Delphine Rumpus has spent a decade watching organizations see trouble coming and fail to move anyway. As Chief Operating Officer of Verrow Mobility, a fleet operations company that runs demand-forecasting models for transit agencies and logistics networks, she argues that the seductive part of forecasting is also the dangerous part: it promises relief from uncertainty without delivering any control.
Today on The Focal Point with Filiberto De La Cruz, Delphine unpacks what she calls the gap between prediction and readiness — the dashboard turning red long after the real failure happened in the assumptions, the handoffs, the decision rights, and the response capacity nobody had funded. Listen in for how she pressure-tests whether a forecast can actually change anything.
Episode Video
About the Guest
Delphine Rumpus is Chief Operating Officer of Verrow Mobility, where she runs the operations side of a business whose product is prediction. Verrow builds demand, capacity, and cost-exposure models for regional transit authorities and mid-market logistics fleets — which means Delphine spends most of her time on the harder half of the problem: what a customer is actually able to do once the model tells them something they don't want to hear.
Before Verrow she led network planning at Tidewell Freight Group and spent six years in operations strategy at the consultancy Brannock & Reeve, where she worked on capacity crises across shipping, utilities, and public health. The pattern she kept meeting — accurate forecasts arriving at organizations with no authority to act on them — became the basis of her work on response capacity.
Delphine holds a B.A. in economics from Wexbridge College and an M.S. in operations research from the Calloway Institute. She guest lectures on decision latency and organizational readiness, mentors founders through the Halcourt Venture Collective, and is a reluctant but persistent open-water swimmer.
Transcript
Forecasts and response capacity
[Music] This is the Focal Point Podcast, an ongoing series where we deep dive into the meat of messy management. Whether it's technology, [Music] business, government, or entertainment, we chat about our experiences and perspectives in life [Music] and leadership. Join us as we learn about their latest insights and hopefully inspire you to have some of your own. [Music] Here's the scary part. The forecast can be right and the organization can still fail. That's uncomfortable, >> right? Because most leaders think the hard part is getting the prediction. Better model, cleaner dashboard, sharper trend line, more data, and all of that matters. But it does not automatically create control. >> So the danger is not bad forecasting. >> Not only bad forecasting, the deeper danger is believing that prediction equals readiness.
A company sees demand coming, sees costs rising, sees risk building, sees energy use spiking, sees staffing pressure ahead, and still cannot respond fast enough. >> So the dashboard turns red, but nothing operationally changes. >> Exactly. By the time the dashboard turns red, the real failure may already have happened somewhere else. In the assumptions, in the handoffs, in the decision rights, in the lack of response capacity, [Music] >> right? Because the dashboard only shows the smoke, not the wiring inside the walls. >> That's the article central warning. Forecasting is seductive because it promises relief from uncertainty. [Music] But prediction is not control.
A forecast can be mathematically impressive and operationally useless if the organization cannot validate it, interpret it, and act when reality moves differently than expected. [Music] >> That phrase operationally useless is doing a lot of work. It is because the article is not anti-foring. It is not saying don't predict. It is saying don't confuse prediction with preparedness. So this is not really a metrics problem. >> Exactly. It is a humility problem. The forecasting humility problem is the belief that better prediction alone creates better outcomes. The stronger discipline is humbler. Plan carefully, test honestly, keep reality close, and build operations that can adapt when the forecast is wrong. >> Okay, let's unpack that because the first move in the article is interesting. It says forecasting does not eliminate planning.
It makes planning more important. [Music] >> Yes, and that runs against how some organizations behave. They think we have predictive analytics now so we can reduce uncertainty, which is partly true. But if the forecast becomes a replacement for planning, that is where the trouble starts >> because planning is where you expose the assumptions. >> Exactly. A good plan makes thinking visible. [Music] It forces the organization to say what does the model believe? What data does it trust? What range of outcomes matters? What actions should follow from each signal? >> Wait, say more about actions should follow because that feels like the part leaders usually skip. That's the part leaders usually miss. A forecast that says demand will increase by 18% is not [Music] enough. The question is at what threshold do we add capacity?
