During my time at Meta, I conducted lots of Product Manager (PM) interviews. We had a particular type of interview which was called the "execution interview" which despite its violent sounding name, it of course referred to the other definition of execution. It was ostensibly about gauging a candidate's ability to build products effectively. However, in practice, it primarily focused on goal-setting.

This focus stemmed from the fact that PMs at Meta primarily operated via goals. The most successful PMs understood how to leverage goals to motivate teams into creating outstanding products. They were well-versed in the art of setting meaningful goals, knew how to avoid common pitfalls, and understood how ill-conceived objectives could lead to adverse outcomes. Without these abilities, a PM would find it challenging to accomplish anything at all.

The interview, in reality, only served as the initial rudimentary test and I spent the next five years being immersed in goal setting at Meta. It was a steep learning curve, involving teams scrutinizing what I initially believed were good objectives, only to realize they were fundamentally flawed or I had missed important nuance.

Throughout the years, my understanding of effective goal setting became far more nuanced. I learned why certain goals succeeded and, crucially, how to discern if I had indeed established a solid goal.

This note serves as a summary of my key takeaways. It presents two sets of six criteria each that you can utilize to evaluate your goals. The first set focuses on the quality of the goals themselves, while the second examines how these goals fit into a system that enables progress. While primarily aimed at product teams, the principles can certainly be adapted to personal goals as well.


To start, let's ensure we're all on the same page with the terminology. Here are the definitions I'll be using throughout this note:

The Purpose of Goals

No goal is perfect. It's crucial to remember that goals are tools, not a universal fix for your team's challenges. However, if your goals assist you in the following ways, they're doing their job. You can set goals for four distinct, yet interconnected, reasons:

  1. To concentrate your efforts on the most crucial tasks that bring you closer to the mission and vision.
  2. To inspire ambition, driving you and your team towards accomplishing more than what you believe is possible.
  3. To communicate your priorities and motivations to others clearly.
  4. To foster accountability and aid in self-reflection for iterative improvement.

Part 1 - The Six Tests of Good Goals

The following criteria aren't foolproof, there are always exceptions. However, if the goal you're proposing doesn't pass one of these tests, it's worth considering why that is and exploring whether there's a better approach.

Nevertheless, there are times when setting an imperfect goal is the right thing to do. For instance, when you don't have the right measurement in place and still need to make progress. The tests below help you to see the flaws in your goal and areas where they might be improved, and you can even use them to set a goal about goaling so that you are in a better position for next time.

1/ Does it clearly define an outcome?

Defining a outcome means that it describes the state that you want your product to be in. It could be growth “grow the product to 100K monthly active users” or technical improvements like “reduce the Android binary size to 5MB”. This is different to a ship goal which usually describes the activity that might lead to an outcome such as “ship my new feature” or “refactor a module to reduce binary size”.

Defining outcomes fosters empowerment. By focusing on outcomes, you're given the freedom to determine and tweak how you achieve something, and pivot your strategy when it makes sense. On the other hand, ship goals can lock you into a specific set of actions, regardless of their effectiveness or impact.

2/ Is it possible to exceed the target?

Once you have a goal, then you need to decide on a target, this is usually where you take a specific metric like 'daily active users' and decide on a number that you want to hit. You should evaluate if you can hit the target, how ambitious the target is and if it's possible to do better than the target and therefore exceed it.

Not all circumstances can be converted to numbers. Even in those scenarios, it should be possible to understand what at least what a good, bad and exceptional result would be for the goal and to set your target accordingly. The goal should be able to express how badly you failed or how well you did. Ship goals, like “launch UI revamp” only leaves you room for failure or completion, a goal with a description of bad, good and exceptional targets, allows you to exceed - a much more exciting outcome to work towards.

Side Note: How to Set Non-Numeric Targets

For instance, consider a manager who has a strained relationship with some of their reports. A good outcome could be an improvement in these relationships, as reflected in peer feedback. An exceptional outcome might involve peer feedback showing that the improved relationships have led to significant enhancements in the product. On the other hand, a poor outcome would result in feedback that remains unchanged from before.

3/ Is it easy to explain why achieving the goal is a good thing?

Goals help focus work on what matters; building a product that people find useful. This means that it should map to the vision, the mission, and the top line goals of the product. Sometimes this process is imperfect, and we need intermediate goals or proxies to better reflect the work that is going on during the time period that you are setting the goal for, but it should always be possible to write a concise rationale for the goal.

