Portfolio thinking across a team's quarter

Most quarters are accumulated rather than composed. The planning review ends, a few commitments get made, and then over the following weeks the quarter fills up on its own: the loudest request, the thing a senior engineer wanted to build anyway. By week three you have a quarter, and no one quite decided it.

A roadmap is a list of features, and a portfolio is the same list with the ratio of risk chosen on purpose. The same set of work can be either one, and the only difference is whether you picked the proportions or let them settle.

Setting a good mix is the easy half of this, and holding it is the hard half, which is the part most of us miss. You can compose a balanced quarter in an afternoon of planning, but over the next twelve weeks surprises and urgent work drag it back toward safe, visible delivery. The bet that would have compounded is the first thing to starve, and almost no one decides to starve it. They just look up at quarter-end and find it untouched.

So your job is less to pick the mix than to defend it.

Name the kinds of work before the quarter fills

Start with a number you can borrow. Corporate innovation strategy has a version of this, called 70-20-10: put roughly seventy percent of your investment into the core, the engine that already runs; twenty percent into adjacent bets that extend it; ten percent into transformational work that might not pay off at all. The same idea goes by other names. The three-horizons model splits work into running today's business, building tomorrow's, and seeding the one after that, and run-grow-transform is the same split again. All of them are making one claim: a healthy portfolio holds more than one class of risk at once.

But these frameworks were written for the whole enterprise, in dollars, across fifteen or twenty active projects. Your unit is engineer-weeks and morale. Ten percent of a forty-person org is a real budget, but ten percent of a six-person team is one engineer, part-time, on something that might fail (a hard line to write in a planning doc). Borrow the discipline of naming the classes, but do not import the percentages whole.

You write the lines down before the work arrives to fill them:

• this much for delivery this quarter
• this much for the slower, less certain work
• this much held back for what you cannot yet see

Name them while the page is still blank. Once the work starts arriving, each arrival looks reasonable on its own, and whatever came in first sets the proportions as a side effect.

Sort by time-to-payoff

Those frameworks sort work by how transformational it is, but for a single team the sharper axis is when the work pays back.

I think about this the way I think about a portfolio of stocks. Some positions are short-horizon: they resolve fast and the return is close to certain. They are instant gratification, and there is nothing wrong with that, because they keep you fed and keep the people around you calm. Other positions are long-horizon: they sit for a long time and most of them do nothing, but the few that work compound into something far larger than the safe positions ever could. A good quarter holds both.

None of this is about size. A large feature with a clear spec and a known design is a big piece of work and still a short-horizon, near-certain bet, while a two-week exploration of an approach nobody has tried is small and genuinely risky. What's risky is how long the payback takes, and headcount tells you nothing about that. Sort the work this way and the real shape of your quarter appears: far more certainty than you assumed, with one or two bets quietly carrying all of the upside.

Running that slow bet well, inside an org that asks for results every quarter, is its own problem. For now it is enough to have it in the mix at all, and to know which line it sits on.

Make the safe work earn the risky bet

Here I break from the standard maintenance advice, which says to reserve twenty or thirty percent of capacity for platform work and tech debt so the rot does not eat you. That is true as far as it goes, but it is defensive, and it misses what the safe work is for.

Shipping reliably buys you the cover to try the risky bet. While the bet is mid-flight and showing nothing, the steady stream of finished, visible work tells the team and your leadership that the quarter is alive, and that credibility pays for the bet: as long as the visible work keeps landing, nobody panics about the thing they cannot see yet.

Both extremes fail, and I have hit each. I have run all-safe quarters, where the team ships every item on the list, hits every date, and the quarter somehow still feels flat, as if everyone stood still while they ran, because nothing compounded. And I have watched all-in quarters go the other way, everything riding on one ambitious thing, until it slips and there is nothing finished to point to, no motion, just a long explanation. Both call for the same sequencing: land a sure win early, not because boring work comes first, but because that early win keeps the slow bet alive long enough to matter.

