A Practical Guide to OKRs

Background: Goals vs. OKRs

Most people have some experience setting goals, either on a professional or personal level. A commonly used framework for goals is “SMART”: Specific, Measurable, Attainable, Relevant, Time-bound. An example of a SMART goal is:

Goal: Reduce my body weight from 86kg to 82kg by March 31st, 2021.

An OKR is simply a SMART goal where you split the “where I want to go” (Objective) from the “how will I know I got there” (Key Result):  

Objective: Lose weight Key Result: Body weight from 86kg to 82kg

This separation allows you to list out multiple initiatives/key results for a given objective, for example:

Objective: Lose weight Key Result: Body weight from 86kg to 82kg Key Result: Exercise for 1 hour, 3 times per week on average Key Result: Reduce average weekly sugar intake from 1kg to 500g

Key Results can become inputs to Objectives for other teammates. For example, based on the above, my nutritionist could take an Objective to reduce the frequency of sugar cravings through adjustments to my diet. Through this alignment structure, large teams can decompose and work together to deliver massive objectives.

The best full primer for the OKR system and methodology is the book written by the creator/promoter of the system: Measure What Matters by John Doerr. However, I have found that there is ittle guidance out there for my next two topics, which can make all the difference between OKRs that help the team focus, and OKRs that are forgettable.

Inputs vs. Outputs

In any goal framework, there are 2 types of goals you can set: input goals and output goals. This is probably the single most important (and probably overlooked) part of setting effective goals, and where context and judgment matters most. Let’s look at an example that illustrates this difference, specifically with key results:

Objective: Increase Rentable Units in all cities
Key Result 1 (output): Acquire 10,000 new Rentable Units with an average publishing time of 4 hours
Key Result 2 (input): Launch a new self-service property listing experience for landlords on mobile app and desktop

Which one is better? Most people would gravitate to the first one, but the second one is almost always a better choice. The first key result is centred around the desired output of what you work on. The second key result describes the changes to the controllable inputs you will execute in order to achieve the objective.

Both key results are measurable and in support of the specified objective. But both key results ultimately depend on controllable inputs, even though the first one only lists the desired output. If last quarter we acquired 5,000 rentable units doing things the way we’ve always done them and you sign up for the first key result above to double that, it’s more of a wish than a goal. However, if you have 3 quarters of experience launching product improvements and acquisition campaigns that led to lifts of 50-80% in rentable units each quarter, then signing up for this key result makes much more sense. While it’s a stretch goal, it’s clear you’ve identified and driven inputs in the past that impacted the desired output. Understanding when to use inputs versus outputs will enable you to write a key result that drives clarity, purpose, and commitment, versus a key result that you set and forget.

The following sections provide guidance on when to choose outputs versus inputs for your key results.

Output Key Results

Output-oriented key results are very easy to correlate to objectives and can be very motivating and inspiring, which is likely the reason they are most commonly used. Output-oriented key results are also the most common type of poorly defined OKRs: there is a fine line between a stretch goal and a wish list. The most important requirement of an effective key result is that you actually feel in control of the outcome. Key results such as “double the company’s revenue” may be bold and inspiring the day you set them, but don’t change behaviour, help make tradeoffs week to week, or provide transparency into the inputs you’re actually working on in hopes of achieving it (which your teammates and stakeholders may not agree with you on, and you won’t even know).

Output key results are effective when: The corresponding controllable inputs are clear and obvious, and/or There is history and data that shows you actually control the output (within a reasonable error margin) Let’s take the following output-oriented key result as an example:

Key Result: Hire 15 engineers in UK and Europe in Q2

Applying the two guidelines above: 1) It’s clear that the inputs here are sourcing and interviewing candidates, making offers, and ensuring they start as soon as possible. 2) We know from the past that recruiters average 2 accepted offers per month, we have 2 recruiters, and 3 months in the quarter: 2 * 2 * 3 = 12, so stretching to 15 is reasonable.