This is also true of the target itself. It is important to have a rationale about why you have chosen a specific number. Is it ambitious enough? Does it force a meaningful change to the product?

4/ Is it easy to understand?

"Metric A divided by Metric B, C & D which are weighted by their importance according to our research on our most recent customers."

There are times when weighted or complex metrics can be useful, but they can have a substantial cost. They take time to rationalise and can cause confusion, especially where the precise definition of the metric is not well understood.

Even when teams think they understand these metrics they often don't, humans don't tend to be very good at this kind of multidimensional problem, keeping track of many variables at the same time with different weights is hard especially when they can all interact with each other.


If it takes longer than 10 seconds to explain a metric to someone on the team who has context, its probably too complicated. It's often easier to keep track of several easy to understand metrics than one compound metric.

5/ Is it hard to game and is the outcome net positive?

You are being paid to achieve the mission of your product not move some needle on a dashboard. Part of the responsibility of working towards a goal is to make sure you are doing it in a way that is not destructive overall.

Ask yourself is there a way that we could achieve this goal but in the process cause harm elsewhere? Goals that can be gamed, either intentionally (e.g. forcing out low engagement users to make a DAU/MAU ratio improve) or unintentionally (e.g. short term increases to MAU by opening up a marketing channel without realising that product is not retentive) can mean at best wasting time, at worst, causing damage which the team needs to fix.

You can set up guardrails to protect from these things, but remember when you do that, you are essentially making a more complex metric (see rule 4). Try to link your metrics to the real impact you want to see and keep them at that level. You can talk about proxy metrics within the team but try not to goal on them directly. Even if its harder to move, a metric that is clearly linked to the outcome you want is better. It means you stay focused on the real challenge.

Side note: Some common traps

  1. Metric A per Metric B (any number divided by another) or as I tend to call it, 'the ratio trap". You have just set two goals, not one. In the worst case both of these numbers can change, and then you open yourselves to a bunch of implicit trade offs that you will have to make on the fly. Is it better to have a 10% user to subscriber conversion rate when you have 100 users and 10 subscribers or 1% with 1000 users and 10 subscribers, does the ratio matter? - This is a trade off worth thinking through. By putting it in your metric you have delayed that trade off.
  2. Metric A per 'a time period' (e.g. users per day), you have to justify why you chose a day, why not 26 hours, or a week? This is one of the places where 'arbitrary metric edges' can catch you out (e.g. a metric with a hard cut off and data points that fall just on the wrong side of the cut off don't count despite being extremely close to the ones that do on the other side of the mark.). You need to understand the distribution of your data and you need to know why a day or a week matters, and where it doesn't.
  3. A weighted score. You take a bunch of metrics and assign weights to them which you combine to create a score which you goal on. It is unlikely that your weighting is correct, usually the weights are based on some sort of intuition or very complex reasoning prone to errors that are hard to spot. You also end up optimising for something that may only have a tangential relationship to the real user experience. They tend to give a false sense of 'covering all bases' whilst really covering none.

6/ Can it can be influenced by the team?

Responsibility requires agency. If you can't do something that will move the product towards your goal then you can't take responsibility for it.

Sometimes it's easy to express goals high-level goals (e.g. increase MAP) that clearly meet the first 5 tests, but in practice your team does not have the agency to move them.

This can be because the team is downstream of other significant efforts or gatekeepers, for instance a goal on the number of support cases resolved may be more heavily influenced by changes to the availability of support than the actions agents on the tickets. It can also be because the effect takes time to see in the metric, a lagging indicator. In these cases, using a leading indicator can help - Boz's post 'The revenue takes care of itself is worth a read on this topic.

Agency can come in many forms and nearly all goals will require some degree of influence on other teams, so you should not be afraid to set goals that are dependent on other teams or outside influences, just be sure you know how you can be influential.

