The Environmental Cost of All That Digital…

When it comes to environmental cost. It is with a small weight of mental dissonance for me, this post is pushed to the internet ether alongside roughly 7.5 million others*. My deliberately selected keywords should make it stand out, pushing aside most other posts of the day? Maybe, just maybe, I bring such a fresh perspective to this topic it will trend above all others. Doubtful.

Did my post even get 0.01 seconds** of its own limelight, for all the digital resource consumption in achieving its publication?  It’s a nailed-on certainty I have missed it, as has everyone else!

Not dissimilar to the fast-moving consumer goods sector, for business and consumers alike we satisfy our cravings from the end of a complex digital supply chain. The challenge of understanding our digital resource consumption still grows, with the (digital) sector increasingly clogged with companies touting their own tools and techniques for a sustainable way forward.

A recent environmental cost whitepaper from Google Cloud Platform (GCP) trumpets their carbon footprint calculator in line with the Greenhouse Gas Protocol (GHGP), but the boundaries of their calculations fall short of allocating the sunk environmental costs of building the datacentre from which their cloud service(s) operate. The path to transparent data-driven decision making, especially for something as top-of-mind as sustainability and climate change, should be clear. It is not!

Countless such examples exist.

Nonetheless, each time we ask a computer to look something up for us, we make our own contribution to the environmental debt pile. If we’re to sustainably continue in a digital world we are compelled to guide behaviour through:

  • measuring what we consume (metrics the key to reducing the environmental impact of digital activity)
  • clearly conveying how digital (habits) can be sustainable
  • continuously encouraging participation in sustainable effort(s) at every level of business, right through to the consumer

There’s a +20-year history of the UNFCCC’s efforts to garner support around environmental key performance indicators (KPIs) with a top down approach. Though KPIs are easy to state, they are challenging to reach consensus thereon and trickier again to implement consistently; successful delivery and long-term practice calls for additional conditions to exist. For example:

  • effective incentives (e.g. executive compensation) as an agent of change
  • individual awareness of how our contributions produce relevant direct benefit(s) and support adjustments in our behaviour / habits for the better
  • for digital sectors (and transformations), the empowerment (skills and freedom) to broadly implement measures and embed delivery practices

Sustainability will not always be the main goal, yet it’s increasingly likely to affect the business’ license to operate. Exchanges (e.g. NASDAQ, ASX) and financial analyst societies (e.g. EFFAS) have, since years, published ESG frameworks for incorporating sustainability KPIs into corporate valuation. Growing recognition of this at the market level indicate that executives (C-Suite and directors) at the enterprise (performance) level are increasingly measured against these sustainability KPIs. Over time, poor performance against these should transfer to lower corporate valuations (public or private), likely increased cost of borrowing and resulting lower executive compensation.

As market sentiment and, more directly, executive compensation, drive change at the enterprise level this transfers to staff through mandate or practice. Projects and technical delivery tend to then implement such change, but lasting practice should be an outcome of such change for example:

  • digital (transformations) should include as practice the measurement of application performance (e.g. DataDog, AppDynamics), infrastructure costs (e.g., native cloud cost estimators), alongside the likely environmental costs of the software and workloads. Marmelab, for example, have made strides with their CLI which translates certain research based methods of compute usage into a carbon footprint.
  • with company owned datacentres, this is an opportunity to reach across a traditional (corporate) divide and ask your datacentres manager for DCIM metrics (e.g. Nlyte, Sunbird)
  • where a project must work with the (major) cloud vendors, more than just a published PUE metric is necessary to get a better understanding of associated environmental costs, so applying some pressure to relationships with these vendors is often required for progress in this area. In the relationship with cloud vendors, it’s important to be clear that virtualization can be a useful tool for workload optimization, but cloud is a distraction for a true understanding of sustainability.
  • individual contributors (e.g. DevOps, engineering, testing) are encouraged and should feel empowered to learn and practice sustainable software development practices, which open up opportunities to save energy along the software development lifecycle. For example, Ecometer, supported and designed by a consortium of French web agencies attempts to analyse the eco-maturity of websites, Intel provides the software development assistant (Intel SDA) and if you’re developing at the interface of software and hardware (e.g. IoT) then Otii Arc from Qoitech is an example for keeping you straight on power consumption.
  • after successful delivery, there’s every reason to extend service excellence KPIs to include sustainability metrics as a practice into the future

But what does all this cost environmentally? In an effort to encourage personal responsibility in sustainable digital habits how do we make this generally meaningful?

In his book “World Wide Waste” Gerry McGovern calculates this in terms of trees. Extending his example of offsetting email spam with 1.6 billion trees, and allowing for variation in tree-planting density per acre, these trees cover Greater London between 5 and 13 times. For the lower end of that range (higher tree density per acre), this translates to approximately 7-8 feet between trees. This is just enough space for your average basketball center to lie down in.

This is a superficial example, but individuals are prompted into action only when their personal value system is somehow threatened. Offsetting, by and large, is not effective in this regard.

Afterword: In an effort to be as environmentally conscious as possible in writing this post, no images are used. Text is the most digitally and environmentally efficient means of making a point. I trust my words are chosen conscientiously enough.

* 7.5m blog posts published daily (

* 86,400 seconds per day / 7,500,000 article = 0.01152 seconds

You can find Justin on LinkedIn here

Similiar Articles


Sign up today for monthly newsletters containing:

  • News and insights from your industry
  • Relevant thought leadership articles
  • Engaging video content
  • Notifications of our upcoming events
  • Networking opportunities with C-Suite leaders