OKR vs KPI: which goal system to use based on your team's size

KPIs measure the ongoing health of the business; OKRs mark the change you want to force in a cycle. When to use each if you're alone, on a team of 3-10 or 10-50, with LATAM examples.

The difference that matters
Which system to use by size
Bad OKR vs OKR you can use
The quarterly cycle, without theater
Three mistakes that burn the system
Minimal implementation: Notion or a sheet
Frequently asked questions

The difference that matters

KPI (Key Performance Indicator) is a continuous indicator. It lives all year. It has an acceptable range, not an expiration date: revenue, margin, churn, NPS, tickets, utilization, cash days. If it goes out of range, something is sick. If it stays inside, the business breathes.

OKR (Objectives and Key Results) is a cycle bet. The Objective is the change you want to see. The Key Results are the numerical proof it happened. At the end of the quarter it closes: fulfilled, half-way or not. It doesn't "maintain". It gets archived, or you decide that front is no longer a priority.

KPIs are the car's dashboard. OKRs are this quarter's destination. You don't turn off the dashboard because you have a destination, and "keep the tank above a quarter" isn't a trip.

That's why they aren't enemies. A KPI warns you churn went up. An OKR might be "make the new client survive their first 60 days". If churn is already in range, you don't need an OKR: stop touching it.

	KPI	OKR
Question it answers	Is the business okay?	Did we achieve the change we said?
Duration	Continuous	One cycle (almost always a quarter)
Healthy amount	Few, stable, always visible	1–3 objectives per team
If it doesn't move	Can be good news	The OKR was decorative, or nobody executed it
Typical owner	Whoever operates the number	Whoever can change the system, not just report it
Used for	Detect, compare, alert	Focus, say no, force a bet

If your "OKR" is "maintain MRR" or "don't lower quality", it's not an OKR. It's a KPI in disguise. If your "KPI" is "launch the new app", it's not a KPI. It's a project.

Which system to use by size

The mistake in copying Google isn't using OKRs. It's using the ceremony of a thousand-person company when you all fit at a restaurant table. The system has to fit in the time you'll actually give it.

1 person (freelancer, solo founder, consultant)

You don't need company, area and individual OKRs. You need 4–7 KPIs telling you if the business holds, and at most one OKR if this quarter you're changing how you make money or how you work.

Typical KPIs for someone alone: collected revenue (not invoiced), pipeline for the next 8 weeks, billable hours vs. operating hours, cash days, concentration in the biggest client, renewals if you sell retainers.

The OKR comes in only when there's real change: moving to retainers, hiring the first person, or launching a product alongside the service. If the quarter is "keep serving clients well", don't write an OKR. A solo founder with 12 key results is doing theater for an audience of one.

3–10 people (agency, studio, small SaaS)

Here a team goal system starts paying off. Not because of "cascading". Because if it isn't written down, everyone optimizes their own craft and the business is left without an owner.

Use:

A company KPI board (weekly, 30 minutes): revenue, margin, on-time delivery, retention, cash, and one quality metric.
1–3 company OKRs for the quarter. No per-person OKRs. If you're eight, an individual OKR is a performance plan in the wrong place.
KPI owners different from OKR owners, when possible. Whoever reports churn doesn't have to be the one redesigning onboarding.

At this size, the value of the OKR isn't ambition. It's the no. If the objective is "get the new client to value without a call from you", then it's not the quarter to redesign the website, open a vertical and attend three trade shows.

10–50 people (several teams, first layer of managers)

Now team OKRs do appear, below the company ones. Individual OKRs are still unnecessary, except for a mandate that doesn't fit the team's.

Company: 1–3 objectives.
Each team: 1–2 objectives that explain how they push the company's, not a parallel menu.
Weekly company KPIs. Team KPIs only if the team operates a system (support, sales, delivery).
One ritual only: a biweekly 25-minute check-in. Not a half-day "OKR day".

At 30 people, 20 OKRs is the symptom that nobody chose: every leader brought their list and the spreadsheet gets updated the day before the meeting. If a team can't point to one number that moves the company OKR, it has a work plan, not an OKR. That's fine. Not everything deserves ceremony.

Bad OKR vs OKR you can use

An Objective answers: what changes in the business if this goes well? Someone non-technical understands it. It's not a project ("migrate to HubSpot"). It's not a slogan ("be leaders").

A Key Result answers: what numerical evidence would prove that happened? It has a baseline, a target and an owner. If you can't measure it on Friday without building a data project, it's not a KR: it's a wish.

Three LATAM cases. The bad one looks like what people put in Notion the first time.

Agency (8 people, retainer + projects, Bogotá or CDMX)

Bad. Objective: be the best agency in the city. KRs: more likes, 20 proposals, hire two creatives. "Best" can't be verified. Likes don't pay payroll. Sending proposals is activity. Hiring is a means.

Good. Objective: make the retainer feel inevitable to the client, not a monthly expense.

KR1: retainer retention from 72% to 85% (6-month cohorts).
KR2: 6 of 10 clients with a reported business result in the first 30 days (today it's 2 of 10).
KR3: revenue per retainer client from X to 1.25X, without adding headcount.

