OKRs v KPIs - How they work together
TL;DR - There is a natural overlap between key results and key performance indicators. A metric can move in either direction depending on the focus of the business and the state of the metric.
KPIs and KRs (the Key Results of OKR) are powerful tools, but I’ve seen an increasing amount of bad advice on the subject. In a desire to make the definitions of KPIs and KRs clean, many observers state that a single metric is either a KR or a KPI, but it can’t be both.
As a definition it keeps things simple, but the distinction is artificial and inhibiting. The main difference between a KR and a KPI is timing and context. Let’s pick this apart with a question.
What job is your metric doing?
I believe there are two main jobs that a metric does in this context.
- Represent the current measurable outcome we are working towards.
- Provide a measure of something that is important to the business, but is not the current focus of your work.
OKR is concerned with number 1 and key results are the metrics you are currently focused on impacting. This is your most important work. Ideally you are reviewing these on a weekly basis.
KPIs are concerned with number 2. Felipe Castro calls them Health Metrics, because they tell you if everything is okay with the business.
The overlap
My argument is that a particular metric can be used for 1 or 2, depending on the focus.
Let’s work through a simplified example.
This is applied to a B2B SaaS I’ve worked with, but could apply to many different products.
This is a mature product and customers are loyal. Churn is less than 5% and we’re happy as long as it remains under 7.5%. We’ve looked at the data behind churn and decided there are too many different factors to make reducing it further cost effective right now.
Naturally we keep an eye on it, every month we look at the churn data and it’s on senior leaders’ radar. It really is a Key performance Indicator.
Our current focus is getting more website visitors to start a trial of our product and our OKRs reflect that priority.
A few month’s later we’ve made good progress on acquisition but churn has ticked upwards. It’s now 11% and it is seriously jeopardising our bottom line.
Time to focus on that KPI
Leaders are concerned and want to focus on this problem. At the next OKR setting a cross-functional team is charged with reducing churn. The logical decision is to make this KPI a key result.
That team includes a product person, an engineering lead and a service desk person. Unless the team stumbles upon an obvious cause and finds a very quick solution, they will treat churn as their key result for a period of time. Identifying and testing hypotheses as they go. As a strong team they are considering several opportunities and solutions at once.
After some great product discovery work the team makes some changes and churn is reduced. Please note, in the real world, churn is usually a lagging measure. To counter this the team may identify one or more leading metrics.
Carry on tracking, but focus elsewhere
With the problem solved sufficiently well, the team can focus on a new problem. Churn goes back to being a KPI and the leading metrics the team identify is added to the set.
The point I am making with this example is that KRs and KPIs are tools. They either help you:
- keep a track on things that are important but not a strategic priority, or
- track progress towards current strategic outcomes.
In either case they’re an important measure of business success. In a future post I will go deeper into this example to explore the connection between leading and lagging measures. It will also help us understand how organisation and team key results can connect.
An engineering focused example:
Load time for our search page is a KPI. We’ve got a sound architecture and this metric has been within threshold for many quarters. We then see load time has crept up and our own research tells us that this is a leading indicator for reduced usage. We make reducing page load time a priority for the next quarter and a KR is created.
A cross-functional team quickly identifies a few contributing factors and tests and builds the required solutions. They soon have the page load time back where it needs to be and the metric is back as a KPI only.
Hopefully these couple of examples help explain why there can be overlap between KPIs and KRs. Quite simply we should only be looking at metrics that are important. If something we track (a KPI) is critically unaligned, then we should focus on it (a KR). No need for dogma.