So the mix is set. Naming the classes, sorting by payoff, sequencing the safe work to fund the risky bet: any competent planner can do all of it in an afternoon, and the quarter is not decided there but in the twelve weeks that follow.

Account for the mix over a horizon

You will never see a clean balance inside a single sprint. Some weeks are all delivery, and some weeks the risky bet eats everything. If you try to make every two-week window individually balanced, you get constant reshuffling and a false sense of control.

Balance is something you judge over the whole horizon, the quarter or the half, rather than week by week. Across the quarter the proportions hold, but inside any given week they do not have to.

You have probably heard that focus and portfolios are opposites, that spreading bets is just a polite name for spreading the team thin. That confuses two different levels. A portfolio of bets at the team level says nothing about fragmentation at the person level, and a team can carry a safe line and a risky line at once while every individual on it works on one thing at a time. Portfolio balance is judged across the whole quarter, work-in-progress limits are held week to week, and you can hold both at once without either giving way.

Seat the reserve as a real line

Every quarter, without fail, something arrives that you did not schedule: a piece of urgent work that cannot wait, or a result that turns a side project into the main one. You know it is coming, just not its name yet, which makes it a line in the plan rather than an exception to it.

So reserve for it on purpose. Leave a portion of the quarter unallocated and named as its own line, the way you would name delivery or the risky bet. A team planned to a hundred percent cannot absorb a single surprise without taking the hit out of something it meant to do, and that something is almost always the long bet, because the delivery lines have people outside the team waiting on them and the bet has nobody to complain when its time disappears. How much to hold and how to protect it is its own subject. The size is something you decide in advance rather than a debt you discover later.

Defend the long bet from accidental starvation

This is where quarters come apart, because the urgent is louder than the important every single day, and the long bet is always the quiet one.

Picture the most common ending: around week four, something urgent lands. It is real, and giving it the time is the right call, but it takes longer than anyone guessed (urgent work always does). The team absorbs it, the safe delivery still mostly lands because people protect their own commitments, and the quarter closes with everything urgent shipped. Then someone asks about the compounding bet, the one the whole portfolio was built to make, and the answer is that nobody has touched it since week three. Nobody decided to drop it. The urgent just won, quietly, a day at a time.

There is a clean version of starving a line and a broken version. The clean version is choosing it: deciding on purpose that something gets nothing this quarter so something else can have everything. That deliberate call is its own discipline. The broken version is the one above, where the schedule makes the call because nobody else did.

To prevent this, watch the long line on purpose, and when it is about to lose its time, make that a decision a person says out loud. Keep it if you can, but if you genuinely have to drop it, drop it on purpose: kill it cleanly against the criteria you wrote in advance, having framed it so it fails survivably in the first place. Any of those is a fine outcome. The bad one is the quarter above, where the drop happened and nobody said so.

When the portfolio is the wrong frame

Two boundaries mark where none of this applies.

The first is the betting-table view, most associated with Shape Up: throw out the roadmap, place one bet at a time, make no forward commitments, re-decide everything every few weeks. It is coherent and it works, but it works for a team that is self-funded and owes no one a date. Most of us are not that team. A normal-org leader carries non-negotiable delivery that other teams are already building on, and does not get to re-bet the whole quarter on a whim. The portfolio frame exists for exactly that situation: real commitments you cannot walk away from, and the upside you are still trying to capture alongside them.

The second is the genuinely existential quarter. Sometimes the honest call is one bet, all of it, because the company will not exist in six months if this one thing does not work. There, diversification is not prudence but denial, and a balanced portfolio is the wrong move. Naming that out loud is itself a portfolio decision, though: you looked at the mix and chose to set it to one.

Every other quarter, the mix takes an afternoon to set, and the twelve weeks after it are spent on small calls about which line gives up an hour this week. Most of those calls will not feel like portfolio decisions at the time.