The team knows what to leave out: a pitch for a client who doesn't retain, an award-winning campaign that doesn't move the client's business, a hire "to keep up" that doesn't attack the cause.

Small SaaS (12 people, invoicing for SMBs, Colombia / Mexico)

Bad. Objective: grow. KRs: 10,000 users, 8 features, press coverage. You can "fulfill" it and the company stays just as fragile.

Good. Objective: make a new SMB issue its first invoice without writing to support.

KR1: activation (first invoice) from 28% to 45% within the first 7 days.
KR2: onboarding tickets per 100 signups, from 40 to 15.
KR3: time to first invoice, from 4 days to 1 day (median).

MRR, churn and margin stay as KPIs. The OKR doesn't replace them: it picks one bottleneck for the 12 weeks.

Freelancer starting to hire (you + 2 collaborators)

Bad. Objective: scale the business. KRs: more clients, more revenue, hire a community manager. "More" isn't a KR. Hiring someone to post doesn't change the fact that you're still the bottleneck of every delivery.

Good. Objective: stop being the bottleneck of every project.

KR1: 60% of weekly deliveries go out without your final review (today, almost none).
KR2: an onboarding playbook used in 5 consecutive projects.
KR3: your execution hours (design, code, editing) from 30 to 12 per week, without dropping collected revenue.

If by the end you hired and you're still reviewing everything at eleven at night, the OKR failed. Even if the Instagram looks more "agency-grade".

The quarterly cycle, without theater

A quarter is the right size under 50 people. A month is too short for real change; a year is so long people pretend there's time. If a more tactical framework helps you, the 12-week method fits underneath: the OKR says what changed; the 12 weeks organize how not to leave it for November.

Week 0 (write). Two hours, not two days. Real problems from last quarter. Pick one or two. Objective in one sentence. 2–4 KRs with a baseline. If you don't have a baseline, the first KR is "measure X for 2 weeks" and only then set the target. Publish the doc.

Weeks 1–11 (execute). Weekly 15-minute check-in if you're under 10; biweekly if more. The OKR doesn't get redesigned. You answer: did the number move? what got in the way? what gets cut this week? The traffic light (red / yellow / green) is enough.

Week 12 (close). Each KR gets scored. The classic convention treats 0.7 as a good result on ambitious goals; a steady 1.0 usually means the target was small. If in your culture 0.7 reads as failure, use "achieved / partial / no". What matters is the retro.

Between quarters, one empty week of new goals. Collect, deliver, breathe. A system with no gap becomes another job.

Three mistakes that burn the system

Twenty OKRs. If everything is priority, there's no system: there's a wish list. An 8-person team with 15 objectives is saying it can't choose. Cut. What's left out goes back to the backlog. The OKR isn't an inventory of what you do; it's what you're unwilling to negotiate.

Vanity metrics. A number is vanity when it can go up without the client or the cash register noticing: followers, pageviews, registered users, features shipped. If it hits the target and the business could still be worse, it's not a KR. "10,000 users" with 80 paying is a press release.

OKR as bonus or as stick. The day salary depends on nailing the OKR, people negotiate small targets and dress up numbers. OKRs are for aligning and learning. KPIs, and a separate performance agreement, are for talking compensation. Mixing ambition with punishment produces prudence disguised as planning.

There's a fourth mistake, quieter: copying the vocabulary and not the habit. Buying software, building 40 pages of "philosophy" and not opening the doc for six weeks. The system is the check-in. The rest is decoration.

Minimal implementation: Notion or a sheet

You don't need Lattice, Ally, Weekdone or an "OKR coach". Under 50 people, a base or a shared sheet is enough.

Columns: cycle, Objective, Key Result, baseline / target / current, owner (one person), traffic light, one-line note, and whether it's a KPI or OKR.

In Notion: one OKR table and one KPI table, cycle filter, Monday reminder. In Sheets or Excel: one tab per cycle and one ugly chart of the KR that matters most. That's a system. A kickoff PDF nobody reopens isn't.

Top of the doc: max 3 company objectives; every KR with number, unit and owner; no OKRs added mid-quarter except crisis; KPIs get watched, not "fulfilled"; the check-in doesn't get cancelled.

If the doc goes unupdated two cycles in a row, it's not that "OKR doesn't work in LATAM". It's that the system weighs more than your discipline. Drop to one OKR and four KPIs.

Frequently asked questions

Can I have only KPIs and never OKRs?

Yes. It's the right call if the business is in operate mode: there's demand, the margin closes, and you're not attempting a change of shape. OKRs show up when you want the quarter to not look like the last one. Without that requirement, they're bureaucracy.

Can a KPI also be a Key Result?

Sometimes the same number lives on both sides. Churn is a KPI all year; this quarter, "bring it from 6% to 4%" can be a KR. Don't duplicate it: clarify whether this cycle is watching it or moving it.

Should every person have their own?

In teams under 50, almost never. The individual OKR becomes a performance review with another name. If someone needs clarity, give them 2–3 results inside the team's OKR, not a parallel set. The task kanban doesn't mix with this doc.

Pick few numbers that say whether the business breathes. Pick even fewer changes worth a quarter. Write them where the team already works. Review them even when the week got ugly. That's a goal system. The rest is vocabulary.