Everything you wanted to know about goals (plus a poorly drawn napkin sketch)
Vision -> Objectives -> KR’s (write, reduce, iterate)
“If you don't know where you want to go, then it doesn't matter which path you take.” - Alice in Wonderland
“You've got to be very careful if you don't know where you're going, because you might not get there.” - Yogi Berra
It seems I talk about goals a lot. Which leads to being asked “do you have something written down I can read/share?” Now that I type that I realize maybe that’s just people politely asking me to stop. But after a friend asked me recently I felt a little sheepish to admit I had little in writing. So I decided to change that. I tried to capture some things I think I’ve learned in a way that could quickly create a shared vocabulary for a group and at least lead to constructive debate.
There are many ways to formally capture goals. This document is grounded in “OKR’s” which is one of many equally valid ways to approach the problem. It has a specific vocabulary - but any approach that follows similar tenets should work similarly well. Those tenets being
Have a shared vision
Define a shared focusing vision set in a future state for the customer. Capture in writing what’s going to be true when that vision is achieved.
Set objectives for your work
Reduce the number of things on that list into a small number of high leverage, observable things that truly bottleneck/constrain achieving that.
Set the “how” in things you have full control over
Create a list of the most important AND controllable things you can do to assist motion on that path. These are the what and how of your work.
Be willing to change as you learn (Do the work while revisiting assumptions with a truly open mind
Execute and revisit periodically (roughly monthly) whether you need to alter any of the above goals and underlying assumptions as progress is made. Checking especially that you’re doing the most valuable thing for the global problem and not just serving progress on a local goal.
To avoid being theoretical for too long I’ll throw in an example before jumping into each of the tenets in more depth.
Example:
Maybe you sell styled boxes of clothes on a subscription basis (let’s call them “fixes”). Your vision (miracle hypothesis) is that 100% of customers love their fixes resulting in a growing population of thrilled customers. Key inputs might include your stylist community, your automated styling algorithms, cost of new visitors to the site, and inventory you have in stock when it comes to style a customer (breadth and depth). If you felt confident that if you improved the styling quality (via human and machine styling) hugely (whatever that means to you) but customers wouldn’t suddenly jump in terms of percent loving their fixes - then neither is likely your constraint. Instead you might believe inventory availability was a bottleneck. Meaning in practice that if you could somehow style people with unlimited inventory then happy customers would skyrocket, even with no other improvements. Then you should probably focus your objectives on this dimension. Setting overall goals then on “inventory availability,” inventory utilization, and velocity at which your execution teams could build in this space. Perhaps adding a fourth objective in savings that freed up free cash flow to extend the company’s runway or fund inventory buying. Each organization/person would then set controllable input targets (KR’s) as to how they would help make more inventory available in their shorter term focus.
Let’s go from this specific example back up a few levels and talk more about the steps in a powerful goals process:
Shared vision
Creating shared vision requires defining the problem you want to solve (aka your goal) very specifically and clearly, and writing it down. This isn’t about having a number for everything - but being able to say what will be observably true in a successful future. That new state may be super specific or a summary of some “new state of being.” It depends a lot on what you’re trying to achieve. Doing this makes it easier to row in the same direction and empower anyone to point out when an effort is wasteful because it isn’t centered in this vision.
An approach popularized by Amazon approaches this as writing a press release to work backward from. That’s an entirely useful/valid approach - but you can also use a less constrained approach of writing down a future vision statement. I find that an easy step in that direction is just to sit down and say to yourself “if I came to work in 12 months and all of our problems had disappeared, how would I recognize it?” - and then write out a short narrative expression of what that day in the new reality would look like. It gets to 80% of the heart of the press release idea likely in 20% of the time.
What’s important is that multiple reasonable people looking at the new state of affairs you’ve written about agree as to how well you did, and agree that end state is awesome (or at least a lot better than the current reality). Focusing too much on the “measurability” here may bite you. The world is littered with many stories of how a group chose to set goaling around something because of its easy measurability. While missing the fact it wasn’t really the right thing to focus on. So it’s worth stating again that your core vision/goal, the thing that needs to be true in the future, doesn’t have to be readily countable.