Decisions behind the forecast
Who makes that decision? What tradeoffs are acceptable? What gets paused? What gets funded? What happens if demand increases by 8% instead or 35%. >> So the forecast is not the plan. The forecast is pressure applied to the plan. That's a useful distinction. The article brings in the idea often attributed through Jim Collins in Good to Great. Planning is priceless, [Music] but plans are useless. Meaning the plan will change, but the act of planning clarifies how the organization thinks before events start moving. >> So planning is not about pretending the future will obey the document. >> No, it is more like rehearsing. A team doesn't run a fire drill because it knows exactly where the fire will start. It runs the drill so people know how to move when something happens. >> And forecasting should work the same way. >> Yes.
Its purpose is not to let leaders stop thinking. Its purpose is to sharpen the questions leaders ask before conditions change. >> Which leads to the next point. Facts are better than confidence. >> Yes, organizations often fall in love with confident predictions. A precise number feels better than a messy range. A dashboard trend feels more authoritative than field knowledge. A model output feels cleaner than a human warning. >> I have seen that the number has a kind of social power. >> Absolutely. Precision can become a custom for uncertainty. A forecast says 73.4%. And suddenly everyone relaxes because it looks scientific. [Music] Even if the inputs are stale, >> even if the inputs are stale, even if edge cases are being averaged away, even if frontline teams are saying this does not match what we are seeing.
That is why the article uses Church Hill's wartime discipline through Collins. Facts are better than dreams. >> So confidence is not the standard. Reality contact is. [Music] >> Exactly. Leaders should ask, "What would make this forecast less flattering but more true? What data is missing? Which assumptions are old? Which observations contradict the model? Which weird cases are being smoothed over? >> That question, less flattering but more true, is powerful >> because many organizations use forecasting as reassurance. They want the forecast to confirm that the current strategy is safe >> instead of using it as a challenge. [Music] >> Right? A forecast that cannot be challenged is not intelligence. It is theater.
[Music] That is a sharp line >> and it matters because when a forecast becomes theater, it actually makes the organization more fragile. People stop looking for disisconfirming evidence. They stop listening to the messy signals. The dashboard becomes a kind of executive bedtime story. >> Everything is fine. The line is smooth. >> Exactly. Until it's not. >> The article then moves into validation, which is where the phrase held out reality comes in. And I love that phrase. In machine learning, you don't just train a model on historical data and then declare victory because it explains the past. You test it on data it hasn't seen. >> Because memorizing the past is not the same as predicting the future. >> Exactly. The model has to meet something it didn't train on. The article expands that idea beyond machine learning.
Test predictions against reality
Organizations need held out reality, too. So what counts as held out reality inside a business? [Music] >> It could be a pilot, a shadow forecast, a customer test, an operational simulation, a post-mortem against actual outcomes. Anything that forces the forecast to encounter the world instead of just the spreadsheet. >> That makes sense because without validation, forecasting becomes storytelling with numbers. >> Yes. And storytelling with numbers can be extremely persuasive, especially when the chart is beautiful. >> The bridge analogy fits here. A bridge can look stable from the road while stress fractures spread underneath. >> Exactly. The visible surface says traffic is moving. But the real question is whether the structure has been inspected under load. Forecast validation is that inspection.
You're not just asking, does this look plausible? You're asking what happens when reality puts weight on it. >> And leaders need to create space for that test before the stakes are too high. >> Yes. Because the worst time to discover that your forecast can't survive reality is when the organization has already committed people, money, and customer promises around it. [Music] >> That brings us to the humbling part. The future refuses ownership. >> Right? Even the best forecast is still a forecast. The world can change faster than the model. The variables that matter most may be the ones the organization can't control. >> This is where humility can sound like pessimism, but it's not >> exactly. Humility isn't saying everything will go wrong. It's saying reality is independent of our confidence.
The article brings in Oliver Burkeman's reminder from 4,000 weeks. You can't entirely depend on a single moment of the future. >> That sentence belongs in every planning room. It really does because organizations often talk as though the future is an asset they can schedule. We will launch here. Demand will rise there. Customers will behave this way. Hiring will keep pace. Vendors will deliver. Costs will stay within this b. >> And then one variable slips. >> One variable slips and the whole operating plan starts to wobble. >> So the lesson is not do not forecast. It is design for movement. Yes, the more uncertain the environment, [Music] the more important it is to build operating systems that can notice, decide, [Music] and adapt quickly.
