Stratrix

Quick reference

Key formula / structure

Horizon one: extend and defend the core. Horizon two: build the emerging businesses. Horizon three: seed the options. Run all three at once.

Memory aid

The horizons are not a schedule. The core pays for the future, so start the future while the core is strong.

TL;DR

List every business and initiative. Sort each into horizon one (the core that brings in most of the profit and cash), horizon two (emerging businesses that need investment to scale) or horizon three (options on the future). Check that every horizon has something in it and that horizon two is big enough to matter. Give each horizon its own owner, budget and measure. Write a test for when each bet moves to the next horizon, and review the portfolio every year.

What Is Three Horizons?

The three horizons sort a company's growth into three groups. Horizon one is the core business, the one most readily identified with the company's name, which brings in most of the profit and cash flow. Horizon two is the emerging businesses, such as rising ventures that could become profitable but still need real investment. Horizon three is ideas for profitable growth further out: research projects, pilot programs, minority stakes in new businesses. The framework's point is that a company needs something in all three at the same time.

Time, as noted on the x-axis, should not be interpreted as a prompt for when to pay attention—now, later, or much later.

— McKinsey Quarterly, "Enduring Ideas: The three horizons of growth"

The drawing that goes with the framework puts time on the bottom axis, and most people read it as a schedule: core now, new businesses later, bold bets much later. McKinsey's summary of the framework says time "should not be interpreted as a prompt for when to pay attention—now, later, or much later," and closes on the opposite instruction: "The three horizons framework offers a way to concurrently manage both current and future opportunities for growth." The reason is money and lead time. Horizon three is paid for by horizon one, so it has to be started while horizon one is strong. A horizon-three bet takes years to grow into horizon two, so a company that starts when the core begins to slip has nothing ready. Meta shows the horizons run at once: a core earning more than $100 billion a year carrying a bet that has lost nearly $77 billion over five years. IBM shows the other path: with nothing in horizon two big enough, it bought Red Hat for about $34 billion at a 63% premium. The framework will not tell you which bets to place. It tells you that you need bets in every horizon, that each needs a different yardstick, and that the time to place them is while you can afford them.

Three horizons, run at the same time

Ventures move from horizon three to two to one as they grow. The core pays for all three.

Horizon 3: options

Research, pilots, small stakes. Measured on learning

→

Horizon 2: emerging

Growing businesses that need investment. Measured on growth

→

Horizon 1: the core

Most of the profit and cash. Measured on returns

Origin & context

Mehrdad Baghai, Stephen Coley and David White, then at McKinsey & Company, set out the three horizons in The Alchemy of Growth: Practical Insights for Building the Enduring Enterprise (Perseus, 1999; Basic Books paperback, 2000). The book's subject was how companies keep growing: by extending and defending the core, building new businesses and seeding options for the future, all at once. McKinsey Quarterly later summarized it in its Enduring Ideas series, presented by Steve Coley.

Builds on: Mehrdad Baghai, Stephen Coley and David White (1999), The Alchemy of Growth: Practical Insights for Building the Enduring Enterprise (Perseus; Basic Books paperback, 2000); McKinsey Quarterly, Enduring Ideas: The three horizons of growth; Steve Blank (2019), The Fatal Flaw of the Three Horizons Model.

Three Horizons in action

Meta

A $100 billion core paying for a bet a long way off

On October 28, 2021, Mark Zuckerberg announced that Facebook the company would now be called Meta. He said he expected to invest "many billions of dollars for years to come" in a 3D internet he admitted was still "a long way off."

In the language of the three horizons, that is a horizon-three bet, placed while horizon one was strong. Reality Labs, the division that holds it, lost $10.19 billion in 2021, $13.72 billion in 2022, $16.12 billion in 2023, $17.72 billion in 2024 and $19.19 billion in 2025: nearly $77 billion in five years, rising every year. Over the same years the Family of Apps, the core, earned $69.38 billion of operating income in 2023, $87.57 billion in 2024 and $101.15 billion in 2025.