An example

Product X has 500 pre-paid annual subscribers
  1. Does it clearly define an outcome? Yes, this is not a ship goal, it describes a product where people are paying upfront for annual subscriptions. This is a sign that they are willing to commit to the product for a long period of time which could be in contrast to a monthly subscription of a pay as you go option. The goal is specific about an outcome that the team wants.
  2. Is it possible to exceed the target? The goal is clearly measurable and it's clear how it could be exceeded - you get more than 500 pre-paid annual subscribers. This leaves the potential for a substantial over performance and pushes people think about how you could go far beyond.
  3. Is it easy to explain why achieving the goal is a good thing? Paying for an annual subscription is a vote of confidence in the product, its a good sign that people see value in it. However there are still ways for it to be achieved that don't always lead to a great outcome. For example, if people are convinced to sign up to an annual subscription, but they don't use the product during the year. That might suggest having a goal around engagement of paid users to make sure the perceived value turns into realised value.
  4. Is it easy to understand? This is a really easy goal to understand, its expressed as an absolute number, rather than something more complex like "convert 10% of registered users to pre-paid annual subscribers". Although there may be hidden dependencies, like the advertising funnel, we can account for these later (see the second section on goaling systems).
  5. Is it hard to game and is the outcome net positive? This is hard to game because it requires a real commitment from customers, it's hard to convince people to pay for something and it's a direct reflection of success for the business. In contrast, if it had been a % conversion target there might be a temptation to boost the rate by turning down advertising or marketing efforts. That might have been something which would have needed to be guarded against.
  6. Can it can be influenced by the team? Clearly this is somewhat dependent on the circumstances of the team, but assuming this is the team in charge of the product then it would appear to be a clear yes. This team can change the product in any way they see fit which would encourage people to subscribe.

This goal has passed all but one of the tests. It has given us a little cause for concern about a scenario where the subscribers go up but the engagement does not track. There are plenty of ways to deal with that problem but this is exactly what these tests are designed to help uncover.

Part 2 - The Six Tests of a Good Goaling System

Having some goals is a great start, but what's next? While setting good goals is priority number one, the way you utilize them is just as crucial. A goal that's constantly being changed every week or simply ignored isn't going to be much help.

Most goal-setting systems can give you a rough idea of how this could work, but I want to emphasize what I believe are the most critical features that good goal-setting processes share. As for which one you choose—be it North Star metrics, 50/50 goals, OKRs, or something else—I don't have any strong preferences. It's seldom the specific framework that makes the difference. You can use these tests to evaluate how to enhance your existing process.

It's also worth noting that I have also seen goaling systems applied (most notably OKRs) with no rigour whatsoever, teams with hundreds of goals (basically a roadmap), or with the rigour in the wrong place (e.g an excessively bureaucratic goal change process). Just because you are giving it the name and following the format does not mean you are following the methodology.

1/ Do your goals have a deadline, expected speed of progress and clear assumptions?

A goal without a deadline makes it hard to make decisions. Getting 5,000 new users in a week is a very different task to getting 5,000 new users in a year. The time you allow to reach a goal will, in part, define the tactics you use. That means it needs to be an intentional decision.

In some situations, you may have the timeframe dictated to you for the planning process. That's ok, you can adjust your goals accordingly 5,000 users in a year might become 2,500 for a 6 month planning cycle or 420 for a month (quick maths). The magnitude will still change the decision making space.


Use time periods to think through how your approach might change. This can help you think of new ways to approach the problem and recognise a wider set of potential influences. What would you need to do if you had to achieve your goal next week? A massive advertising campaign? Take a punt on the new architecture you have been thinking about? Even if you don't take the more extreme option its helpful to know what it would be. The same is true of extending the deadline. Would organic growth be enough, would you then focus more on new features or extending to a new market?

Once you have a deadline you also need to define the speed of progress. Some work will result in progressive gains. For instance, if you are growing your user base or making a bunch of small optimisations to your app to make it faster. In these cases you can see the improvements with every item you ship. In the simplest of cases, this will be linear change over the period and therefore nice to represent on a chart with a nice straight line.

Other times, you might not hit the goal until the end of the period or you may not be able to break it down into smaller incremental chunks. This is not the ideal scenario as it makes monitoring hard but it happens. In these cases I would suggest trying to find something that measures incremental progress, even if it's not a perfect representation of your outcome (e.g. we have 6 things to do and 3 are done).

But between these two extremes, is where most situations actually land, some items are bigger than others, there may be some seasonality in the period which could cause a spike of usage, etc. There are many ways to represent and manage these within goals, which could probably warrant its own note, but the point to remember is that thinking through how you expect progress to evolve over time will force you to think through these questions and help you to communicate that effectively.

When you do work that out, you should express it, ideally in the form of a chart and I would recommend reading a great post about creating the perfect goaling chart by a former manager of mine, and great PM Simon Cross: The Perfect Chart - by Simon Cross - Tradeoffs and Payoffs.

Tip: using goals to think through risk

One way to prioritise your roadmap is to use the insights you have gained from thinking through the speed of progress to decisions about how you want to approach the risk of not achieving your goals.