As an important sanity check, double check the vision is aligned with a real problem that people have. These people are your customers. If you cannot explain the problem you’re solving to customers so clearly that they’ll agree “that’s right” (ideally shaking their head either that that was super obvious OR that they’re so happy someone finally understood them) then you’re not done. And possibly/likely dangerously off course. These customers can be external customers, internal ones (for example: you’re working to solve problems in finance so ask the people doing the work if they’d be thrilled if you achieved this goal), or your teammates/manager.
Objectives
Once you’re pretty sure about the vision then write out what success looks like for the problem (aka - your list of Objectives). Almost any enterprise should have 5 or less of these. Much less is much better.
Do a bit of confirmation if people see the same things in what you wrote by asking for feedback. If different people interpret the goals differently then keep looping. This is time well spent as it ensures you’ll pass the test of “if I ask 10 people across this group of 100 randomly what their goal is - will they all say basically the same thing?” This step will sound a lot like the first step in the vision setting part, and it basically is. But now you’ve got more insight, more help, and more practice. The first step will be at a higher level than what you need to execute on for say the next year. So we’ll call the beginning your vision, and what you’ve got at this stage your Objectives.
Objectives (the “what”) while more expansive and longer running than execution goals (the “how”) should still be observable and not overly broad. This can often be accomplished simply by stating the directional objective first and then followed with the phrase “as measured by ______.” “As measured by” does not have to be a number. But it should be a state that reasonable and informed people could agree on when it comes time to assess our accomplishments later on.
The objective’s core purpose is to share the intent of the team, organization or individual. It should be about the “what needs to be true”, “why it’s important” and any guardrails that might constrain the “how” of achieving. Your objective (O) represents a future state that’s important to you to achieve, KR’s are the controllable things you’ll try to do to move in the right direction.
Key Results (how are we planning on making progress on the objective)
The next step is to take your longer running goals (Objectives) and for each period of time (say quarterly) define “how” you plan to contribute to the goal. These form the key results (KR’s) that you’ve autonomously determined are the best way to contribute to the larger goals. Each group or person should focus on the unique ways that they can contribute to these objectives, writing out how they will (in the shorter term) help move things in that direction. At this stage it’s likely a specific example might help.
Let’s imagine you’re part of a team who builds pricing models for bidding on the right to do units of work based in part on how much it will cost the company to provide that service (maybe where the service providers are sourced from a dynamic marketplace. To get even more specific let’s imagine that your work on setting objectives narrowed in on three core ideas;
Improve price prediction accuracy such that the error in our top five business segments is < ___ during normal operation
Avoid short term catastrophic predictive errors.
Increase the velocity that the science team can iterate on testing new approaches.
This might result in one of the objectives taking the form of #2 being, “Reduce negative events with our pricing model as measured by the number of times the model deviates > 50% from reality for more than 5 minutes.” In which case possible Q1 KR’s might include;
KR1: Implement data consistency checks on daily retraining that stop training if input data changes > (insert reasonable but gross measure; average, distribution, etc) day/day by 1/15/xx
KR2: Implement alarming such that the team is paged if average price predicted within top 5 business segments (3 geographies, drop shipping, live load) changes by 30% vs. prior week in a 30 minute trailing window.
KR3: Launch ____ tools to reduce time to debug pricing question from operations from 6 hours to 15 minutes by March 1st.
Note, that the KR’s (mostly) takes the preferable form of “improve ____ from X to Y an improvement of Z by _____.” This is a quick check to confirm you’ve got something observable. KR1 and KR2 are observable in the sense that people reading them in the beginning likely will agree what the end state should look like, and you can explain to someone after the quarter has ended why you believe you’ve achieved the goal. Surfacing these success criteria early allows for active debate if they are the “best” versions of the team’s intent, if this intent is the best use of focused time to improve their objective, and of course if the objective’s juice is worth the squeeze.