The horizons explain why the bet could be placed at all: only a core that strong can carry a loss that size. Meta also reports Reality Labs on its own line, so the bet is neither hidden in the core nor judged by it. The cost of carrying it showed up elsewhere: about 11,000 layoffs in late 2022 and about 10,000 more in March 2023.

The framework's next question is the one only Meta can answer: what evidence would show Reality Labs moving from horizon three into horizon two, and by when?

Read the full story: Meta's bet on Reality Labs →

IBM

When horizon two is empty, you buy one

On October 28, 2018, IBM offered $190 a share in cash for Red Hat, whose stock had closed two days earlier at $116.68. It was a 63% premium, about $34 billion in all, funded with debt and cash, and IBM said it would suspend its share buybacks in 2020 and 2021.

What IBM bought was a horizon-two business it did not have: Red Hat's revenue had grown 21% to $2.9 billion in its 2018 fiscal year, and subscriptions made up 88% of its revenue in the fourth quarter.

Buying a horizon two is a legitimate move, and it worked: by mid-2023 Red Hat was leading IBM's software segment, which grew 7.2% to $6.6 billion. But the premium is the price of not having grown one.

The deal closed on July 9, 2019, and Red Hat led IBM's software growth by 2023.

Read the full story: why IBM bought Red Hat →

Using it in the room

Use it in the annual strategy and budget cycle, before the core's budget is set, and in any board review of where future growth will come from. Bring it out again when the core starts to slow and someone proposes buying growth.

Questions to ask

  • ?What is in each horizon today, and how much are we spending on each?
  • ?If the core stopped growing today, what would we be selling in five years?
  • ?Is horizon two big enough to matter, or is it a list of small pilots?
  • ?What would have to be true, and by when, for each horizon-three bet to move into horizon two?
  • ?What moved between horizons this year?

The 30-minute version

Ten minutes to list every business and initiative. Ten minutes to sort each into a horizon, by maturity, not date. Five minutes on the gap: is horizon two big enough? The last five minutes to name an owner and a graduation test for the two biggest horizon-three bets.

Core Components

1

Horizon one: the core

The businesses most readily identified with the company's name, which bring in most of the profit and cash flow. The job is to extend and defend them and get the most value out of them.

Example

Meta's Family of Apps, with $101.15 billion of operating income in 2025.

2

Horizon two: the emerging businesses

Ventures that have customers and momentum and could become a future core, but still need considerable investment to scale. The job is to build them.

Example

A new product line with paying customers that is growing fast but not yet profitable.

3

Horizon three: the options

Ideas for profitable growth further out: research projects, pilot programs, minority stakes in new businesses. The job is to seed several, learn cheaply and keep the ones that show promise.

Example

Meta's Reality Labs, home of the bet on a metaverse that Zuckerberg said in October 2021 was still "a long way off."

4

All three at once

The horizons are not stages a company passes through. They are three parts of one portfolio, run at the same time, with ventures moving from horizon three to two to one as they grow.

Example

A yearly plan that names the core's targets, the two or three businesses being built and the options being tested, side by side.

5

A different yardstick for each

Horizon one is measured on profit and cash, horizon two on growth and the path to profit, horizon three on what has been learned and whether the option is still worth keeping. Judging a horizon-three bet on horizon-one measures kills it; never judging it at all lets it run forever.

Example

Reporting a new venture as its own line, as Meta does with Reality Labs, so its numbers do not hide inside the core's.

Simon Wardley, the creator of Wardley mapping, calls the three horizons "useful as a high level concept but overall weak in practice on its own," because without a map of how markets evolve it is easy to miss what is coming.

When to Use Three Horizons

Scenario 1

The annual strategy and budget cycle

Problem it solves: Every dollar goes to the core because the core has the clearest return, and nothing new gets funded.

Real-world application

The plan sets aside money and owners for horizon two and horizon three before the core's budget is set.

Scenario 2

A board review of the growth pipeline

Problem it solves: The board sees one blended growth number and cannot tell where the next core will come from.

Real-world application

The CEO shows each horizon on its own line, with its own measures, and what has moved between horizons this year.

Scenario 3

Deciding whether to buy growth

Problem it solves: The core is slowing and there is nothing in horizon two large enough to replace it.

