Before you draw anything, write down what would have to be true for this work to have been worth doing. Not "users will love it". A behaviour, a number, a date.
Teams skip this because early on the goal feels obvious and the quarter feels long. Then the thing ships, the demo goes well, everyone feels good, and six weeks later nobody can say whether it worked. The argument about whether to keep investing gets settled by whoever is most senior in the room.
Success defined after the fact is always achieved. That is precisely why you define it first — it is the only moment when you can be honest, because you have nothing to defend yet.
Goal Setting
A goal is a sentence containing a metric, a direction, a size and a date. "Raise the share of new teams who invite a second member within seven days from 22% to 35% by the end of Q3." Anyone on the team can now judge whether a design idea serves it.
The common failure is writing outputs and calling them goals. "Ship the new onboarding" is a plan. You can deliver it on time, in full, and change nothing. Attach the number that makes shipping worth the money.
The second failure is having six of them. Six goals is no goals — the first time two conflict, the loudest voice picks. Choose one you would be embarrassed to miss, and let everything else be a thing you merely measure.
- Setting Product Goals: A Strategic Guide for Product Teams — article
- What Is a Product Goal? Definition, Examples & Best Practices — article
Product Vision
A vision describes what the world looks like once you have won, three or four years out. It is not a slogan and it is not a roadmap. It sits above the roadmap so that when the roadmap changes — and it will — the direction survives.
The test of a vision is whether it rules something out. "The leading platform for modern teams" rules nothing out; a competitor could hang it on their wall unedited. "Any freelancer can send a compliant invoice in a country they have never worked in, in under two minutes" tells you to build tax logic and not to build a time tracker.
Write it so a new engineer reads it once and can correctly predict which of two features you would cut.
- What is a Product Vision? — article
- Product Vision — article
Value Proposition
A value proposition names one benefit, for one kind of person, against one named alternative. The third part is the one teams leave out.
Your real competitor is rarely the product in your benchmarking deck. It is a spreadsheet, an intern, a WhatsApp group, or carrying on exactly as before — and the last one is undefeated, because choosing it asks nobody to do anything. The status quo has a cost; it just never arrives as a decision. If your proposition only beats a rival product, you have not explained why anyone changes their Tuesday.
Say it in the customer's words, with their number in it: close the month in two days instead of nine, without hiring a second bookkeeper. If you cannot name what you are displacing, you do not have a proposition. You have a feature list with adjectives.
- Value proposition: the key to winning customers and driving business growth — article
- How to Create a Compelling Value Proposition, With Examples — article
Product-Market Fit
Fit shows up as pull. Usage carries on without you prompting it, people come back on their own, and they hand the product to a colleague without being asked. Support changes shape — fewer people trying to get unstuck, more asking for what comes next.
The trustworthy signals are behavioural. The strongest is a retention curve that flattens rather than decaying to zero: a stable share of each cohort still active at week eight. The "how disappointed would you be if this disappeared" survey is a cruder proxy, but it costs an afternoon.
What it is not: signups, press, a waiting list, or growth that evaporates when the ad spend stops.
Before fit, your job is to change the product quickly. After fit, your job is to stop breaking it. Designing as though you have fit when you do not is how a year disappears into polishing something nobody would have missed.
- What is product-market fit? Examples and strategies to find it — article
- How to Find Product-Market-Fit as Fast as Possible — video
Success Metrics
Pick one primary metric and at least one guardrail. The primary metric moves in the direction you want. The guardrail names what you are not willing to break — support contacts, refund rate, time to first value, p95 load time.
Without a guardrail, almost any metric can be moved by making the product worse. Push harder in the signup flow and activation rises; so do uninstalls. A metric optimised alone will eventually be gamed rather than earned.
Prefer metrics that only move when someone actually got something. Sessions per week is not one — it climbs when people cannot find things. Invoices sent, files shared, second-week return rate: each costs the user effort, so they rise only when the effort paid off.
Write the number and the date down before the experiment runs, where the whole team can see them. A target agreed in advance is the only one that can still tell you no.