Don’t skip the iteration part
Both the objective and the KR’s should go through rigorous, transparent debate both upwards (to higher management layers) and downward/laterally (within team and adjacent partner teams). That will confirm the contracts of focus make sense given disparate perspectives on your reality. This iterative sharpening step is often given short shrift - but it’s truly the most important part of the process as it’s where assumptions are broken and thinking is expanded. When it’s slipped over it’s usually out of a concern to “make thing go faster.” But much like the idea of limiting meetings to 30 minutes you’re going to pay and pay and pay later if you skip the tough parts here. You want to go a little slower now so you can go hella faster later.
For larger, complex orgs there’s a simple corollary tool that is underutilized but can have dramatic impact. I suggest that teams say out loud what goals they’d like to set for other teams to enable getting their work done better (obviously also writing them down). Then look closely at these because that’s where you may have golden ideas - or at least mismatched assumptions. Make sure you discuss these goals with those other teams as part of your goal cross briefing process.
Making it personal
OKR’s are key for any larger, complex endeavor. But they’re also super useful tools for aligning personal work with longer term organizational goals. You just need to reframe them as “given what I know is important overall, how can my unique set of skills most impactfully be applied to busines needs. If you worked in a large hotel as a bellhop and the key objectives were around a perfect (defect free) experience for the guests then you might focus on the percentage of times you could bring a guests bags to their room before they arrived as an objective. Setting KR’s focus you then to continuously get better at that in a way that you control.
Stepping out of the work world with a much used example - if you’re trying to be healthier (maybe in support of a vision to live as long as possible and have energy to spend meaningful time with your kids/grandkids) then you might arrive at the following OKR’s
Objective: Improve my physical fitness as measured by weight < 165 lbs, resting pulse < 65, and increased strength (bench press 180 lbs 3x within 5 minutes).
KR1: Increase weekly gym visits from 3x per month to 15x per month by end of Q1.
KR2: Increase cardio activities from 60 minutes per week to 120 minutes per week.
KR3: Take a comprehensive nutrition class by ____
KR4: Sign up for a daily meal plan to ensure I have healthy dinners that can be prepared in < 15 minutes 5 days/week.
This is an example, and an imperfect one - but it should include the gist of the strategy. Write out what you want to achieve, by when, and how you’ll know you’ve achieved it. Then write out your shorter term strategies, each of which you can control for some next period of time. The Objective likely will not change, but the “how” (KR’s) will continue to be something you iterate on. Either because you complete each or come up with better ways to proceed.
You almost certainly have good, specific ideas of what’s important for whatever you’re trying to achieve. Framing what you think needs to be done in terms of the team, org’s or company’s objectives also makes it more likely folks will get excited about what you propose. Saying you need more complete test coverage to enable true continuous deployment will likely make more sense among competing priorities if stated as “increase the number of experimental turns focused on improving pricing predictions by 5x” (as the objective) by KR: “increasing the number of deployments from 1/week to 5/day by <insert cool thing about test coverage>.”
Recap and practical practices via bullet points
The KR’s define ‘how’ the org will make progress on the intent (objective). They should be as specific and as observable as possible.
KR’s should represent “input” drivers and not outputs. “Make more money” is the output of something. Ensuring that customers have a good NPS score is closer to an input - as it’s one of the factors that likely results in you making more money. Reducing defects that we know contribute to a poor experience (and hence NPS score) is an even better controllable input as it’s more specific, and more upstream from the state you want to impact. If it’s a constraint to your system being able to produce more of the output then it’s a high leverage thing to focus a lot of attention on.
The reason for KR’s to be specific is because the KR’s represent the org’s best current approach of “how” to satisfy the intent of the group (the objectives). Therefore being succinct and clear allows other people in the organization to check if these plans match their assumptions. Catching assumption mismatches early increases the odds of fast and effective execution.
