How to Create KPIs From Your Strategy

The fastest way to learn how to create KPIs is to stop starting from a blank page. You already have a strategy: a plan, a board paper, a few slides from last year's offsite, a set of priorities the CEO repeats in every all-hands. Good KPIs come out of that material by asking four plain questions of it, in order: Where are we now? Where are we going? What could affect us? What must we deliver?

The answers give you a short list of company KPIs. Each department then translates them into five to seven of its own, and every one of those gets a baseline, a target and one named owner before the year starts. That is the whole method. The rest of this article walks through it with a single worked example, so you can see what each step actually produces.

What a KPI is, and what it is not

A KPI is a number that tells you, while there is still time to act, whether the organization is getting where it said it was going. "Improve customer experience" is an intention. "Orders delivered on time and in full" is a KPI. And if you cannot say which priority a number serves, it is a report, not a KPI.

What a KPI is not: a count of activity. Calls made, tickets touched, emails sent and meetings held all describe effort. They are worth watching as early signals, and they are very easy to grow without anything improving. A dashboard full of them is how every status report stays green while the year drifts; vanity metrics covers how to spot them.

A KPI is not an objective either. Objectives say what you want; KPIs say how you will know. If your team already runs OKRs, OKR vs KPI explains where each one earns its place.

How to create KPIs: start with four questions, not a template

Most KPI projects start from a list of example metrics and pick the ones that sound right, which produces numbers that are individually sensible and collectively unrelated to the strategy. The four questions reverse that: the strategy goes in first. They work as a KPI framework because each one produces a different kind of measure:

Where are we now? produces your baselines.
Where are we going? produces your outcome KPIs and their targets.
What could affect us? produces your early-warning and guard KPIs.
What must we deliver? produces the commitments you cannot miss, and the department KPIs that protect them.

Ask them in that order. Teams that skip the first question set targets with nothing to measure them against. Teams that skip the third get blindsided by something they already knew about.

The example we will use

To keep this concrete, we will follow one fictional company all the way through. This is an example, not a client, and every figure in it is illustrative.

Harbor Supply is a regional distributor of plumbing and heating parts. It has 200 staff, three branches, a central warehouse, and about 2,800 active trade accounts, mostly contractors. Revenue last year was $60 million. The leadership team has a strategy document from its last offsite, and the headline in it reads: "Grow to $75 million in three years by winning more of each contractor's spend, not by chasing new contractors."

That sentence is a strategy. It is not yet measurable. Here is how the four questions turn it into KPIs.

Question 1: Where are we now?

This question is about honesty, and it is the one most often skipped. Before you set any target, write down where the organization actually stands on the things the strategy cares about. These figures become your baselines. Answer with the numbers the strategy needs, not just the ones you already report. Harbor Supply's strategy talks about "each contractor's spend", so the leadership team needs to know what share of that spend it already wins, which is not a number its finance pack contains.

Harbor Supply's answers looked like this:

Measure	Where we are now	Source
Revenue	$60.0M	Finance, last full year
Gross margin	22.4%	Finance, last full year
Revenue from top 300 accounts	$33.6M (56%)	Sales ledger
Estimated share of wallet, top 300	~35%	Account reviews with 40 contractors, extrapolated
Orders delivered on time and in full	88%	Warehouse system, last 6 months
Warehouse staff turnover	31% a year	HR records

The share-of-wallet figure is an estimate, labeled as one. A rough baseline everyone understands beats a precise one you will have "in a few months". If you have no way to measure something the strategy depends on, the first KPI for that area is simply to start measuring it. Give it a month, then set the target.

Question 2: Where are we going?

This question turns the strategy's ambitions into destinations with numbers and dates. Most leadership teams have answered it already, in words that mean slightly different things to each person in the room. Test that before you go further. In a 2018 MIT Sloan Management Review study, "No One Knows Your Strategy — Not Even Your Top Leaders", Sull, Sull and Yoder found that only 28% of executives and managers responsible for executing strategy could name three of their company's strategic priorities. If the people running departments cannot list the priorities, they cannot build KPIs from them. Asking each leader to write down the top three on their own, before the meeting, is a quick check; strategic priorities has more on what to do when the lists disagree.

Harbor Supply's leadership agreed on three priorities and one number to sit above them:

The company's main measure: revenue from the top 300 accounts, from $33.6M to $45M in three years. This is the number the whole strategy is trying to move. If you want a single measure that every department can see itself in, a north star metric is the idea to borrow.
Priority 1, share of wallet: from roughly 35% to 45% across the top 300.
Priority 2, reliability: on-time-in-full delivery from 88% to 96%, because contractors do not give more of their spend to a supplier that makes their crews wait.
Priority 3, margin: hold gross margin at 22% or better while growing, so the growth is not bought with discounts.

Each destination has a starting point from question 1. That pairing, baseline and target, is what makes a target mean something. The difference between measures that predict and measures that confirm is covered in leading and lagging indicators.

Question 3: What could affect us?

