Dark Data
In the physical world we care about cleaning up after ourselves; if we see someone throw litter on the pavement, we get annoyed. But in the digital world we do not see the rubbish we leave behind. Dark Data is the term for information that is collected, stored and organised, but never used. In the physical world it is called hoarding. Dark data is digital hoarding, and it is costly in many ways.
How is dark data created?
Dark data is created by almost every digital interaction. Here are some examples:
Customer logs and web analytics: Every click and action is stored for future analyses that rarely happen.
IoT & streams of sensor data: Thousands of sensors in smart factories or cities stream data 24/7. Often only 1% of it is used for analysis.
Duplicate files: Several versions of the same presentation or document stored across different data centres (which some still call the Cloud)
Old communication: Attachments in Slack, Teams or old e-mail threads that are never opened again.
Redundant, obsolete and trivial (ROT) data: Duplicate files and former employee accounts remain active «just in case».
We never delete data
A worrying feature of dark data is its scale and permanence. According to the Veritas Global Databerg report, an estimated 52 % of all data held by organisations is «dark», and a further 33 % is «ROT» (redundant, obsolete or trivial). Only around 15 % is identified as business-critical.
Because cloud storage has historically been cheap, companies rarely put deletion protocols in place. This data sits in «cold storage», where it still needs constant power for servers and cooling systems. And hardware and software must be upgraded regularly to keep storing it.
Digital smog: the CO2 comparison
Data may feel weightless, but the infrastructure needed to store it requires huge amounts of energy.
Power consumption: Data centres account for roughly 1 % to 2.5 % of global greenhouse gas emissions, a footprint now on a par with the aviation industry.
The car comparison: Storing dark data alone creates an estimated 6.4 million tonnes of CO2 a year. To put that in perspective, research from Tessi and academic studies suggests that if the growth of unused data continues unchecked, emissions from dark data could scale to match the impact of 80 million fossil-fuelled cars driving for a whole year.
Key statistic: Digitalisation was estimated to account for 4 % of global greenhouse gas emissions as early as 2020, and with the rise of AI-generated data in 2026 this figure is under growing pressure. As Jason Miklian says in this post, AI-generated content is a danger to democracy.
Why does it matter?
Dark data is not only an environmental problem, it is also a security risk. Every gigabyte of dark data is a «blind spot» where sensitive information – such as PII (personally identifiable information) – may be hidden and represent a security breach.
To solve this we need «digital decarbonisation». As experts at Loughborough University have written, businesses must move from data hoarding to data hygiene. Deleting an old e-mail or emptying a redundant cloud folder is a small but necessary act to protect the environment. And it is good for the wallet. Who wants to pay for something they do not use?