Data analytics in its many forms has crucial roles to play in enabling the achievement of Net Zero but should not be considered a silver bullet. These roles include collating and integrating data; modelling data; reducing uncertainty; understanding current state; trend analysis; predicting outcomes; and holding decision makers to account.
All of these roles have challenges, given that analytics are generated in a broader context of human decision making; societal judgments; competing priorities; disparate data (if held at all); and politics. This context can mean that even the best, most robust, analytics can struggle to have the desired impact.
What is Net Zero?
Net Zero is defined by UK law as “the net UK carbon account” and is based on a commitment to reduce emissions by 100% than the 1990 baseline.
This requires countrywide accounting mechanisms which, as we have seen with UK economic statistics, is not a straightforward or completely accurate exercise with statistics taking months to prepare and still being subject to revision for over a year after release.
Carbon accounting offers many of the same complexities and adds more – for example when calculating the UK trade balance imports and exports have a financial value attached to them so their impact can be incorporated. How can we determine carbon flows between countries?
'Useful' analytics
How useful analytics results are depends on the quality of the underlying data and any assumptions applied to that data.
DAMA describe six key aspects of data quality:
| Aspect | Definition | Net Zero Challenges |
| Completeness | Are all records present? | How do we collate carbon impacts of every activity in the UK and every activity carried out abroad for the UK? |
| Uniqueness | Is there any duplication in records? | Accounting for carbon lifecycles of products and services means potential duplication hidden within data collated. For example, a petrol retailer could account for customers use of fuel purchased while the customer also accounts for it in their business mileage. |
| Consistency | Values in a data set should not contradict other values representing the same entity. |
Does every gallon of petrol with the same octane rating have the same carbon value in all reported datasets? |
| Timeliness | Is the time lag between collection and availability appropriate for the intended use? |
Introducing carbon reporting is a multi-year process requiring agreement of standards, information dissemination and time for implementation. |
| Validity | Is data in the range and format expected? |
Is carbon being reported under the correct definition? Based on ONS reporting, in the UK in 2016 the estimates of UK greenhouse gas emissions in 2016 ranged from 473m tonnes on a territorial basis to 784m tonnes on a footprint (consumption) basis with the bulk of the difference being composed of air travel and imported emissions. |
| Accuracy | Does the data match reality? | It is not technically feasible to measure carbon flows with the same level of accuracy as financial flows. |
The roles of analytics in net zero
In 2024, most people think of analytics as AI and machine learning yet if we define analytics as “the pursuit of extracting meaning from data” many more opportunities arise.
| Role | Net Zero |
| Collating and integrating data |
Applying standards to the collection of data to improve quality. Automatically ontologies to further improve quality. An ontology is a set of categories that help classify items - for example animal, vegetable, mineral. |
| Modelling data |
Using actual data to create assumptions that can be used to create digital twins to test policies, incentives and optimisations. Using machine learning to identify patterns in the data that would not necessarily be visible to the human eye. |
| Reducing uncertainty |
Combining the information derived from the other roles provides a consolidated view. |
| Understanding current state |
Using the data derived from the other roles and required under regulations such as the EU Corporate Sustainability Reporting Directive. |
| Trend analysis and outcome prediction |
Machine learning, stochastic modelling, Monte-Carlo simulations and Bayesian modelling can be used to generate trends and uncertainty bands, providing forecasts for different scenarios to support decision makers. |
| Report writing |
Using generative AI to quickly provide first drafts of information and analysis. |
| Holding decision makers to account |
Publishing agreed data along with the agreed targets enables detailed analysis of what went well and what needs further work. |
Challenges in applying analytics to Net Zero
Even if the underlying data is of sufficient quality and can support the various roles, there are still many things that would stop the analytics enabling delivery of Net Zero. These include:
| Role | Net Zero |
| Carbon emissions of analytics |
Storage and analysis of data is not carbon neutral. An International Energy Agency report shows that data centres generate 3.5% of global greenhouse gas emissions. And this does not account for the energy required to power the computers, networks and peripherals required to transmit and make use of that data. |
| Human decision making |
Human decision making is prone to confirmation bias. This means that people look for data which supports their beliefs. If someone is a climate change sceptic, then analytics alone are unlikely to convince them. |
| Competing priorities,, politics and social judgments |
No matter how compelling the analytics (and the famous climate stripes chart created by Professor Ed Hawkins at the University of Reading is about as compelling as they come!) any action still requires a positive decision to be taken. There are many competing priorities for attention and funds. There is only a finite level of both available and if these are consumed by immediate problems rather than long-term issues such as Net Zero then nothing will happen. |