Strategies are written as if the world will hold still. This question makes the leadership team name the things that could knock the plan off course, inside and outside the company, and then decide which of them deserve a number of their own.

Harbor Supply's list had eight items. Four were worth measuring:

A national chain is opening a branch in its largest territory. Measure: revenue from the 40 accounts nearest that branch, reviewed monthly, so a loss shows up in weeks rather than at year-end.
One supplier provides 38% of purchases. Measure: share of purchases from the largest supplier, target below 30% by year-end.
Copper prices swing. Measure: gross margin on copper lines specifically, because a company-wide margin figure would hide the problem for a quarter.
The warehouse cannot keep staff. Measure: warehouse turnover and days to fill a vacancy. At 31% turnover, the reliability target in question 2 is not realistic.

These are guard KPIs: they stop a good plan failing for a known reason, and most move before revenue does. The other four risks were given an owner but no KPI. Not every risk needs a number; the test is whether watching one would change a decision in time.

Question 4: What must we deliver?

The first three questions are about choices. This one is about commitments: the things the organization has already promised to customers, lenders, the board or regulators, which the plan has to protect no matter what.

Harbor Supply's commitments:

To customers: next-day delivery on any order placed by 3 p.m.
To the bank: debt below 2.5 times earnings, a loan condition.
To the board: earnings of $3.6M this year.
To staff: the safety record. A warehouse running harder to hit 96% on-time delivery must not start having more accidents.

This question is also where department KPIs come from. Once the leadership team knows where it is going and what it cannot drop, it can ask each department: what must you deliver for this to happen? Sales must deliver more revenue from existing accounts. Operations must deliver the reliability. Purchasing must deliver the supplier spread and protect the margin. People must deliver a warehouse that keeps its staff.

The four questions at a glance

Question	What you decide	Example output (Harbor Supply)
Where are we now?	Your baselines, including the ones nobody has measured yet	On-time-in-full 88%; share of wallet ~35%; warehouse turnover 31%
Where are we going?	The main measure, three priorities, each with a target and date	Top-300 revenue $33.6M → $45M in three years; on-time-in-full 96%
What could affect us?	Which risks get a guard KPI and which just get an owner	Largest-supplier share below 30%; copper-line margin reviewed monthly
What must we deliver?	The commitments to protect, and each department's part in them	Next-day delivery for orders by 3 p.m.; debt below 2.5× earnings

If you want to fill this in for your own organization, the KPI template is the same four questions laid out as a worksheet, one row per KPI.

From company KPIs to department KPIs: five to seven each

The company set above has around a dozen measures. Departments should not inherit all of them, and they should not copy the company's main number onto their own scorecard either. Giving every department "revenue" tells nobody what to do differently on Monday. Each level translates the one above it into something it can move. Company, department and individual KPIs covers how the levels relate, and Hoshin Kanri is the long-established version of the same idea, where each level proposes its own measures and negotiates them with the level above.

Five to seven KPIs per department is the range that works. Fewer than five and you are measuring one part of a job that has several. More than seven and the department will quietly pick the three it cares about without telling you which. A mix that tends to hold up is one or two outcome measures, two or three early measures that predict them, and one guard measure that stops the outcome being hit the wrong way.

Here is what Harbor Supply's operations department ended up with:

KPI	Type	Baseline	Target (year-end)	Owner
Orders delivered on time and in full	Outcome	88%	93% (96% in year 3)	Head of Operations
Orders picked same day by 5 p.m.	Early	81%	92%	Warehouse Manager
Stock availability, top 500 lines	Early	94%	98%	Inventory Lead
Delivery routes leaving by 7 a.m.	Early	70%	90%	Transport Manager
Warehouse staff turnover	Guard	31%	20%	Head of Operations
Lost-time safety incidents	Guard	4 a year	2 or fewer	Health & Safety Lead

Six KPIs, each tied to a priority or a commitment, each with a starting point. Sales, purchasing, finance, customer service and people each got a similar table. For starting points in other functions, KPI examples by department has five to seven per department, and how to set team KPIs covers the same translation one level further down, with the team in the room.

Every KPI needs a baseline, a target and one owner

A KPI without these three is a wish. In practice the gaps are easy to spot.

A baseline is where the number stands today. Without one, a target is a guess: "improve on-time delivery by 10%" means nothing until someone knows whether the current figure is 88% or 58%.

A target is where the number should be and by when. Set it with the baseline in front of you. Harbor Supply's year-three goal for on-time delivery is 96%, but the year-one target is 93%, because the warehouse turnover problem has to be fixed first and everyone knows it.

One named owner is a person, not a department or a committee. "Sales" does not own share of wallet; the Sales Director does. The owner need not move the number alone, but they report on it and explain it when it slips. A KPI owned by three people is owned by nobody, which is why accountability in the workplace starts with one owner per KPI.

The five-point KPI check

Before a KPI goes on the list, put it through five questions. This is the familiar SMART test applied to KPIs, and it catches most of the bad ones. A KPI that fails a point is not necessarily wrong, but it needs rewriting before anyone is held to it.

