Do not confuse Input & Output metrics in your OKRs

At Velocity Global, we practice quarterly planning as a way of chunking work into manageable segments with tangible outcomes. A quarter is a Goldilocks amount of time — short enough that usually we can be deliberate in committing to clear outcomes, and not too long that the market or tech changes significantly requiring a course correction.

Quarterly planning is done using OKR’s: What are the objectives? What are the key results that get us to the objective? The key results — which are metrics — determine the goal and the progress towards that goal

An observation I have had is how people can confuse output metrics with input metrics. They are distinct, and confusing the two results in poor outcomes. In addition, much thought has to be given to the construction of the output metric and its relation to the inputs. Sometimes, you can get to the output metrics and it will be a pyrrhic victory

The second problem with a lack of understanding of output metrics and what they represent.

What is a good output metric?

Here are examples of poor output metrics:

“Build the new expense management feature”

“Complete the integration of accounting with core software”

Why are these bad output metrics? Output metrics must move business, customer or employee outcomes otherwise they are not good output metrics. In order to do so, the organization has to be aligned behind outcomes that ignores org boundaries: Engineering saying they will get something built does not imply that marketing will share it or sales will actually sell it. Outcomes that line up with customer outcomes force the organization to solve for the true metric that we need to solve.

So if we say ‘Build the new expense management feature’ will likely do exactly what it says: build the expense management feature. It does not care about whether the rest of the organization has lined up to support it and that can be fatal. The new invoice management feature gets built, and the team celebrates. In the meantime, there is no adoption because sales has not been informed about it. Customers are not adopting it because they did not really want this feature. Instead, if we rewrite the output metric as

“20% of customers are using the new invoice management feature with an NPS ≥ 20”

This is

1. Customer centric
2. offers a clear way to measure adoption
3. Is clear on the quality metrics expected
4. Forces alignment across the organization to get to the outcome.

You cannot get there without having alignment across all the different stakeholders.

Be aware of the importance of input metrics

While the focus is on the output, the input metrics — the ‘how’ — matters.

On one occasion, a leader was asked to staff up his platform team as the objective, and sent a key result of ‘hire X number of engineers’. So the team went out and hired. And then celebrated the successful completion of hiring X engineers. But wait — how was the quality assessed? Were the engineers a culture fit? Did they have the right growth mindset? Were they in locations we wanted to be in? None of that was managed. So the successful outcome was hiring the engineers — which was not a success, because six months later, many of them were let go because of poor fit

The inputs matter

When constructing your output metrics, pay attention to the inputs. Think about a fishbone, with the output metric at the head of the fish, and the inputs (e.g. location = North America and India, ratio of eng:sr eng: staff eng = 5:2:1, interviewers for technical, culture fit defined are possible inputs)

Here is what an example Fishbone might look like, based on the expense management output metrics of “20% of customers are using the new invoice management feature with an NPS ≥ 20”