Most innovation programs in operations-led companies fail for one boring reason: innovation competes with the day job for the same people and the same hours, and the day job always wins. The quarter has a number. Innovation has a hope. So the workshop happens, ideas land on sticky notes, and three weeks later everyone is back on the P&L.

As COO, your job is not to be the company's idea person. It is to build the system that lets a good idea survive contact with a busy quarter — a funded pipeline, a way to kill weak bets early, and a slice of protected capacity the operating plan cannot raid. Manage innovation the way a fund manages positions, not the way an office manages a suggestion box.

This guide covers the mechanics: structuring the pipeline, gating decisions with real criteria, budgeting so innovation is not the first thing cut, and measuring whether any of it reaches a customer. It is the operating twin of the innovation culture playbook — culture makes people willing to try; management makes the tries add up.

Innovation management vs. innovation culture

These get conflated and they are different jobs. Culture is about whether people feel safe proposing an odd idea and admitting a failed one. Management is about what happens next: does the idea get evaluated, funded, staffed, and either shipped or stopped on a clear timeline?

You can have a warm culture and no output — lots of enthusiasm, but nothing reaches a customer because nobody owns the pipeline. You can also have tight management and a cold culture, where only the safe ideas surface. A COO needs both, but they are separate levers. When output is thin, work out which lever is broken before pulling the other: a strong culture with a broken pipeline needs process, not another all-hands about "being bold."

Structure innovation as a portfolio

A single innovation budget spent on a single big bet is how operations companies lose money quietly. The discipline is to hold a spread of bets at different risk levels, the way you balance a set of investments — mostly safe, some medium, a few wild.

A useful frame is three horizons: improvements to the core business, adjacent extensions, and genuinely new plays. A common starting split is roughly 70 / 20 / 10 by budget — most on the current business, a fifth on adjacent moves, a tenth on far bets that mostly won't pay off but occasionally change the company. Treat those percentages as a starting point to argue with your CEO about, not a law.

The point is not the exact numbers. It is that every idea gets a horizon label and you fund all three deliberately. Without the label, safe H1 efficiency projects starve the H2 and H3 bets every time, because they always look like the responsible choice this quarter. Naming the horizon protects the risky money from the safe money.

Use stage gates so weak ideas die cheaply

The most expensive failure mode is a mediocre project that never dies — it eats a little money and two people's attention for eighteen months and ships nothing. A stage-gate process kills those early, on purpose, and frees the capacity for something better.

An idea passes through a small number of stages, and between each is a gate — a decision point with real criteria and a named owner who can say stop. Ideas advance only by clearing the gate. Killing a project at gate two is a success, not a failure: the system worked and you spent little to learn the bet was weak.

Weak version: every idea gets "keep going" because saying no feels discouraging, so the pipeline clogs with zombie projects. Strong version: each gate has explicit go/kill criteria set before the review, a decision owner, and a visible kill count. If your innovation process has never killed anything, it is not a process, it is a queue.

Protect capacity, not just budget

Money is the easy resource to promise and the useless one to promise alone. You can hand a team a budget and they will still do zero innovation, because their calendar is full of the operating plan. The scarce resource is protected time.

The concrete move is to ring-fence a defined slice of capacity — a fixed percentage of a team's week, or a small team seconded out of line duties — and defend it when the quarter gets tight. The test of whether it is real: when a delivery crisis hits, does the innovation time get raided first? If yes, you do not have protected capacity, you have a sentence in a strategy deck. Make the ring-fence explicit, tell managers it is not a buffer for overflow work, and back them when they hold the line. This is a capacity and prioritization problem — the same discipline you apply to any operations bottleneck.

Fund it so it is not the first thing cut

If innovation is a line inside each operating budget, it dies at the first cost review — a manager under pressure always protects this quarter's delivery over a speculative bet. The structural fix is a separate innovation fund, governed centrally, that operating pressure cannot silently drain.

