Making the OKR framework the actual operating system: Zulema Quintáns of Noda on focus, flexibility and wasted energy

Most companies adopt the OKR framework and then quietly stop using it by week four. Zulema Quintáns has it running as the actual operating system at Noda, and her account of how is the most practical one I have heard.

Quintáns is COO of Noda, an energy management software platform for commercial buildings. She arrived there by an unusual route: Juilliard-trained professional ballet dancer, then Bain, then American Express, and now a climate tech company. At the time we spoke she had just passed her one year mark in the COO seat.

Focus and flexibility, held at the same time

The through line from dance is not discipline, which would be the obvious answer. It is holding two things at once.

Ballet is a creative profession that also demands extreme focus and structure. Her point is that the artistry arrives in the moments you are not expecting, when there is room for improvisation, and that this only works because the discipline underneath is total.

She sees the COO job the same way. You get cast as the focus person, the one driving OKR discipline and repeatable operating processes. But you also want enough flexibility to anticipate change and stay open to the moments that were not on the plan.

Her rule for which mode to be in is time. When you are moving fast you have to be directive and push people to the outcome. When there is a little room, you give teams the space to reach the conclusion themselves, even if it costs an extra day or an extra week.

Why 70 percent is the target

The number that demoralizes people is the one she defends most clearly.

You are not trying to hit 100 percent. Success is around 70 percent, because that means the goals were stretchy enough to push everyone past the comfort zone.

I pushed on whether that produces complacency, and her answer was a Roger Federer story. In his Dartmouth address he points out that even at the top of his game he wins only just over half the points he plays. The skill is putting the lost ones in the rear view mirror fast enough to focus on the next one.

Her related point is that the misses are information. If you have to change your product OKRs mid-quarter, that tells you that you did not have a firm enough grip on your product roadmap going in. That is a thing you can get better at.

Where OKRs actually fail

Her diagnosis is specific and it is not about the goals.

They fail when the objective is not linked to a project, a roadmap, or somebody's actual job. You said you wanted the thing, but you did not identify the work, you did not prioritize it inside the team, and you did not make it anyone's priority. So it does not happen, and everyone is surprised at the end of the quarter.

Which is why she insists on both directions. Top down for where the strategy places emphasis, bottom up so the goals connect to what people are really working on. In a startup you are constantly toggling between the people closest to the customer and where you are trying to be in twelve or eighteen months.

She is also honest about the pressure on that process. Internal stakeholders want internal tools. Customers want features. Investors want their thing. Her phrase for it is surround sound, and the job is to bring it together well enough to make an honest trade-off rather than just serving the loudest voice in the room.

The mechanics that make it stick

This is the part worth stealing wholesale.

Objectives are set on a six month horizon, twice a year. Key results are developed quarterly, and that part is collaborative: senior leadership proposes some, then teams work to identify their own.

Every OKR gets an owner. They get published. They get talked about at the monthly all hands. She spends real time on how to reinforce them in internal communications and how to celebrate the wins that ladder up to them.

And then the move that changed behavior: they put the OKRs into the performance management system. There is now a record of your goals, visible to you and your manager at review time.

Her summary of why it works is three words. Collaboration, reinforcement, visibility. The performance link is what converts the third one into attention.

What Noda actually does

Worth being precise, because the distinction matters to her. Clean tech is what powers buildings. Climate tech is what optimizes them.

In the US alone, commercial buildings spend around $200 billion a year on energy, and somewhere between 30 and 50 percent of that is simply wasted. Globally, commercial buildings account for roughly 40 percent of greenhouse gas emissions.

Buildings are messy. Aging equipment, messy data, and many hands touching both over decades. Inefficiency is close to inevitable.

There is also a labor problem arriving at the same time as rising electricity prices and heavier compliance reporting. One in three building engineers is retiring, and those roles are not being backfilled.

Noda connects into a building through whatever is available, from the building management system to smart meters to utility bills, then standardizes and cleans that data against an ontology of equipment. Her line on why that comes first is one every operator should keep: AI is only as good as the data you feed it.

The product then works in layers, because customers are at different points. A reporting layer where the operator finds and acts on problems. An analytics and service layer where Noda's team combs the data and drives projects. Then the automated layer, which is the interesting one.