If you have one big project which will give you most of your gains, should you do that first? What if it doesn't pay off? Will you have run out of time to do the other things on your list? If you do all the small things first will they add up to big enough gains.

There isn't a right answer to this question, but it is something you should think about. Using this lens can force you to re-evaluate your projects, be specific about the riskiest parts and think of reasonable mitigations.

My general advice is two fold. Firstly, lean towards redundancy, in other words make sure you find multiple ways to achieve your goal and try more than one option. Secondly, and especially if the situation does not allow you to do that, communicate the risk, make sure everyone understands the shape of the risk you are taking on.

Once you have a deadline and the expected speed of progress now you can make sure you have thought through the goal's assumptions. These are the things that you, your team, your organisation's leaders, investors and whoever else is involved are all assuming to be true but influence your goals. For instance, if you were working with the public sector, you might assume that their funding situation remains consistent or if you have substantial hosting costs that your provider isn't going to double their fees overnight.

Although many of these can seem obvious, and you don't have to be totally exhaustive, there are two things you are trying to get from making this list. Firstly, a clear accounting of the factors that you are dependent on and you should be on the lookout for change. That means if they do change, you can be quick to use it to your advantage and if you need to no one will be surprised when you suggest changing goals. Secondly, that everyone is on the same page, what might seem to be an obvious assumption to you might not be to your investor, director or team member.

It also cuts short a huge amount of post event debate. You can say, either it was foreseen and we decided not to act, or it wasn't but we all missed it. As I have often said to PM's in my teams, don't take risk alone, share the burden and work out how to tackle it together, never get yourself in a situation of hiding a potential problem and have it blow up in your face.

Side note:

In my experience, poor leaders have a tendency to play the blame game after a negative event. They should have been setting their teams up to make sure they do have redundancy in the case of inevitable negative events and supporting them when they happen. The blame game is usually a form of regret, a reflection of their lack of confidence and the fact that they did not help the team to avoid the situation in the first place.

If you are a leader, ask your team to think through assumptions and actively contribute to finding potential problems. Having a discussion about risk in advance is a healthy practice for teams, it helps build better products and reduces stress. A problem shared is a problem halved.

So your final goal statement should have the following:

  1. A clear statement of the outcome you want to achieve
  2. The deadline by which you want to achieve that outcome
  3. The speed of progress over the time between now and the deadline
  4. What shared assumptions you all have for the goal.

For those of you wondering how hard this is and how much time it would take. The answer is that it's going to take you a significant chunk of time the first time you do it (a day or two probably for a set of goals). That's because you are learning how to do it and everyone around you is learning how to do it but once you have done it becomes much quicker for the next round. In my experience it has always been worth the investment and it's always much less time that making a bad decision on your goal and dealing with the consequences.

2/ Do you have clarity on how and when to change goals?

There are times when you think you need to change goals. This can be tricky. How do you know if you are giving up too early or you're are heading in the wrong direction? Honestly, changing goals, especially in larger organisation always comes with the suspicion that the team can't make the goal because they don't know how. This is why having a clear way to change goals can help make this process fair, clear and productive.

There are scenarios where changing a goal is not just helpful but categorically the right thing to do. I have personally been in situations where we chose to improve something but did not realise the negative impact it had elsewhere in the system, continuing to optimise that metric would have delivered harm overall. You have to change that goal.

There are also times where the situation has changed and the chances of meeting the goal are so low its deeply demoralising to the team. These situations are much harder, you don't want to take away the incentive to defy the odds and pull of something amazing, on the other hand a demoralised team can turn into a burned out team pretty quickly.

This is where assumptions are particularly helpful, if any of your assumptions change then its grounds for a review of the goal, and given that everyone has bought into the assumptions its usually an easy case to make.

I really don't think you need anything complicated here, simply the following:

  1. A shared recognition that it is possible to change goals - especially if assumptions have changed.
  2. A requirement to write (yes I think writing it is important to clarify your thoughts) a case for the change.
  3. A time to discuss with a requirement that the meeting ends with a decision to change or not - decision making delay can be worse than working on the wrong goal.

3/ Do you know how and when you will monitor goal progress?

So you have clear goal statements, clarity on changing goals but now you need to agree how you will monitor them together. To get the most out of goals you need to be regularly checking how things are going, you need to be held accountable to them and you often need to change your approach to make the most progress.

