Business objectives and key results

Product strategy is the bridge between a vision and the objectives: a few explicit choices about which customers and which problems to focus on next, and what the company will deliberately not do. The vision says where you're heading. The strategy picks the road. The objectives name the measurable results that road should produce.

A real strategy rules something out. “Be the best cat food service” chooses nothing, so it can't stop a single request. “Grow by keeping subscribers and through vet clinics, not by buying new owners with ads” makes two bets and names the route it drops.

At CatChow those two bets become the two business objectives: keep subscribers longer, and grow the vet clinic channel. Each one then gets key results that show whether the bet is paying off.

Two quick checks before the objectives go on a slide: can you name what the strategy says no to, and does every objective trace back to one of its choices? If not, the objectives are a wish list.

An objective says in words what the business wants to achieve: “grow subscription revenue by keeping subscribers longer”. A key result shows how you'll know you got there: a measurable change with a baseline and a target, such as “monthly churn falls from 9% to 6%”. Write two to four per objective, so no single number carries the whole story. The pair is often called OKR, objectives and key results.

Most business objectives belong to one of three families: protect what makes the product valuable to customers; grow, with more customers, new markets or more repeat revenue; improve profit through price, lifetime value, cost, or getting more from what you already have. If the why ladder from Outcomes over outputs, run on a request, ends in none of them, it isn't an objective yet.

Keep the levels apart. “Keep subscribers longer” isn't a key result: it has no number. Work such as “launch reorder reminders by March” or “replace the warehouse system” isn't one either: it can be finished while nothing changes for customers.

A key result measures a change, not a delivery. “Launch the clinic ordering portal by June” is specific and dated, but it's an output with a date: if the portal launches on time and clinics still run out of prescription diets, the key result turns green while the problem stays.

Run the output test from Outcomes over outputs on every draft: if you could tick it off while nobody behaves differently, it's a task. Then ask what should be different at PurrPoint, the 14-clinic chain in CatChow's pilot, if the portal works: fewer stock-outs, fewer phone orders. Measure those. The portal is one way to get there, and the pilot in two clinics will show whether it's the best one.

The launch items don't disappear. They go under the themes as candidate solutions, and the ones the team picks move into the release plan, which tracks what ships and when. The OKR keeps measuring what customers do; the release plan keeps tracking what the team builds. Each document answers its own question.

Is churn alone enough for the subscription objective? Not quite. Churn is a lagging number: by the time it moves, months have passed, and if the work isn't helping, you find out far too late to change course.

Pair it with a leading signal that moves earlier and points the same way. For CatChow, that's the share of cancellations that cite running out of food, currently 30%. If reminders and flexible delivery dates work, that share should drop within weeks, long before monthly churn shows anything.

That answers the board's likely question, “how will we know by spring?” In spring the board won't see churn at 6% yet, but it can see whether cancellations citing running out are falling towards 15%. If they aren't, the team knows to try another solution while there's still time.

A good set of two to four key results usually mixes both: one lagging number that proves the business result, and one or two leading ones that show early whether you're heading there.

A KPI (key performance indicator) is a health number you keep in an acceptable range all the time. A key result is a change you set out to make in a given period. The same metric can be either: churn is a KPI while you only watch it, and it becomes a key result the moment you decide to move it from 9% to 6% next year.

CatChow tracks many KPIs. One of them is the share of orders delivered on the promised day, about 97% every month. Nobody plans to raise it next year; the job is to keep it there.

That makes health KPIs useful guards beside the key results. While the team chases lower churn with flexible delivery dates, on-time delivery must not slip. If it drops below its usual range, that's a signal to stop and look, even while churn improves.

So each objective on the roadmap carries a few key results, with a handful of health KPIs listed next to them, not dozens of metrics dressed up as goals.

A key result needs a real starting point, and sometimes nobody has measured it yet. Sofia wants a key result about clinic phone orders, but “every week, one clinic at a time” is all CatChow knows.