Specific. Is it defined clearly enough that two people would calculate the same number? "Customer satisfaction" fails. "Share of delivery surveys scoring 4 or 5 out of 5" passes.
Measurable. Can you get the number without a project? If producing it takes an analyst a week, you will see it once a quarter at best.
Achievable. Given the baseline, is the target a stretch rather than a fantasy? On-time delivery from 88% to 93% in a year is hard; to 99%, with a third of the warehouse leaving, is not a target.
Relevant. Can you name the priority or commitment it serves? "It's always been on the dashboard" fails.
Time-bound. Does the target have a date, and the KPI a regular review?

Harbor Supply ran 61 candidate KPIs through the check across six departments. 38 survived, some of them rewritten. The 23 that failed were mostly activity counts and measures nobody could produce without a new system.

Settle conflicting KPIs between departments before the year starts

Department KPIs drafted separately will conflict. That is normal, and far cheaper to find in a planning meeting than in the third month of the year. Harbor Supply found three the first time the lists were put side by side:

Purchasing vs operations. Purchasing wanted to cut inventory to release cash. Operations needed 98% availability on the top 500 lines to hit its delivery target. The resolution: purchasing's inventory target excludes the top 500 lines.
Sales vs finance. Sales planned to win share of wallet with volume discounts. Finance owned the 22% margin floor. The resolution: discounts above 3% need the Finance Director's sign-off, and sales also reports margin on the top 300 accounts, not just revenue.
Customer service vs operations. Customer service measured how fast it answered calls; most calls were about late deliveries. The resolution: customer service also owns "calls per 100 orders", which only falls if the delivery problem is fixed.

The method is the same each time. Put every department's KPIs on one page, read across the rows, and for every pair that pull against each other, agree which one gives way, or add a second measure that keeps both honest. When a leadership team says it is aligned but everyone means something different, leadership alignment is the conversation to have. A strategy map helps too, because drawing the lines from each department's KPIs to the company's priorities shows the conflicts immediately.

Do this before the first quarter of your financial year starts. Once the year is running, every conflict becomes a negotiation about whose quarter gets worse. If your planning cycle is still ahead of you, annual planning covers the few weeks where these decisions are cheapest, and an offsite agenda built around the four questions ends the retreat with KPIs rather than a document.

Review KPIs quarterly, not annually

A KPI set is a plan, and plans go out of date. Review the whole set every quarter, with the owners in the room.

Ask three things of every KPI. Is it on track? If not, what is the owner doing about it, and what do they need? And is it still the right measure? Harbor Supply retired "routes leaving by 7 a.m." after two quarters at 95%. A target set on a bad baseline is better changed openly than quietly ignored.

A quarterly business review is the natural home for this, as long as it checks the strategy rather than just reading out the figures. It needs a one-page view that the leadership team and board can follow in five minutes; KPI dashboard examples shows what that page can look like. Whatever system holds the numbers afterwards, the leadership team decides what goes into it. Performance management software can track KPIs and connect them to people's goals, but it cannot tell you which KPIs to have.

Using AI to draft KPIs without handing over the decision

Drafting a KPI set is slow work, and AI does it well. Given your strategy document and the answers to the four questions, a capable model will propose candidate KPIs, suggest an early measure for each outcome, and flag where two departments' drafts conflict. What it cannot do is decide: it does not know that warehouse turnover is your real constraint, or that the board will not accept a margin below 22%. AI drafts, leaders decide, and every KPI that survives has a human owner who agreed to it. AI for strategic planning covers where it helps most, and if your leadership team is going to use it regularly, agree the ground rules first with an AI governance framework.

If you would rather do all of this in one sitting with the strategy you already have, a strategy workshop built around the four questions is designed to end with KPIs for the organization and every department, each with an owner.

Frequently asked questions

How many KPIs should a company have?

Around three to five at company level, sitting under one main measure, plus five to seven per department. A company of 200 people with six departments ends up with roughly 35 to 45 KPIs in total, but no single person is watching more than seven of them. If the leadership team's own list runs past ten, some of those numbers belong to departments.

How do you create KPIs if you have no data yet?

Make "start measuring it" the first KPI. Pick the simplest reliable way to collect the number, run it for four to six weeks, and set the target once you can see the baseline. A rough baseline from a sample, clearly labeled as an estimate, is better than waiting months for a perfect one.

What makes a KPI SMART?

A SMART KPI is specific, measurable, achievable, relevant and time-bound. In practice that means one clearly defined number, data you can actually get, a target that stretches from a known baseline, a link to a named priority, and a date with regular reviews. The five-point KPI check above runs every candidate through those five questions before it goes on the list.

How often should KPIs be reviewed?

Owners should look at their own KPIs weekly or monthly, depending on how fast the number moves. The leadership team should review the whole set every quarter, checking progress, retiring measures that have done their job and correcting targets set on a bad baseline. An annual review alone is too late to change anything that year.

How to Create KPIs From the Strategy You Already Have