchased. Moving forward, we will count the lifecycle cost of each product as it’s purchased.

Our calculated 2020 lifecycle emissions were 8.96 tCO2e.

Cloud hosting (53.77 tCO2e)

We use two primary platforms to host our applications: Amazon Web Services and Digital Ocean. (We use AWS both directly through EC2, Lambda and API Gateway services, as well as through Heroku which is built on AWS.)

Our research has shown that it’s quite difficult to get accurate emission estimates for the various cloud providers, as they do not provide detailed energy usage information based on the type of virtual instance being used. Increasingly, a number of entire platforms (eg Google Cloud, Microsoft Azure) or regions within those platforms (eg US West on AWS or AMS2 on Digital Ocean) are being powered by renewables.

Given the difficulty of getting detailed information from cloud providers, we’ve chosen to take a simplified approach to calculating our cloud hosting costs that makes a number of assumptions, but errs on the environmentally conservative side of those assumptions at each step. The assumptions we’ve made are:

• All our our virtual server instances not located in regions confirmed by their provider to be 100% renewable are powered by 100% non-renewable energy;
• The carbon footprint of a virtual instance is 0.2987 tCO2e/year per virtual CPU and 0.0747 tCO2e/year per GB of available RAM. These numbers are derived from a number of sources, including AWS’s own sustainability pages, green computing blogs, and the LCA assessment of servers from manufacturers like Dell. These estimates also include the “embodied” emission cost of cloud resources (the emissions generated in the manufacture of the hardware involved), assuming a 4-year replacement cycle.

Using these (admittedly very rough) figures, our cloud hosting emissions from platforms not known with certainty to be powered by carbon-neutral energy were 53.77 tCO2e.

We have sanity checked this number against other companies that have published their estimated hosting emissions and are confident this is a very safe upper bound for Disco when adjusting for our relative size and hosting needs.

Travel (6.98 tCO2e)

Business-related travel during 2020 has been extremely limited due to the COVID-19 pandemic. In the normal course of operations, Disco does see employees traveling domestically and internationally for the purposes of company gatherings, conferences, and events.

To date, Disco flight travel emissions have always been offset directly with our airline of choice (Qantas) when purchasing tickets. We estimate our 2020 flight emissions at 4.28 tCO2e, recorded here for the purpose of tracking flight emissions over time, but we do not include this figure in our total emissions requiring offset.

Historically, we have not tracked or offset emissions from ground transport related directly to Disco’s business activities. That transport has been extremely limited in 2020, so we use a very conservative estimate of 1,000 kilometers of rail travel and 500 kilometers of car travel for a total of 0.22 tCO2e.

Likewise, hotel stays have been extremely limited in 2020. We used an estimate of 0.02 tCO2e for each night of a hotel stay for a total of 0.20 tCO2e, bringing our total business travel emissions to 0.42 tCO2e.

We had an internal discussion on whether to include personal travel to and from the office in the course of normal business. There’s an element of personal choice and circumstance around how someone travels to work that feels outside the remit of a workplace, but it can’t be denied that Disco is encouraging that travel in the first place.

In the end, because work-related travel emissions were so low in 2020 due to COVID-19, we decided we’d use a rough upper bound on emissions linked to travel to/from the office, offset that, and make this more of a focus for 2021. To get that upper bound, we assumed all employees drove a mid-sized car to and from work five days a week for 20 weeks of the year, at a distance equal to the employee that lives furthest from the office — a pretty generous bound considering most employees walk, ride or use public transport and live closer than 25km to the office.

Plugging these numbers into a calculator gave us 6.56 tCO2e for the year, taking our overall total travel emissions to 6.98 tCO2e.

Buffer (7.50 tCO2e)

This is the first time we’ve undertaken a carbon footprint assessment, so we have likely missed something. We have tried to err conservatively on all metrics, but to account for the unknowns, we’ve added a buffer of approximately 10% of our other estimated emissions.

For 2020, that buffer is 7.50 tCO2e.