Don't invent a number. A key result without a real baseline is a guess with a percent sign: if you assume “two hours a week” and set a target of one, any result is meaningless, because you never knew where you started.

Don't drop it either: it's a real clinic problem, and replacing it with “launch the portal” just swaps in an output. Instead, make measuring it the first step. The two pilot clinics can count their phone orders, or the time admins spend ordering, for a month. Then you set the target on real numbers. Saying “baseline to be measured in November” on the slide is honest and perfectly acceptable.

And keep the source next to every baseline, as in the input summary: where the number comes from and what period it covers. If it's based on only one month, say so.

Not every key result is a promise of the same strength. Eighty partner clinics feels like a stretch; losing any PurrPoint clinic would be a disaster.

Many teams separate two kinds of key results:

• Committed: the business expects to hit it in full. Keeping all 14 PurrPoint clinics is committed: missing it is a real failure.
• Aspirational: a stretch where getting about 70% of the way is already a good result. Growing from 54 to 80 clinics is aspirational.

Marking both as committed turns a stretch into a promise. Marking both as aspirational quietly makes PurrPoint optional. Dropping the stretch because nobody is sure throws away the ambition.

Say which is which on the slide, or the board will read every number as a promise: a disclaimer, but for numbers.

One more habit: set the key results for the year and check them every quarter. If quarterly OKRs appear later, they should be checkpoints towards these, not a new list each time.

OKRs tend to go wrong in the same four ways. Lucy, CatChow's tech lead, saw several at her last job: thirty OKRs, cascaded from the CEO down to every team, and more time spent updating the spreadsheet than building anything.

• Output key results: the team hits every key result, and nothing changes for customers.
• Too many OKRs: nobody can say which two or three things matter most this quarter.
• Cascading from the top: every team's key results are the CEO's numbers split by department, with no say from the people doing the work.
• Tying OKRs to bonuses: teams set targets they already know they can beat, and bad news stops being reported.

Each one breaks a different part of the method: outputs stop measuring change, too many objectives erase priorities, cascading ignores what teams can actually influence, and bonuses reward safe targets over honest ones.

The better way: two or three objectives. Leadership sets them, teams propose the key results they can influence, and the two meet in the middle. Then check that every key result measures a change, not a task.

Tying OKRs to bonuses is the one failure that usually arrives with good intentions. Ian likes the key results and has an idea for the board: pay the team for hitting them, because nothing motivates like money.

The motive is good: he wants people to care about the numbers. The effect is predictable. When pay depends on a target, people protect the pay:

• they set safe targets they already know they can beat;
• they stop reporting bad news early, because an honest red status now costs money;
• the aspirational key results quietly disappear, since a 70% result would look like a miss.

OKRs only work if they're honest, and money makes honesty expensive.

How do you say that to a CEO the day before the board? Don't argue with the motive. Acknowledge it, name the concrete risk in one sentence, and offer something he can say yes to: keep performance reviews separate from OKR scores, review the key results openly every quarter, and celebrate what the team learns as well as what it hits. Two or three sentences are enough.

For the objective “grow the vet clinic channel”, good key results are the ones CatChow can measure today: partner clinics from 54 to 80 (lagging), prescription-diet stock-outs from 6 a month to 1 (moves first), all 14 PurrPoint clinics kept. “Build the clinic ordering portal” is an output, “hold 12 meetings with clinic directors” an activity, and “the clinic channel is improved” has no number at all.

The warehouse replacement is months of Lucy's team's work that no customer will see. As “replace the warehouse system” it's an output, but it's what lets stock-outs fall and 80 clinics be served. Enabling work is judged by the key results it unlocks: link it to them, so everyone sees why it matters and nobody mistakes finishing it for success.

How it fits together: Vision → strategy → two objectives → key results, lagging and leading, with real baselines, marked committed or aspirational, and health KPIs beside them. That's the “why” layer of roadmap version 2. Module 3 turns customer needs into themes under these objectives.