Goal Setting

An eight-person company has nowhere to write goals down. What matters most this month gets settled in a single meeting, and after that everyone can see what everyone else is working on. When the founder says the priority right now is getting free teams onto a paid plan, that sentence is the goal. Turning it into a document would add nothing.

Past forty people the method breaks in three places at once. Every team works hard by its own standard, and those standards point in different directions. When the period ends there is no basis for saying whether it went well. And because there is no basis, evaluation falls back on memory and impression.

At Pagedeck the problem first surfaced in the all-hands during the last week of a quarter. Grace asked whether the quarter had gone well, and every team arrived with a different number. Theo said they had shipped forty-one releases. Ravi said visitors were up 30 percent. Sofia said she had closed nine contracts. There was no way to add those three numbers together and no way to compare them. The harder part was that none of them connected to whatever the company had been trying to accomplish that quarter — because what the company had been trying to accomplish had never been written down anywhere.

The same gap showed up again in performance reviews. Noa had spent the quarter pulling scattered design elements into order, and in her review she was told the impact had been small. Nobody had asked her to do that work at the start of the quarter, and nobody had asked her not to. When the standard appears after the period closes, any piece of work can fail it, and the person who fails has no way of knowing what would pass next time.

Goal setting exists to move that standard forward in time. Before the period starts, the work to be accomplished and the way of checking it get written down together, and when the period ends people are measured against those sentences rather than against somebody's recollection. The People axis follows a person from the job opening through to the exit and splits the path into nine branches, and the evaluation branch is the only one holding two terms. This one owns the front half, deciding what will be accomplished, and performance management owns the back half, checking results against what was decided and adjusting from there. Drop either half and the other spins. Without written goals a reviewer has nothing but personal impression to work from, and without a checking routine the document written in week one sits unopened until the quarter ends.

Seven terms sit under goal setting, in four branches: what a single goal is made of (components), what format holds it (frameworks), how high it gets set (ambition), and how goals connect across the company (operation). The first two shape the document, and the last two decide how that document behaves once an organization uses it.

Components cover the parts that make up one goal. There are two, split by whether they are written in words or in numbers.

1. Objective — a sentence describing the direction worth heading in. It is written in words, not numbers. Pagedeck's objective last quarter was a single line: get teams who use the product for free to start paying and stay. A sentence like that tells a reader which way to go, but on its own it cannot settle whether the goal was reached. It always needs a partner.
2. Key result — a measurable outcome showing whether the company actually moved in that direction. The objective above carried two or three of them: raise the free-to-paid conversion rate from 4 percent to 7 percent, keep three-month retention among converted teams above 80 percent. There is one rule for writing them. Describe the change the work produced, not the work itself. "Redesign the onboarding screens" is not a key result. The redesign can ship while conversion stays exactly where it was, leaving the company with new screens and an unmet goal.

Frameworks decide what format and what rhythm hold those two parts. There are three, and they do not compete — each covers a different job.

1. OKR — objectives and key results, one objective carrying two or three key results, run on a quarterly cycle. What distinguishes it is that company-level and team-level entries stay visible to everyone. When a team can see what the team next door is attempting this quarter, two teams are less likely to build the same thing separately, or to wait on something the other one already dropped.
2. KPI — a key performance indicator, a number chosen once and watched continuously. If OKRs are the numbers a team is trying to move this quarter, KPIs are the instrument panel that stays on. Monthly churn or response time are not picked because someone wants to improve them but so that a decline gets noticed immediately, which is why they are usually written with a threshold not to cross rather than a target to hit. Quarterly objectives get replaced when the quarter ends; the indicators stay in the same place and get read again.
3. SMART goal — five tests applied to how a goal sentence is written: whether it is specific, measurable, achievable, relevant to what the team is doing now, and time-bound. It is not another format but a ruler held against a sentence already written. "Improve customer satisfaction" fails three of the five, because it never says what improves, by how much, or by when.

