How to set growth goals: six lessons learned

Goal setting is hard. But not only is it a powerful way to focus a growth team’s efforts in the right direction, it can also be a forcing function to better understand your business and how it grows.

So if it’s a little painful – especially the first few times around – that might be a sign that you’re on the right track.

Through many mistakes and a few successes at Thumbtack and now working with a bunch of other teams, here are the biggest lessons learned.  

1. Start with your north star metric(s)

A primary growth goal – often called a north star metric – serves as the backbone of the goal setting process by providing a consistent yardstick by which to measure opportunities.

What makes a good north star metric? It should describe the core output the growth team aims to drive and span the full growth funnel from user acquisition to retention. One common form is weekly (or daily or monthly) active users. For marketplaces, this metric is commonly a measure of transactions between supply and demand, such as total rides taken (Lyft), nights booked (Airbnb), or projects completed (Thumbtack).

Effective north star metrics also measure the value delivered to your customers first, value to the business second. For example, in a subscription business, monthly active subscribers is a better metric than monthly paying subscribers, because without engaged subscribers, you won’t have revenue for long.

The prevailing wisdom is to use a single goal – “one metric to rule them all” – and that often works well. But I’m all for throwing out this constraint if it obscures important nuance. For example, at Thumbtack it was more effective to use both a demand-side metric (number of projects) and a supply-side metric (weekly active professionals). And for a specific period early on, we demoted both of these to focus primarily on marketplace health, measured as the % of projects that received a sufficient number of quotes from professionals.

2. Focus on the highest leverage inputs

To drive the more tactical planning process (e.g. quarterly goals at the working team level), you can break a north star metric down into its component inputs. Just as important as knowing how to improve those inputs is the process of determining which to focus on.

In the case of a travel marketplace like Airbnb, ‘nights booked’ could be broken down into traffic, conversion rate to first booking, bookings per customer, and nights per booking.

And you can further break out individual traffic sources, break down the steps of the conversion funnel, account for referrals, and so on until you get to the level of fidelity that is most effective for planning.

By then estimating the amount you can move each input, you can size the impact that it would have on your north star metric and find where you have the most leverage.

Of course, this example is simplified. But even when done at a pretty basic level I’ve seen this modeling exercise shift a team’s focus, quite often away from customer acquisition and toward metrics deeper in the funnel.

It’s also important to run some longer range scenarios of this model to avoid over-optimizing for short-term opportunities. For example, if paid marketing is 50% of your traffic, it might be a win to grow that by 20% this quarter. But at some point, you’re probably going to want to invest in diversifying, e.g. through virality or SEO, even if doubling those in the early days does less to move the bottom line.

3. Goals are a great sense check on your org (and vice versa)

The process of aligning on the inputs to tackle, and then figuring out who should work on what is a great opportunity to calibrate the growth org, especially as it relates to the number of goals relative to the number of working teams. (By “working team” I just mean any self-contained unit which has the technical, analytical and creative resources needed to attack a problem.)

More than one core goal per working team? That may be a sign you need to focus. In early stages, growth teams are most effective when laser focused on one or two goals. They’re typically areas with high reward to effort ratios where you can get quick wins, like conversion rate optimization or customer referrals.

Multiple working teams on the same goal? That might be an opportunity to collapse those teams into one. This frequently happens across product and marketing, e.g. a marketing team working on user retention via email and push notifications, and a product team working on retention via recommendation algorithms and in-product up-sells. Despite urging these two teams to just “operate as one” this setup can lead to poor resource allocation and a disjointed customer experience. This is a big part of the rationale for a “full stack” growth team: regardless of what you call it or what function they sit on, the most effective growth teams own all the available levers to attack the problem they’re focused on.

While written in a different context, Fred Brooks’ concept of “conceptual integrity” from the Mythical Man-Month – the idea that good design comes from one mind or a small number of agreeing minds – has always seemed very relevant to me as it relates to org design. One “mind”, one goal.