Exclusions

There are some things we know we haven’t taken into account, either by choice or because their overall contribution to emission totals are so small we are confident they are covered by our buffer. These known exclusions include:

• Embodied emissions for non-workstation physical items, such as office furniture, merchandise and gifts. These are very minimal for Disco’s operations.
• Emissions from third-party software vendors. We use a large number of third party software services in our business, and we’ve seen some other companies consider their hosting emissions as part of their own footprint. We’ve excluded these from consideration as Disco is likely a very small percentage of those emissions, and we have no way of knowing which hosting providers are in use or whether they are using renewable energy sources or not.
• Banking. Our business accounts are managed by the Commonwealth Bank of Australia, which does not have the best record when it comes to sustainable investment practices. We haven’t managed to find a good resource for estimating the indirect emissions stemming from our deposit accounts, but are keen to explore this further in 2021, either in the form of offsetting these indirect emissions or by moving to a different bank.

Total (85.89 tCO2e)

Summing the above, our total estimated carbon footprint for 2020 is:

• Offices: 8.68 tCO2e
• Equipment lifecycle: 8.96 tCO2e
• Cloud hosting: 53.77 tCO2e
• Travel: 6.98 tCO2e
• Buffer: 7.50 tCO2e

for a total of 85.89 tCO2e.

Offsetting our carbon emissions

There’s an ever-growing number of companies, technologies and brokers in the carbon market. While this is overall a good thing, it can be a bit bewildering for small companies like us unable to conduct detailed assessments of all possible projects or companies.

In the end, we decided to work with Pachama, a carbon offset provider focused on carbon capture through forestry projects in Africa and the Americas. Trees are currently the most scalable and efficient way to remove carbon from the atmosphere, and while there is a need for longer-term capture solutions down the road, the project bundles offered by Pachama were a straightforward way for us to offset our emissions.

Both Who Gives a Crap and Shopify, two companies we respect and work closely with, have offset emissions with Pachama, which strongly influenced our decision.

We purchased carbon credits for 86 tonnes of CO2 from Pachama at a price of USD$10.80 per tonne, for a total of USD$928.80. We’ve made our certificate available here.

Reducing our carbon footprint

An ongoing theme in our research for this project is that while carbon offsets are valuable, they are not a panacea. Reducing overall emissions is just, if not more, important.

Looking ahead to 2021, the biggest impact Disco can make in reducing our footprint is moving all of our cloud hosting to providers and regions using 100% renewable energy in their operations. While this won’t eliminate our cloud hosting footprint entirely (there will still be the embodied emissions in the hardware), doing this could save over 50 tCO2e/year.

For technical reasons, we are already considering a wholesale move to Google Cloud Platform (100% renewable) for all hosting needs — the environmental benefits reinforce this as an attractive option. In the interim, a less involved approach could be to move all instances currently hosted in non-green regions on AWS or Digital Ocean to green regions. Either way, there seems no reason for us not to aim to have all of our infrastructure hosted on 100% carbon neutral platforms by the end of 2021.

Outside of that, as we filter back into the office more frequently, the emissions attributable to home office usage (which aren’t applicable at The Commons) should drop. We may also consider encouraging or incentivising employees who work from home frequently to swap to a renewable or carbon-neutral energy provider.

Other initiatives

Putting together this report has sparked a few other ideas or initiatives that we’re keen to explore further in 2021:

• We often send physical gifts to employees, clients, and partners. We could do more to consider the sustainability and impact of those gifts when selecting them, and avoid purchasing merchandise or goods that won’t be used.
• We don’t currently have an estimate of the indirect impact of our banking deposits. We hope to clarify and understand this more in 2021.
• We could add some incentives for employees to swap to renewable electricity suppliers if they’re not already using them, minimising the impact of WFH.

References

These are some of the online resources we found helpful in compiling this report.

• Measuring Wildbit’s environmental impact (Wildbit)
• Towards carbon negativity (Basecamp)
• Our decision to make shipping carbon neutral (Who Gives a Crap)
• We need to talk about carbon (Shopify)
• Apple’s environment page (Apple)

Two books that were influential in putting this report together (if you’re in Melbourne, Australia, you’re free to borrow them from us):

• The Uninhabitable Earth by David Wallace-Wells
• How To Talk About Climate Change In A Way that Makes A Difference by Rebecca Huntley