Her image for it is a knob on your desk that controls every thermostat in the building, making small adjustments all day. Cooling earlier when energy is cheaper and cleaner, easing off later. The adjustments are small enough that nobody in the building notices, and the savings are real. Marriott and Hilton are among the customers.

AI pointed at her own operations

Her top AI priority as COO is internal, not customer-facing.

Engineers on her team used to spend hundreds of hours mapping the systems and points in a building. They now use AI to read the names and descriptions and assign those points automatically. Work that took weeks takes hours.

She is prototyping an agent that identifies cost savings from a smaller data set, which compresses onboarding and time to value, and is causing her to rethink the whole choreography of what happens after the sale.

Her broader prediction is the most useful frame in the episode. Historically you had two options: bespoke and expensive, or one size fits most. AI creates a middle segment where tooling delivers customization at a cost that used to be impossible.

But she will not go further than that. Most of her team are engineers who spent years learning building equipment, and she does not think AI replaces that. Which is why the service layer around the customer keeps mattering, and why she expects a correction in which the companies that got the technology right and paired it with a real service proposition are the ones left standing.

The warehouse full of sensors

My favorite question produced the best story.

New in the job, turning over stones, she discovered Noda had a distribution warehouse for sensors in the UK. It puzzled her. Why would a software company be running a logistics supply chain?

It took her back to her first day at Bain. As a new consultant you do a strategy workshop, and she walked in expecting complicated frameworks and graphs. It opened with a simple question instead: what business are we in? The point being that this is where companies make hundred million dollar mistakes.

Noda is not a logistics company. It is a software company. The warehouse was a vestige of a broader IoT product portfolio in European markets that they have since exited, and she is currently finding the sensors a new home.

The 5 things I took away from this conversation

1. Put the OKRs in the performance review system. Everything else on the list is advice I have heard before. This one is mechanical, and it is the one that moved behavior at Noda. A goal that shows up at review time gets attention that an all-hands slide never will.

2. Six month objectives, quarterly key results. Resetting the whole structure every quarter is most of why OKR programs collapse from fatigue. Zulema changes the measurable part quarterly and leaves the direction alone for six months, which is a much lighter thing to sustain.

3. The failure is never the goal, it is the missing link to somebody's job. If a key result is not attached to a specific project and a specific person's priorities, it will not happen. That diagnosis is more useful than any amount of rewriting the objective.

4. Use time as the dial between directive and permissive. When the clock is short, be directive and accept it. When there is room, let the team get there themselves. Framing it as a resource question rather than a leadership-style question is what makes it usable.

5. What business are we in. A warehouse of sensors survived inside a software company because nobody asked. It is the cheapest question on this list and the one most likely to surface something expensive.

FAQ

What is the OKR framework? A goal-setting system pairing an objective, which states the direction, with key results that measure progress toward it. At Noda the objectives are set on a six month horizon and the key results are developed quarterly with the teams, with every OKR assigned a named owner.

Why is 70 percent the target for OKRs? Because goals set to be fully achievable are not ambitious enough to move anyone. Quintáns treats roughly 70 percent as success, on the basis that it means the goals stretched past the comfort zone, and she points to Roger Federer winning just over half his points as the analogy for handling the misses.

Why do OKRs fail? Most often because the key result was never linked to a specific project, roadmap or person's priorities. Quintáns also notes that having to change product OKRs mid-quarter is a signal the product roadmap was not well enough understood before the quarter started.

What is the difference between climate tech and clean tech? Clean tech generally refers to the technologies that generate and supply energy. Climate tech, in the sense Quintáns uses it, covers the technologies that reduce emissions by optimizing how existing infrastructure is operated, which is where an energy management platform for commercial buildings sits.

How much energy do commercial buildings waste? US commercial buildings spend roughly $200 billion a year on energy and somewhere between 30 and 50 percent of that is wasted, according to the figures Quintáns cites. Globally, commercial buildings account for around 40 percent of greenhouse gas emissions.

Also mentioned

• Noda, its independent data layer and equipment ontology
• The Buildings IoT acquisition, and the controls and expertise it added
• Roger Federer's Dartmouth commencement address on losing points and moving on
• Bain, and the first-day question about what business you are actually in
• Marriott and Hilton, among Noda's customers
• The one in three building engineers retiring without a backfill

Listen to the full episode

Zulema Quintáns on Between Two COO's

Between Two COO's is hosted by Michael Koenig. Subscribe on Apple Podcasts, Spotify, or wherever you listen.