Real-world application

The team compares the cost of building a horizon-two business with the premium it would pay to buy one.

Scenario 4

Reviewing a long-running bet

Problem it solves: A horizon-three project has absorbed money for years and nobody has said what success would look like.

Real-world application

The project's sponsor is asked to name the evidence that would move it into horizon two, and the date.

Horizon two is the one that goes missing

Horizon one has its managers and horizon three has its champions. Horizon two, the businesses that are past the idea but not yet profitable, has neither, and it is where the next core has to come from. If your plan cannot name what is in it, that is the finding.

How to Apply Three Horizons: Step by Step

Before you start

  • →A list of every business, product line and growth initiative, with its revenue, profit and investment
  • →Agreement on what counts as the core today
  • →Someone willing to say a project has not earned more money
Tools:A list of the portfolio with revenue, profit and investment for each lineA one-page chart of the three horizons
1

Sort everything into a horizon

Put every business and initiative into horizon one, two or three, by how established it is, not by when it will pay off.

Tips

  • ✓A business with paying customers and growth, but no profit yet, is usually horizon two

Common mistakes

  • ✗Calling something horizon three because it is new to the team, when the market already knows it well
2

Look for the gap

Check that every horizon has something in it, and that horizon two is big enough to matter if horizon one slows.

Tips

  • ✓Ask what the company would be selling in five years if the core stopped growing today

Common mistakes

  • ✗Counting a long list of small pilots as a pipeline
3

Give each horizon an owner, a budget and a yardstick

Fund the three separately, and measure each on what matters at its stage: returns for horizon one, growth for horizon two, learning for horizon three.

Tips

  • ✓Report horizon three on its own line so its losses neither hide in the core nor sink it

Common mistakes

  • ✗Holding a new venture to the core's profit targets in its first year
4

Write a graduation test for every bet

For each horizon-three bet, write down what would have to be true, and by when, for it to move into horizon two, and for each horizon-two business, when it should join the core.

Tips

  • ✓Make the test something you can observe: customers, revenue, a working product

Common mistakes

  • ✗Leaving a bet open-ended, so it is never judged
5

Review the portfolio every year

At each planning cycle, move ventures between horizons, close the options that failed their tests, and decide whether to build or buy what horizon two is missing.

Tips

  • ✓Track what moved between horizons this year; if nothing did, the pipeline is not working

Common mistakes

  • ✗Treating the first sort as permanent

Showing the result

One slide for the board: three columns, one per horizon. In each, the businesses or bets, the money going in this year, the measure each is judged on, and an arrow for anything that moved between horizons since last year. A single line at the bottom answers the question the board cares about: if the core slowed, what would replace it?

On the slide

  • ✓Every horizon, even if one is nearly empty
  • ✓The investment and the measure for each
  • ✓What moved between horizons this year

Left off

  • ✕A blended growth number that hides the horizons
  • ✕Dates on the horizons, which invite the timeline reading
  • ✕Pilots too small to change the answer

Value & Outcomes

Primary benefit

It shows, on one page, whether a company is building its next core while its current one can still pay for it.

Additional benefits

  • ✓Protects new ventures from being judged on the core's measures
  • ✓Makes the missing middle, horizon two, visible
  • ✓Gives the board a way to see growth beyond one blended number

What you'll learn

  • →How to sort a portfolio by maturity rather than by date
  • →Why the three horizons are run at once, not in turn
  • →How to set a test for when a bet moves from one horizon to the next

Typical outcomes

A portfolio chart with every initiative in a horizonSeparate budgets, owners and measures for each horizonA graduation test for each horizon-three bet

Best Practices

📋 Preparation

  • •Start from a full list of businesses and initiatives, with their numbers
  • •Agree the definition of each horizon before sorting
  • •Include the people who run the new ventures, not only the core's managers

🚀 Execution

  • •Sort by how established a business is, not by when it will pay off
  • •Fund and measure each horizon separately
  • •Write a graduation test for every horizon-three bet

🔄 Follow-Up

  • •Review the horizons at every planning cycle
  • •Close the options that failed their tests and say so
  • •Decide each year whether to build or buy what horizon two lacks

💎 Pro Tips

  • •If horizon two is empty, that is the most important line on the page
  • •Start horizon three while the core is strong; it is the only time you can afford it
  • •Ask what has moved between horizons this year
Write the graduation test in advance

For each horizon-three bet, write down what would have to be true, and by when, for it to move into horizon two. It is easier to agree the test before the money is spent than to argue about it after.