Set it up like a small internal venture fund: a defined pot, clear criteria to draw from it, and staged release tied to gate progress — a little to investigate, more to prototype, real money only to pilot. Staged funding is itself a control: you are never all-in on an unproven idea, and a project that stalls at a gate simply stops drawing. Keep this fund distinct from your normal operating budget process so the two never compete line by line, and put release decisions at the gates, not with whoever shouts loudest in a planning meeting.

Measure what reaches a customer, not activity

Innovation metrics quietly drift into vanity: ideas submitted, workshops run, people trained. All of that can be high while zero value reaches a customer. Measure the pipeline like a sales funnel — by what comes out the far end, and how fast.

Track three things. Throughput: how many ideas move from stage to stage, and time-to-decision at each gate — a slow gate is a hidden cost, because ideas rot while they wait. Output: what actually shipped, and the share of revenue or cost savings from things launched in the last two or three years. Portfolio health: the spread across horizons and your kill rate, because a process that never kills is not selecting, it is hoarding. Wire these into your regular operating metrics review rather than giving innovation its own reporting no one reads. The revenue-from-new-products number lags by years, so judge this quarter on throughput and use revenue to check the whole machine over the long run.

Look outside before you build inside

Not every idea should be built in-house, and a COO who defaults to "we'll build it" burns years reinventing capabilities that already exist. Partnering, licensing, or acquiring is often faster and cheaper than internal build, especially for bets outside your core competence.

Make the build vs. partner vs. buy call per idea. Build when it is core to your advantage and you have the capability. Partner when someone else already has the technology or market access and a structured partnership gets you there faster. Buy when speed matters more than cost and a mature capability is acquirable. Treating internal build as the only respectable option flatters the org's ego and wastes its calendar. Screen external partners like any dependency — their reliability is now your risk.

Key takeaways

• Innovation fails in operations orgs because it competes with the day job and loses. Manage it as a funded portfolio with protected capacity, not a suggestion box.
• Label every idea by horizon (core / adjacent / new) and fund all three, or safe efficiency projects starve the risky bets.
• Use stage gates with real go/kill criteria and a named owner. Killing weak projects early is the point — a process that never kills is just a queue.
• Ring-fence capacity, not just budget, and hold the line in a crisis. If innovation time is the first thing raided, it was never protected.
• Measure output and throughput — what reached a customer, and how fast — not activity like ideas submitted or workshops run.
• Decide build vs. partner vs. buy per idea; partnering is often faster for bets outside your core.

Frequently asked questions

How much of the budget should go to innovation versus running the business? There is no universal number, and anyone who gives you a precise one is guessing. A common starting frame is roughly 70% of innovation spend on core improvements, 20% on adjacent moves, and 10% on far bets — but that is the split within your innovation budget, not the innovation share of total spend. Set the total by how fast your market is changing: a stable, regulated business can run lean; a fast-moving one cannot. Argue the number explicitly with your CEO rather than letting it default to whatever is left over. What is the single most common reason innovation programs fail? Capacity, not ideas. Most companies have more ideas than they can act on; what they lack is protected time and a process to move ideas through decisions. Innovation gets scheduled around the operating plan, the operating plan always wins, and the program quietly starves. Fix the capacity and the gates before running another ideation workshop — more ideas into a clogged pipeline just makes the clog worse. Should the COO or the CEO own innovation? Both, on different axes. The CEO owns the ambition and the risk appetite — how bold the bets should be and how much the company will spend to find out. The COO owns the machine that delivers against it: the pipeline, the gates, the funding mechanics, and whether ideas actually reach customers. When the two blur, you get either ambition with no delivery or a tidy process producing only safe wins. Keeping the split clear is a specific case of the broader CEO–COO division of labor. How do I stop good ideas from dying between departments? Cross-departmental handoffs are where most ideas die, because no single function owns the whole path from idea to customer. Give each surviving idea one owner who carries it across functions, and make the gates cross-functional so operations, product, and finance all sign off at the same review rather than in sequence. The failure pattern is an idea that passes engineering, waits a month for finance, then dies in an operations capacity fight nobody flagged early. One owner plus a shared gate kills that pattern.