OKRs for platform work
Measuring success for platform work can be difficult, especially since these projects are often take longer and are further away from end users and business impact. Below are some ideas on how you might come up with good OKRs for your work. If you need to measure progress of a 2-3 month project within a quarter you can probably take combine of the above into milestones and measure if you hit 4/5 milestones.
Efficiency
If feasible measure the time it takes to complete a few typical tasks or features before and forecast the time saving your new APIs will build. Your goal can be to reduce completion time by 20% on sample tasks. You’ll obviously want to get alignment on what are good sample tasks to measure.
Planning milestones
As architectural designs are completed reviewed and approved. Sometimes your plan is to have a plan by a certain date. This can include alignment milestones from internal stakeholders
Survey oriented KRs.
If you’re early you can get feedback from internal teams to see if your work is positively affecting them, even if the whole thing hasn’t shipped yet
Completion
As end points become available you can count those towards KRs. You can track stubbed end points separately if that’s a meaningful milestone.
Adoption
Adoption KRs as teams start using your new API. You should make sure teams are lined up to use what your build ASAP and this makes sure it isn’t something shiny but unused
Usage
Usage KRs. This can be % of new APIs used over old APIs if you’re replacing an old API. Or volume of traffic goals
Enablement KR.
Any amount of revenue engagement or other business impact you’re able to enable or unblock. This can help capture any interim workarounds your team needed to build for other internal teams.
Rubric KRs.
Grade various aspects of your API, (e.g. performance security, extensibility, maturity) on from A-F. Have multiple senior engineers grade independently and resolve any grades off by more than one letter. Put a KR to improve certain aspects from say D to B. You can also break this down into say performance by component or extensibility for X and Y that are in high demand.