It's up to you to decide on a cadence that works but set it up at the start of the period and stick to it. There is nothing worse than the "goals review meeting" that appears out of the blue and its far too easy to put off thinking about goals for a few weeks, hoping that things will get better if you don't have a time set up. Get used to the awkwardness of not making as much progress as you would have liked and talking honestly about it. When you do this you access everyone's ability to think about the problem and help. The PM who keeps it to themselves whilst hoping everything is fine or will work out in the end, is almost never fine in the end and they caused themselves a whole bunch of stress in the process.


When you do this you should also align on the terms you are going to use and how you are going to talk about risk. Make sure you all agree on what the status of a goal means (e.g. “on track”). This helps you communicate effectively but also communicate nuances like the difference between being “behind” or a goal being “at risk”.

Ask people you work with to give a percentage value for common terms like "not very likely to happen" or "probably will happen" and see quite how different they can be. Some people place "probably will happen" closer to 80%, others have it at 51% but that is a huge difference.

This is why having a shared terminology and referring to it when using terms like "at risk" is so important. For instance you could say "At risk" means we think there is a less than 80% chance of getting it done or we shift projects to "At risk" if we can see any reason why they might not complete. Doing reviews where everything is green and 'on track' is pointless, it doesn't mean you are doing well, it means you haven't thought hard enough about it.

4/ Do you have too many goals?

There does not have to be a hard limit on the number of goals, but less is usually better. Goals should help us focus, lots of goals means you don't have focus. It's important to remember that not every single piece of work needs to be covered by a goal there are many things which are part of what teams but aren't linked directly to goals, that does not mean they are not important (e.g. team meetings, oncall rotations, small UI fixes).

I often intentionally made sure that I felt like the team would have a decent chance of achieving its goals with 80% of the time available. Not only did that give us capacity when things went wrong, but it also allowed us some chance to experiment, fix small things and give people time to think of the next big thing.

5/ Do you have clear expectations for goals including when you will do a final review of the goals and reflect on the results?

Good goaling systems demand that there are consequences. It doesn't mean people should be fired for not meeting goals or excessively praised for exceeding them but it is important that the results are celebrated, commiserated and used as an input for the next round.

A good goaling process is clear about the expectations for achieving or exceeding goals. So that goals can help you be ambitious there needs to be a healthy doubt that you will be able to achieve them.

Making it so you have to achieve all the goals or you fail, is a recipe for encouraging people to write easy goals so they can be sure of their success.

50/50 goals are a elegant way of avoiding that trap. 50/50 goals means that you think you have a 50% chance of achieving the target when set. This means that you are not expected to achieve all your goals all the time. OKRs use a more complex 0-1 scoring method, with 0.7 as a target, but it amounts to the same thing, it's a way of setting expectations for goals and the final review is where you make the judgments over those goals.

Side note:

Some people get into prioritising goals, which is a bit of a pet peeve of mine, because I don't think it's very effective. Instead set the first goal and if you get it done, add another goal. Prioritising goals means you and/or your organisation are having trouble making clear decisions, and want to be able to say 'we are doing all of these things'.

So always make sure that the following are true:

  1. You have clear expectations about hitting goals
  2. You know when you will finally review the progress
  3. You use the results to celebrate, commiserate and feed into the process of setting the next round of goals.

6/ Are your product goals separated from you and your team's personal goals?

Teams should be rewarded for their overall product strategy and execution, especially how well they adapt and learn, not just whether they hit a specific target.

As stated above, evaluating the performance of team against their goals requires nuance and understanding of the specific context. The team should be able to make progress or explain why they were unable to get to where they wanted. Teams should constantly be considering their progress towards goals and the impact of external forces and be transparently communicating this beyond their team. There is much more to overarching performance than just goals.

It is also worth noting that team performance does not always equal individual performance. It's perfectly possible to make amazing individual performances but the team misses it's goal. As well as the opposite, the team exceeds it's goal despite individual performances.

Cheat Sheet

Six tests for goals

  1. Does it clearly define an outcome?
  2. Is it possible to exceed the target?
  3. Is it easy to explain why achieving the goal is a good thing?
  4. Is it easy to understand?
  5. Is it hard to game and is the outcome net positive?
  6. Can it can be influenced by the team?

Six tests for goaling systems

  1. Do you goals have a deadline, expected speed of progress and clear assumptions?
  2. Do you have clarity on how and when to change goals?
  3. Do you know how and when you will monitor goal progress?
  4. Do you have too many goals?
  5. Do you have clear expectations for goals including when you will do a final review of the goals and reflect on the results?
  6. Are your product goals separated from you and your team's personal goals?