Limitations & Pitfalls

⚡

It sorts but does not choose. The framework says you need bets in every horizon; it says nothing about which bets are good.

Mitigation: Pair it with a tool for choosing where to grow, such as the Ansoff matrix, and a way to value uncertain bets, such as real options

⚡

The time axis misleads. Steve Blank argued in 2019 that "the three horizons are not bound by time": a new rival can bring a horizon-three idea to market as fast as an incumbent improves its core.

Mitigation: Classify by maturity, not by date, and watch what rivals are doing in your horizon three

⚡

It says nothing about how markets change. Simon Wardley calls it weak on its own because, without a map of how markets evolve, it is easy to miss the obvious.

Mitigation: Use a Wardley map or scenario planning to see which parts of the market are moving, then place your bets

⚡

It has no stopping rule. A horizon-three bet can run for years without anyone saying whether it is working.

Mitigation: Write a graduation test for every bet and review it every year

What comes before and after

Run first

GE McKinsey matrix →

Score the existing businesses first, so you know how strong horizon one really is.

Ansoff matrix →

Generate the options for growth that will fill horizons two and three.

Run next

Real options →

Value and stage the horizon-three bets, so each one buys the right to invest more later.

OKRs →

Set each horizon's goals with its own measures, so a new venture is not held to the core's targets.

Where this comes fromMehrdad Baghai, Stephen Coley and David White (McKinsey & Company) introduced this (1999).

Sources & further reading

More Management Frameworks

Other frameworks in the Stratrix library.

Strategy Development

Core Competency Analysis

A framework for identifying an organization's unique bundle of skills and technologies that provide access to multiple markets, contribute significantly to customer value, and are difficult for competitors to imitate.

Strategy Development

Strategy Map

A strategy map, from Robert Kaplan and David Norton (2000), draws a strategy as four layers of objectives joined by cause-and-effect arrows. The boxes are easy; the arrows are the strategy, and each one is a bet you can price and own.

Strategy Development

Blue Ocean Strategy

Blue Ocean Strategy creates new demand instead of fighting rivals for existing demand, by eliminating, reducing, raising and creating what an industry competes on. A blue ocean is a move, not a place: it fills, and the grid must be run again.

Strategy Development

Playing to Win

Playing to Win writes a strategy as five linked choices, from a winning aspiration to the management systems that support it. Its two core choices, where to play and how to win, only work when they are made together, not one after the other.

Strategy Development

Business Model Canvas

A visual strategic management tool that describes, designs, challenges, and pivots a business model through nine interconnected building blocks covering value creation, delivery, and capture.

Strategy Development

OODA Loop

The OODA loop is John Boyd's cycle of observe, orient, decide and act, repeated as a feedback loop. Most people remember it as a race to decide faster. Boyd put the weight on orientation: the picture of events in your head, which is where a fast team goes wrong.

Tools to try

More from Stratrix

Ask Stratrix

Three questions a month are free.

Test a decision, free

Set a live decision against the companies that made it before you.

AcquisitionsThe Distance TestDivestituresThe Divestiture TestDiversificationThe Diversification TestMake or buyThe Make-or-Buy Test

Stratrix Team

Every Casebook, three Strategy Packs a seat each year and Ask Stratrix, for teams that face these decisions all year.

$1,999 a year · $499 a seatWhat is included →

New analyses, by email

The defining moves in business, and what they teach, as they publish.

We'll email you a link to confirm. Unsubscribe anytime. Privacy policy.

See Three Horizons in the wild.

Follow Three Horizons across the companies and lenses where it actually shaped the strategy — the decision, the rationale, and how it played out.

Explore the lenses →