Ambition covers how high a goal gets set, and it holds one term. A stretch goal is deliberately set beyond what the current approach would reach. It only works with one agreement attached: reaching 70 percent of it counts as a good quarter. Without that agreement people write down whatever they are sure of clearing, and the goal-setting meeting turns into a negotiation. Pagedeck set paid teams at 100, up from 40, and finished the quarter at 72. Had the target been 50, the quarter would likely have stopped near 52. This is the one place where a missed goal leaves better results behind than a met one. The agreement has to exist before the goals get written, though. Lowering the bar after the quarter closes buys nothing, because nobody believes the promise the next time around.

Operation covers attaching those goals to the organization, and it holds one term. Cascading goals means carrying company goals down to teams and individuals. The sequence has three steps: the company-level goals get fixed first, each team writes in its own words which part of them it is taking on, and individuals take their share from the team's version. Letting teams write their own and then reconciling upward lasts longer than slicing goals at the top and handing pieces down, because people reread sentences they wrote and rarely reread sentences they were handed. The risk to watch is drift: every step down moves further from the original. If the company goal is paid conversion, the engineering team takes on reducing payment failures, and one engineer's goal becomes cleaning up the billing module, that third-level item may no longer touch conversion at all. Many companies handle this by asking every individual goal to name, in one line, which team key result it moves.

The four branches run in a line. Components build a single goal, frameworks give it a format and a cycle, ambition sets its height, and operation attaches it to the rest of the company. Leave one out and the others stall. A direction with no key results gets read differently by everyone at quarter's end. With no format or cycle, the writing never happens at all. With no agreement about height, everyone writes down what they are certain to clear. And with no alignment, every team does something defensible while the company's numbers stay flat.

"That's a task, not a key result. Write what changed, not whether you finished it."

An entry phrased as completion needs to be rewritten as a measurable change.

"Let's cut this quarter down to three goals. Eight means we haven't prioritized."

The complaint isn't the count. It's that nobody has decided yet what to drop.

"Keep that number as an indicator we watch and leave it out of this quarter's goals."

A request to separate the numbers a team is trying to move from the ones it monitors.

Pagedeck wrote its first quarterly goal document the year it reached 45 people, and the first attempt came close to failing. Every team was asked for five, the company ended up with twenty-three, and by the end of the quarter roughly half had never been touched. Nobody could even ask why they went untouched, because everyone had known while writing them that twenty-three was more than the company could do.

The next quarter Grace capped company goals at three. Instead of writing their own, each team picked one of the three and attached its key results to it. Something nobody had planned for came out of that. Ravi and Sofia landed on the same goal, which forced them to agree for the first time on which of the leads marketing generated sales would actually take. Until then each had been watching only their own number, and neither had raised the fact that marketing's best month could become a sales problem. Cutting the number of goals exposed the seam between two teams. Mia summed it up afterward: the value of goal setting is less in the finished document than in the conversations that writing it forces.

The most common mistake is reading a goal as a to-do list. Ask people what they will do this quarter and a task list is what comes back, and tasks are comfortable to write because they are within one person's control. But the goal is the place to record what will be different afterward, not what will be done. Fill it with tasks and the quarter ends in a strange state: every item marked complete, not one company number moved, and nobody able to object, because the work was delivered exactly as written. Two quarters of that and the goal document becomes a record of effort, at which point no decision gets made from it again.

Turning attainment percentages directly into review scores is the second frequent error. The goals were set in advance, so scoring against them looks fair, and the number makes the reasoning easy to explain. The problem is that the moment attainment becomes a score, the act of setting goals changes. People write down what they are sure of clearing, and when circumstances shift mid-quarter they leave the original wording alone, because revising it costs them. Stretch goals disappear from that company. Left alone, the pattern ends with company-wide attainment at 105 percent and the business sitting exactly where it sat last quarter. So attainment gets used as material for a conversation rather than as the sole basis for a rating: the number is read next to why it came out that way and what changed during the quarter.