Three Horizons

A new business put on the same page as an established one almost always loses, because the columns on that page are revenue forecast and payback period. A business with customers can write a number and point at the reasoning behind it. A business without customers can only write a guess.

The scene played out in Pagedeck's annual planning meeting. One proposal was to test a new area, workflow automation; another was to add enterprise permission management to the existing document tool. The first wrote a guess of roughly $2.3 million in three years. The second wrote a calculation of about $3 million next year. Grace put them side by side and there was nothing to choose. The budget went to the second.

Once is a judgment call. Repeated every year it becomes structure. The state where resources flow only toward what a company is already good at while a new current passes it by is called the innovator's dilemma, and this scale is how that dilemma operates in a meeting room. So the method that came out of it splits the work into boxes and puts a different scale in each box. Keeping the boxes from competing on one page is the whole of what the framework does.

Five steps.

1. Put everything you're doing and everything proposed into one list. Live products, features in progress, ideas — all of it, with no sorting yet.
2. Distribute the items into three boxes. The first holds what already earns revenue, the second holds expansion with real grounds for becoming revenue in two or three years, and the third holds exploration that may or may not work. At Pagedeck the document tool is first, enterprise permissions second, workflow automation third.
3. Write a separate evaluation criterion for each box. The first is judged on profit and customer retention, the second on growth rate and contract pipeline, the third on what got learned this quarter. Attach a revenue target to the third box and that project is being judged by the first box's rules, which returns you to the original problem.
4. Write down the share of resources in advance. The 7-2-1 split so often quoted in resource allocation came from this framework. What matters is less the exact figures than nailing the third box's share to something other than zero — and naming people, not only budget.
5. Split the reporting cadence. Monthly for the first box, quarterly for the second, twice a year for the third. Call an exploration project in every month to report progress and its owner starts manufacturing things to show.

Give the third box graduation conditions too: what finding moves a project to the second box, and what finding kills it. Without them, projects sit in the third box permanently and collect a little budget every year.

This is a framework for a company whose core business still earns. The structure routes part of the first box's profit into the second and third, so a company where the first box is wobbling has no room to split boxes at all.

At a Series A company of around forty people, the use changes shape. There aren't yet three businesses to sort, so the boxes get drawn inside one product at the feature level: features that defend today's revenue, features that open next year's contracts, features nobody knows will sell. For a company still validating an idea it doesn't fit, because the whole company is the third box.

Three limits are worth knowing. First, box assignment itself becomes politics. Landing in the third box means not being judged on revenue, so there's pressure to file a project with second-box characteristics as exploration. Write the rule down in advance — anything with paying customers already doesn't go in the third box — rather than deciding case by case.

Second, people get moved quietly. Budget is an accounting line, so it shows when it flows elsewhere; people simply relocate when something urgent appears. Someone named on a third-box project can spend four days a week on incident response with nothing wrong on paper. That's why many teams count hours per box rather than money per box.

Third, the boxes don't learn from each other. If what the third box discovers never reaches product decisions in the first box, exploration becomes a side hobby. Reading it against a product portfolio — the table of what products the company runs and what role each plays — and writing down movements between boxes each quarter is what turns that link into a process.

"On three horizons this is a third-box project. Drop the revenue target and review it quarterly on what we learned."

Separate its evaluation criteria from the first box's.

"The third-box budget is untouched but there's nobody on it. Both names are still on incident duty."

The resource split is holding only on paper.

"This has paying customers already, so it's second box. File it as exploration and we stop watching growth rate."

Keeping box assignment from going slack.

Pagedeck wrote the boxes out in the following year's plan. The third box got two people, a budget of roughly $115,000, and one question to answer within six months: what percentage of existing customers actually have the kind of repetitive work automation would address.

Six months later the answer was 18 percent, and half of those said they would pay. The automation project moved to the second box and started being judged on growth rate. In the same period another third-box project was shut down, and because the kill condition had been written in advance, that decision took one meeting.

The most common reading is that this is a classification scheme. There are three boxes and items get sorted into them, so it looks like taxonomy. But knowing which box an item sits in changes nothing on its own. What changes things is putting a different scale in each box. Left uncorrected, the company draws a tidy diagram and still ranks all three boxes by one revenue forecast, which leaves the outcome identical to before the framework arrived.

Treating 7-2-1 as the correct ratio is the second misreading. It's quoted widely enough to look like a standard. The right ratio depends on how fast the market moves and how solid the first box is. Some companies need 20 percent in the third box; for others 5 percent is right today. Writing the number down in the meeting and checking each quarter whether it held is worth more than importing a memorized split. The only fixed condition is that keeping even one moonshot — an attempt likely to fail that changes the game if it works — requires the third box's share to be something other than zero.