Now let’s look at another example:

Key Result: Complete 2600 Bookings on a mobile app in Q2

Applying the guidelines above again:

1. The inputs here could be anything - product changes to the mobile app, marketing channel mix changes, business model changes, booking process improvements, new supply strategy, etc.
2. Unless you have 1-2 quarters under your belt changing inputs that you clearly attributed to closing bookings on a mobile app, you’re not in a position to sign up for this output; you don’t control this outcome.
   

In situations such as this last example where you have an output-oriented key result for which you can’t defend 1) and 2) above, you need to identify the controllable inputs that you will execute against to achieve the desired output, and define your key result(s) based on those. The next section provides guidance and examples.

Input Key Results

In situations where you want to impact an output but it’s unclear how (a common example is financial outputs such as revenue), focusing on controllable inputs in your key results is the most effective strategy. Key results focused on controllable inputs are the best option when for a given output, nobody knows what inputs matter most. Setting key results based on the controllable inputs and later monitoring the impact it has on the outputs will enable you to learn what inputs matter most. This is a process that has been effective at Amazon for more than a decade.

Taking the output key result from the previous section again:

Key Result: Close 2600 Bookings on mobile app in Q2

Today, it’s unclear what inputs will have the biggest impact on closed bookings originating from booking requests on the mobile app, so taking this goal is a good intention, but not much more. A more effective approach is to set one (or more) key results centered around controllable inputs. For example:

Key Result: Launch 5 or more mobile app features that target the top 3 customer pain points from user research studies Key Result: Run 3 experiments that change the payment amount due by the tenant upon Booking Request Key Result: Launch an MVP of a monthly payment system by June 1st

The above key results are all targeting more bookings on mobile apps, and while we don’t know how much impact each will have, we fully control the success of these goals. We can provide data and reasoning that correlate these inputs to closing Bookings on mobile app.

Input goals are much harder to set than output goals because they require you to outline your assumptions and high-level strategy to achieve a given output such as closed bookings. This requires talking to more people, getting feedback, and engaging in healthy debate. Since the most important inputs are unknown, your peers and stakeholders may have different assumptions and strategies than what you propose, and this discussion and subsequent alignment is arguably the most valuable thing OKRs drive.

Explaining How You Measure

All properly defined key results are measurable, either through a metric or through a clear binary outcome (e.g. this happened or it didn’t by the said date). While it’s good to have clear and concise OKRs, how you measure each key result is worth elaborating in a column of its own to ensure that everyone is one the same page with respect to what this goal really means. Oftentimes, what needs to be measured may not be possible and requires some work itself, which in turn becomes a key result you depend on from a partner. Sweating the details and insisting on the highest standards when it comes to how you will measure your key result will eliminate ambiguity and sudden misalignment down the line.

Even goals that seem obvious in how they’re measured can be ambiguous. For example:

Key Result: Hire 15 engineers in UK and Europe in Q2

What counts as a hire in Q2? Does their start date have to be within Q2, or just offer accepted? What about candidate offers accepted in Q1 that have a start date in Q2? Creating a supplemental “How This Result is Measured” validates that 1) there is a mechanism in place to measure it, and 2) it’s clearly understood how it works:

How This Result is Measured: Start dates for hires counted toward this goal can extend one month into Q3 if the offer is accepted and the contract is signed within Q2. New hires that join in Q2 with offers accepted in Q1 will not count toward this goal.

Note that this level of detail would not be automatically inferred by everyone reading this key result. With this detail, other team members can align and set their key results properly, and when we report on the status we know clearly how it’s measured and if it’s on track.

Conclusion

The OKR methodology is only as effective as the quality of the OKRs themselves. The process of defining great OKRs is as valuable as the output. Writing effective OKRs should feel hard. It requires a lot of thought, discussion, and judgment to 1) identify and clearly specify what controllable inputs you are going to focus on, and 2) achieve the necessary specificity and detail on how you’re going to measure progress. There is a fine line between a set of challenging stretch OKRs and a wish list. Make sure that you’re setting goals where you feel you’re in control, and it’s clear what your controllable inputs are. The best measure of a great OKR is if half way into the quarter, we can definitely say whether a given OKR is on track or not, and what we can do to change its course. If you or others aren’t sure, then the OKR needs more work. Execution is everything, and it begins with great OKRs.