Demystifying Objectives and Key Results (OKRs)

There is a whole book written by the Oracle of OKRs (“Measure What Matters” – John Doerr), and how it made a “10x” difference to companies like Intel, Google, The Gates Foundation, Intuit and several startups at a much earlier stage of their evolution. That’s the best place to go for inspiration on why OKRs matter and how they help.

In my personal experience, OKRs helped me in three specific ways:

• To simplify and clarify my own priorities in actionable and measurable language (rather than a complex strategy document or mission statement that few understand practically)
• To translate the 3-5 priorities each quarter into an org-wide agenda. So everyone knows what we are going after, how we plan to get there and where they fit
• To make the hard trade-offs. Capacity is not infinite. Continuously adding “urgent” priorities confuses an org and detracts from building real capability (and perhaps most importantly makes it harder for team members to deliver and feel a sense of achievement). Adding something must lead to removing something.

If you think your org needs some version of these benefits, read on. This short document is intended as a workmanlike “playbook” for OKR implementation – based on my personal experience (at Flipkart and Ola), the experience of friends (thanks!) across several startups, and VC firms who have implemented OKRs with varying degrees of success. Thanks also to friends from Google (today’s global inspiration for OKR implementation) who happily shared their experience after only a few reminders! 

For a simple example of what an OKR looks like – to clarify “O” (higher level goal) vs. “KR” (specific, measurable steps to get there), here’s an example:

OKRs for everlasting youth

Objective: To appear at least 5 years younger than my peers, as measured by

Key Results:

• Use a 5-year-old photograph as my Facebook/ WhatsApp/ LinkedIn profile photo
• Drink two glasses of coconut water/ warm milk every day
• Do not consume high sugar/ caffeinated drinks and processed foods
• Use Bombay Shaving Company for personal grooming, Upstox for stock trading, Darwinbox for HR, UrbanLadder for furniture, OYO and Treebo for vacations, and Xiaomi for smartphones. This won’t make me look younger but all the young founders at these companies will say I do
• OKR (verb) all the important things I want to get done. If it doesn’t matter, don’t do it – that’s probably 70% of what I do now. Reinvest the time saved in getting 8 hours of beauty sleep every day

As the example shows, having personal OKRs can also help to create the habit (e.g., “O” – Get fit; “KR” – Swim 40 lengths, 5 days a week) – worth a try!

The boring stuff: How to run a typical OKR cycle (T = start of the quarter/ year)

1. Evangelise: Explain why this matters and how this will help everyone (has to be the Founder/ CEO, with all senior staff in sync)
2. Identify an OKR shepherd: If you don’t have a Chief of Staff, nominate a CXO as OKR manager for each cycle (can be done by rotation) – without a senior shepherd, OKRs are unlikely to succeed especially initially. Side note – why don’t you have a Chief of Staff?
3. T – 4 to 6 weeks – Brainstorm: Discuss company goals in the top team. End product – a simple statement of what we would like to achieve as a company. Use the company mission, strategy, inspiration from competition or high-impact problem statements as a good place to start defining company level OKRs
4. T – 4 weeks – Involve teams to flesh out high-level goals into focused objectives and measurable KRs: CXOs involve their respective teams. Important to focus – “less is more”: 3 to 5 Os per cycle (at company, team and individual level), with 3 to 5 KRs each is a good number. All Os and KRs should be objectively measurable.
5. T – 2 to 3 weeks – Crystallise company level OKRs and communicate: Workshop to crystallise company level OKRs with top-down guidance on priority/ stack-rank. Communicate these OKRs to all teams/ team leaders.
6. Detail: Teams work out the next level detailing of their respective OKRs (this would have started with step #4) including resource requirements and rough timing. Typically a “KR” at one level will be an “O” at the next level. Horizontal OKRs (e.g., a financial goal or people goal) can be reflected across multiple teams. The KR detailing must include specific and complete answers on “how we will reach this key result”. (This is what makes OKRs different from goal setting!)
7. Align & break open dependencies: Call out clear dependencies on other teams, respond to incoming OKRs (for your team). Hold a dependency breaking meeting for all multi-team OKRs (usually helps to have a Chief of Staff/ CXO who can tie-break in charge of running this half-day meeting).
8. T-1 week – freeze: teams freeze their respective OKRs along with resource allocation and publish
9. Track & adjust (in-quarter): Ideally weekly, with a more formal check-in mid-quarter, feel free to adjust as you go along (OKRs are not cast in stone). Scoring and self-assessment are critical to drive the process to maturity (0 = no progress, 1.0 = fully achieved). Good scoring must involve judgment, not just mathematics.
10. Reflect and learn: The All Hands is a great place to openly share how the company is doing on its OKRs and to encourage all individuals to do the same for themselves. Use the OKRs as one input for manager – team member feedback conversations.

