feedback on Implementation Pillar

[seanmcilroy29]

Parul Jalota - Bloomberg

• There is a lot of focus on the set up, data collection, aggregation and validation stages. It can be very expensive to collect the metrics across the organization and ensure they make it into established databases. It would be helpful to expand more on how to establish the KPIs / OKRs i.e "What does good look like?" . Ex : SCI may increase due to pure business growth/adding more users but could also increase due to a sub optimal implementation. Some examples on - what a good KPI/metric an look like will help.

• A lot of companies are tracking carbon emissions at an organizational level, and have strategies to reduce by virtue of switching to renewable energy / purchase of PPAs. But that is not enough, we need to go more granular than that. We can start by breaking down emissions as described in Stage 7 - but this will require a lot of coordination and collection efforts from various systems and teams. Our focus instead has been how can we find ways to make an impact as soon as we can i.e. can we track alternate more granular improvement metrics even though we haven't yet been able to tie back to scoped emissions. Ex : efforts to reduce idle services ? Number of services migrated to cloud ? Etc.

• Post tracking the above granular metrics, we want to "automate" changes to the systems instead of "providing recommendations" as we are not sure recommendations are always accepted.

[PindyBhullar]

Thank you for the thoughtful feedback. You’ve raised some excellent points that are crucial for refining our approach to sustainability metrics.

Establishing KPIs/OKRs: I agree that defining what "good" looks like is a critical piece of the puzzle. We can start by framing KPIs around both desired business outcomes and technical implementation quality. For example, the SCI (Software Carbon Intensity) can indeed increase due to business growth, but to distinguish between growth and implementation quality, we could introduce a secondary metric like the efficiency ratio (carbon emissions per user or transaction). This would help pinpoint whether emissions are rising due to suboptimal implementation or genuine business scaling. We could further refine KPIs by incorporating benchmarks and thresholds for acceptable emissions relative to business size, as well as the quality of implementation (such as server optimization or load balancing). I will add examples like this to clarify how good KPIs can be structured.

Granular Metrics: Your point about going beyond organizational-level emissions tracking is valid. While many companies have adopted high-level strategies like renewable energy adoption, the real impact will come from addressing granular metrics like service optimization, cloud migration, and idle service reduction. Tracking these can not only provide immediate results but also lay the groundwork for more accurate emissions tracking. An example metric could be energy efficiency per service or energy consumption per transaction. These granular metrics can be a bridge until more comprehensive emissions data is available.

Automation vs. Recommendations: Automating changes based on granular metrics is a proactive approach. We could integrate this by establishing triggers for automatic optimization, such as reducing idle time for cloud resources or automatically scaling services based on usage patterns. This would align with a more agile sustainability strategy, where actions are taken automatically to improve efficiency without relying on manual intervention or recommendations, which might not always be accepted.

[birtheb]

I especially like the “granular metrics” approach because it is really close to reality. At least in bigger organizations, it is definitely necessary to aggregate or break down metrics.