How to Set Goals That Are Actually Ambitious (Not Just Optimistic)

There's a question that shows up constantly in PM interviews and in H-planning conversations: "How do you set goals for your team?"

Most people answer it by describing SMART goals or OKRs. That's the wrong level. Those are frameworks for documenting goals. They're not a diagnostic for figuring out what the target should actually be in the first place.

This post walks you through a diagnostic framework - the inputs and thought process for setting targets once your north star is already defined.

First, separate the goal from the roadmap

A common mistake: treating goal-setting and roadmap-building as the same exercise. They're not.

The goal is the outcome you're trying to achieve. The roadmap is how you get there. You need both, but you can't set an honest goal by listing your roadmap items and reverse-engineering a number.

In interviews, you can surface this with a simple clarifying question: "Are you asking about the target I'd set, or the roadmap I'd build to get there?" It's a good question in real planning conversations too.

Anchor on reality, not aspiration

The most common goal-setting failure is setting a target that's disconnected from how the surface actually moves.

If you're working on a highly optimized, at-scale product - say, Instagram Newsfeed - you're not in 10-20% growth territory. You're in the 0.5-3% range. In some cases, 0.1-0.2% can be defined as genuinely ambitious.

Start by looking at:

- The current level of your north star metric

- The recent trend

- The maturity of the surface

A mature product's levers are different from a growth product's levers. That context shapes the entire target-setting exercise.

Work backwards from the northstar to what you can actually move

Let's say your north star is "Daily active people with a meaningful interaction on Feed" (likes, comments, reshares, bookmarks) and that it isn't feasible to influence that metric directly.

Think about what moves it: share rate in top positions of feed, comment depth on friend posts, Reels replay rate. These are the laddering metrics - the things your team can actually run experiments against.

Your 2-3% north star improvement (as it's an optimized mature surface) should be the aggregate effect of moving those levers, not a number you picked because it sounds ambitious.

You can also break the metric down mathematically. If your north star is DAPMI (daily active people with meaningful interaction), then:

DAPMI = DAP x %DAPMI

This decomposition helps you build concrete hypotheses: what moves daily active people overall, and what moves the share of them who meaningfully interact?

Sanity check against what your team can actually ship

Before you land on a target, pressure-test it against three things:

Throughput How many experiments and launches can the team realistically ship in a half? What's the historical launch rate? A 2% target built on 15 winning experiments when your team ships 20 experiments and wins 40% of them doesn't math out.

Team health Have there been big changes to productivity - AI tooling improvements, recent tech debt paydown, or the opposite? Do you have new hires who need ramp time? Churn that's pulled out institutional knowledge? These all affect what you can reasonably expect per engineer per half.

Risk and guardrails Integrity and user experience constraints put a ceiling on how aggressive you can be. Know where the guardrails are before you commit to a number you'll need to hit regardless of what you learn.

Reconcile top-down and bottom-up

Good goal-setting requires holding two lenses at once.

Top-down Where does the business need to go? For a mature newsfeed product, the answer is probably "protect and deepen engagement, don't try to grow your way to 10%."

Bottom-up Given this team's capacity, constraints, and mix of big bets versus optimizations, what can we actually move? For each bet or optimization, you're stating a hypothesis: "If we do X, we expect roughly Y lift on this input metric." Sum those hypotheses. That gives you a credible range.

In an interview, you want to show how you reconcile the two: "Top-down, we're focused on retention and sharing depth on a mature surface. Bottom-up, given our mix of big bets and optimizations and their expected impact, that leads me to a 1-3% target on the north star."

That's a real answer. "3% because it sounds good" is not.

Make the goal hard to game

A goal that can be hit the wrong way isn't a goal - it's a loophole.

Call out explicitly what you're *not* counting. "Cheap" interactions - ones that drive the metric without driving value (like making a like button 20% bigger) - should be excluded or flagged. Pair your primary metric with guardrails: an engagement quality score that mixes likes, comments, and shares, plus integrity metrics, so that hitting the number through surface manipulation still counts as a miss.

What a real answer sounds like

"Given a mature, optimized surface, the current baseline, and a mix of big bets and optimizations we can ship this half, I'd target around 1-3% improvement. Top-down, the focus is retention and sharing depth on a very mature product. Bottom-up, given our hypotheses of impact from ranking, integrity, and creator-supply changes, 2-3% is credible if we have strong iteration capacity and team health - and closer to 1% if we're more constrained. I'd pair it with guardrails so we don't win by degrading experience or integrity."

That's not a number. It's a reasoned range with a logic structure behind it.

Why is it important?

Setting goals too low gets you an uncomfortable conversation with leadership who'll push a more aggressive number down to you anyway. Setting them too high gets you in endless review meetings explaining why the team is behind. The goal is to find the target that's ambitious and fair - the one your team can believe in and execute against.