You may choose a different frequency (from quarterly) for your company – one quarter tends to be a good balance between annual (too long, too many things change) and monthly (too much overhead to do an OKR exercise every month!).

Checklist – “Dos”

• Differentiate between 100% vs 70% expectation: Differentiate between commitment OKRs (typically BAU, have to deliver 100%) where the path is relatively clear vs. aspirational OKRs (path not fully clear, attempted “10X” breakthrough) where a 0.7 score is “green”. It is helpful to define aspirational OKRs over multiple quarters.
• Use OKRs for performance management: As a part of it, not all of it. Use as a source of objective data about what has been accomplished. Performance management in addition should include feedback on leadership behaviours, competencies, and stakeholder input. Do away with bell curves once you have objective data on these (and OKRs can be a big part of having objective data) – but that’s a whole different subject!
• Ensure cross-functionality: The real nuance in OKR definition is to get ops, product, data science, technology, finance, HR all into a coordinated effort to chase big, ambitious goals, even if across different time horizons. It is very easy to default to operational target setting only (for the immediate quarter) and think “I have a great OKR process”. You’ve got it right when all teams have a clear understanding of goals and are delivering in sync across different time horizons. Getting this right also depends on how the org is structured (e.g., pod structure for functional orgs).
• Leave meaningful room (and time!) for bottom-up: vs. doing OKRs only top-down. It is helpful for all team leaders to have some time with the company-level OKRs to create room for ingenuity and dialogue. Even if the OKRs are set top down, there must be room to discuss and adjust – a lot of the richness comes through this process. Provide time for iteration – don’t rush the process!
• Visibility and communication: Define the visibility level you are comfortable with – when in doubt, the more visible the better (of OKRs across teams and levels) – OKRs are a great way to communicate what’s important and who’s doing what
• Ensure clear ownership: Have a clear owner for each OKR, with all the relevant teams needed to support it reflecting their specific responsibilities in their respective OKRs.

Checklist – “Don’t…”

• …take up OKRs where even half the path to achieving the objective is not clear – these can be taken as “solutioning or experimentation OKRs”. For complex/ difficult problem statements it’s usually a good idea to take up a solutioning/ experimentation OKR first (in the current quarter) ahead of an execution OKR (in the next quarter).
• …abandon OKRs as a process if the first quarter is messy. Be patient, it takes a couple of quarters to mature – with time and experience, goal setting and KR definition get better.
• …write OKRs in “internal” language. The Os must be in the language of why this is “good for the customer” or “good for the company”.
• …wait for the right time or company size to introduce OKRs. There is no perfect moment at which OKRs suddenly become relevant – the earlier the better. With larger companies, be mindful of cultural barriers (e.g., trust, willingness to collaborate and share information).
• …experiment in only one silo. Misses the point of what OKRs can achieve cross-functionally. If you want to tread cautiously, take one OKR company-wide and learn from that experience.
• …allow sandbagging. A team continually achieving a “perfect 1.0 score” is probably taking goals that are too low or not breakthrough enough. The tone of the discussion for “yellow” OKRs will define the norm for how comfortable people are with taking a genuine stretch goal in their OKRs.
• …allocate 100% of capacity to pre-quarter OKRs. Leave some room to think and to adjust to market realities/ new priorities (these tend to happen!).
• …use OKRs as an excuse to not accommodate urgent work or adjust priorities. “I won’t do it, it’s not in my OKRs” has unfortunately become an excuse for some teams. That completely violates the spirit of OKRs – to have focus and method, but also to have the ability to quickly adjust based on market realities.

This article reflects the input from many conversations I’ve had on how OKRs have been used in different contexts (e.g., large vs. small companies, top-down vs. bottom-up, ops vs. tech). Every conversation adds something meaningful to my own experience and to what I’ve heard before. I suppose what we can learn from all of this is that OKRs really help and inevitably you will need to find your version of this that works best for you. All the best!