It is not expected that all KR’s will be achieved. They should be important, achievable, but create some tension (ie; we don’t get benefit from sandbagging them). You don’t want your team to work on all of their KR’s in parallel, but you do want to have more than just the ideas they’re sure they can accomplish captured at the start. But don’t go overboard or you’ll have a bear of a time keeping work in progress (WIP) low and multitasking minimized [more on that evil another day].
Some ad hoc rules
If you are improving something countable, write the KR in the form; “Improve ____ from X to Y an improvement of Z.” Including the current performance (where we start from) provides additional context beyond the size of the improvement (Z%).
For countable goals include when you expect the goal will be achieved and over what measurement period. As an example “improve split shipments from X% of orders in January to Y% in December” is likely a fully specified KR.
When speaking of percentages include the absolute values associated with the change. This reduces predictable questions such as “hey - if you are improving oncall by 15% does oncall consist of 1/2 an engineer or 10 engineers?” While also giving additional context as to the size of the problem/area to a reader.
If your KR’s are to “release” or “demonstrate” something then strongly consider if you can be more specific about what functionally will be true when the KR is completed. “Launch ___ on ___” is probably less helpful than “be able to have one person in merchandising operations do _____ in production with full error checking by ____.”
No matter how much it would be nice if it were not so - a percentage of goals are likely to be completion based. In those cases if you plan to be done before the end of the quarter then write down a specific target date. Otherwise it’s safe to assume your plan is to finish the last day of the quarter. It’s helpful to see if a plan assumes a cascade of last minute things coming true. This is especially useful in predicting excessive WIP that can impede flow. But really, really try to find a functional definition of “completion” vs. some unobservable state like “project super-terrific-re-write of X is done.”
As time has gone on I’ve become an advocate for including next to each KR the approximate estimation effort the team has assigned to the work. This has several advantages including (a) makes transparent that the team has estimated the work and the size of the estimate (b) forces a discussion about whether there is likely an unrealistic amount of work planned (c) along with target dates tends to flag highly parallel efforts (ie; 5 projects with 4 engineers) which increase bad multitasking (d) proactively answer one of the most common “resource allocation” questions, thus building trust upwards.
I probably should have put this first - AVOID USING ACRONYMS in your goals. If you must please define it. We should assume that over time smart, informed people will get value and improve our OKR’s. Using acronyms that are more internal or even company wide rule out that feedback from those outside the tribe. OKR’s serve a critical assumption and intent sharing purpose. Gumming it up with local team lingo tends to reduce their effectiveness because you’re baking in assumptions along with the acronyms.
Thanks for reading this far. There are lots of other aspects of this topic, and I’m aware I left some out - such as how often, and how to review goals. If there’s interest in things I’ve glossed over let me know. If you’d like an external perspective on your goals or related process then please reach out and let’s chat.
End Notes:
Personal vs company goals: If you’re at work and someone is asking you to set goals for your own (personal) focus and you don’t have a clear understanding of the company’s objectives then you should probably stop and ask about those. It’s certainly possible to write well structured goals for yourself. But if you’re not sure what the company’s most important outcomes are then it’s likely pretty hard to be sure you’re planning on doing the most important thing you can control to assist. This maybe sounds obvious - but I’ve seen places get the order of these things wrong.
Goals vs objectives: Some early readers have asked about how “goals” relate to objectives, KR’s and vision. So I reread the text with an eye toward how I used the term “goal.” I realized that given the way I’m throwing it around that vision statements, objectives and KR’s are all “goals.” In that they’re statements of intent to achieve something. Sort of like how chickens are all chickens - but hens and roosters are also chickens. And which is which matters depending on what you’re trying to solve for. I realize that explanation would make more sense if I’d bet someone I could discuss a work related issue using farm terminology. But (perhaps) sadly I have not.
Reading list:
Some further reading that does a better job than me overall include - admittedly with a higher page count:
Measure what matters
4DX - The 4 Disciplines of Execution
The Art of Action