4. Metrics > milestones

It’s often less work to create a milestone-based goal (accomplish x by y date) instead of a metric-based goal. But they’re also usually less effective for growth teams.

In some cases, milestones are necessary. They’re useful for new initiatives with no baseline, especially when it will take a while to have an impact. When kicking off an SEO initiative in a competitive category, for example, success is going to be measured in years not months, and milestones will likely make better early goals. Milestones are also helpful for infrastructure projects, which a focus on short-term metrics makes it very hard to prioritize.

Otherwise, metrics win. By describing the measurable outcome you care about rather than the steps you want to take to get there, metrics are both more clearly linked to impact and allow for more flexibility in execution as the team learns what works. They also force a better understanding of the product and how it grows – so much so that a lack of metrics is often a symptom of not really understanding the problem.

Metrics are also just harder to sandbag, whether we intend to or not. At Thumbtack (a team full of thoughtful, well-intentioned people) I once looked back over multiple quarters of goals and found that we had hit 80% of milestones and only 55% of metrics.

5. When it comes to metric definition, the devil is in the details

A couple of common mistakes:

Poor measurability – It’s worth thinking about the tension between how well a metric measures the value you’re trying to drive, and how easy it is to measure. For example at Thumbtack, the number of times a customer hires a professional is a better measure of marketplace value than the number of projects they start. However, because transactions happen off of the platform, we couldn’t directly track which customers go on to hire a professional. We chose the upstream metric of projects started to allow for clean measurement and faster feedback. All of this said, if you find a better metric you can’t track today, it’s worth exploring if there’s a way to better measure it in the future.

Not accounting for changes in baseline – How often have you seen a goal that is simply impossible to achieve, or that can be achieved without any effort? The problem is often a poor projection of the baseline. One common form of this seasonality: beware the Q4 goal in an e-commerce business that hasn’t been seasonally adjusted. Another is mix shift: if the paid marketing team is planning to test a new, low intent channel, that is going to make it tough on the conversion rate optimization team unless it’s adjusted for.

6. You probably need a guardrail

This last one must be important because Andy Grove talked about it, and then Marc Andreessen tweeted about it, and then Tim Ferriss published that tweet in a book:

“For every metric, there should be another ‘paired’ metric that addresses the adverse consequences of the first metric”.

Guardrails are critical in growth because very often one goal is in tension with some other, important goal.

There is the narrower, short-term application of this principle. You might want to make sure, for example, that as you increase the number of signups from paid marketing you hold customer acquisition cost constant. Or that as you increase conversion rate, you’re not just converting low-quality customers and decreasing activation rate.

More important is the broader application of this principle. By optimizing on highly measurable metrics, growth teams are at risk of driving short-term gains while eroding long-term customer satisfaction or brand value. This is a tricky problem that I’m not going to do full justice here. However, here’s one example of how I messed this up at Thumbtack: initially goaling the email marketing team on increasing projects per customer (not a bad idea) without a customer experience guardrail (bad idea). An incremental email will always drive some incremental engagement, but too much will damage the long-term health of the customer relationship. We ultimately balanced this by setting a guardrail around “email quality score”, or the ratio of good to bad email interactions for a given campaign: (clicks) / (unsubscribes + 10x spams).

One of the best long-term guardrails I’ve found for growth teams is net promoter score (NPS), a clean and powerful measure of customer satisfaction. Monitoring it over the long term helps make sure you’re moving in the right direction, and measuring the NPS of experiment buckets can help guard against shipping short-term winners that may be long-term losers.


So that’s it. Work on the highest leverage inputs to your north star metric, map those inputs to teams 1:1, and use well-defined metrics (not milestones) with the appropriate guardrails.

When done right, goal setting can be a lot of work. But hopefully, it becomes not just something you have to do so that you can get back to execution, but a valuable process in its own right.