Hi Everyone, with most technical platforms being i...
# general
s
Hi Everyone, with most technical platforms being internal with developers as their primary customers, curious to know what approaches did you take in measuring the value. Unlike external facing products, they may not always have direct revenue impact but are enablers in deliver faster and more frequently (ex: kubernetes platform). Did anyone use value modeling to measure and track value out of internal technical platforms? How did you get the buy in from executive leadership to secure, prioritize and measure value out of the platform investments?
s
There are business metrics you can use to drive the leadership alignment. DORA is a good way, for kubernetes, you can use cost and utilization, which is even more straightforward.
s
Yes, we used DORA metrics but realized that platform is just one of the contributors to those metrics. For instance we can have a smooth CICD pipeline to build/deploy and kubernetes platform for run time but if the team is delayed to due to bad requirements or quality they end up impacting DORA metrics. Also, DORA are lagging metrics which can be measured after the fact, right? I have seen leaders struggling to justify the platform investments in order to secure funding. I was curious to know what approaches helped in that case.
s
For your case, deployment failure rate could be one part of it. And you will need to divide this metrics into different reasons, and use that reason metrics for improvement.
Once you have a predictable deployment frequency, you will also need to setup a deployment schedule, so new requirement will also go along with the deployment window.
This will need to work along with product management team or the agile people
Platform engineering need to be defined as a product to incorporate business requirements and connect people together. Use collective voice to justify, metrics is a result of driving the org improvement
DORA is not a lagging metrics, it is a result metric and North Star to guide the direction, but not the only ones. For each organization, try to find the metrics already has a consensus in your organization, use that to drive.
For example, the deployment frequency called “version release time” here in Tencent, because one version can be released multiple times.