>> That phrase operating systems is important because people often treat forecasting as an analytics function. [Music] >> Right? But the article is really saying forecasting is also an operating discipline. Who sees variance? Who interprets it? Who is authorized to act? [Music] How fast can the organization change course? If the model says something important and no one can respond, the forecast is basically a weather report for a building with no doors. >> You know the storm is coming, but you cannot move. >> Exactly. >> The article also makes a subtle emotional point. Outcomes are usually less extreme than the story. >> Yes. Forecasts can create overreaction in both directions. A favorable projection can create complacency. An unfavorable one can create panic. >> Both distort judgment. [Music] >> Right?
Use forecasts without theater
When the forecast looks good, leaders may stop asking hard questions. [Music] When it looks bad, they may make frantic changes before they understand the system. >> So, optimism and fear both become bad operators. >> Exactly. The article uses Morgan Howell quoting Scott Galloway, "Nothing is as good or as bad as it seems." Forecasting humility means using that idea to temper both excitement and dread. >> So the forecast is an input, not a verdict. >> Yes, that might be one of the most practical sentences in the whole piece. The forecast is an input, not a verdict. Leaders still need ranges. They need thresholds. They need contingency plans. They need disciplined review >> and that keeps the organization from worshiping the forecast or ignoring it. >> Right? The best organizations do neither. They don't treat forecasts like prophecy.
They also don't dismiss them as guesswork. They use forecasts to improve preparedness. >> Let's make this concrete. Suppose a company forecasts a surge in customer dem. What would a humble forecasting discipline ask? It would ask, "What evidence would tell us the surge is real? What would tell us it's not? What operational bottleneck appears first? Support tickets, fulfillment delays, cloud costs, sales capacity, inventory, onboarding time, >> and who gets to act when that signal appears?" >> Exactly. [Music] Because if every response requires three executive meetings, the forecast may be accurate but still useless. >> That's the response capacity issue. Yes. Prediction without response is fragile. [Music] Response without prediction is reactive. The advantage lives in the relationship between the two. >> Say that again.
>> Prediction without response is fragile. Response without prediction is reactive. The advantage lives in the relationship between [Music] the two. >> That feels like the thesis. >> It is. The forecast clarifies possible futures. The operating system prepares the organization to move when one of them arrives. [Music] >> So for leaders listening, what questions should they take back to their teams? >> First, what assumptions are hidden inside our forecast? >> Not just the data, the beliefs >> exactly what are we assuming about customers, capacity, costs, staffing, timing, vendor behavior, market conditions. Second question. >> What would make this forecast less flattering but more true? >> That one is uncomfortable in a useful way. >> It is because it invites contradiction.
It gives people permission to bring up stale assumptions, missing data, frontline observations, and edge cases. >> Third, [Music] >> where is our heldout reality? Have we tested this forecast against anything it did not already know? a pilot simulation post-mortem shadow run. >> Exactly. Fourth, who notices variance and what happens next? >> Because noticing without authority is just anxiety. >> Yes. If a frontline team sees the forecast breaking, do they have a path to escalate? Can they adjust? Can they slow something down, add capacity, change sequencing, or trigger a decision? And the last question, >> what happens if the model is right but we cannot respond? >> That is the painful one. >> It is because it exposes the gap between intelligence and action.
Build an adaptable operating system
Many organizations are investing heavily in better prediction while leaving decision speed, crossf functional coordination, and operational flexibility untouched. >> So they're installing better smoke detectors but not fixing the wiring. >> Exactly. And then they're surprised when the alarm works and the building still burns. >> That is the final warning, isn't it? Forecasting is not the finish line. >> No, [Music] the goal is not perfect foresight. It's disciplined readiness. >> Better forecasts help. >> They do, but better response capacity matters more >> because the future does not belong to the organization with the cleanest prediction. It belongs to the organization that can stay close to reality, challenge its own confidence, and move before the cost of being wrong becomes too high.
>> So the next time a dashboard turns red, don't just ask why did the forecast miss. >> Ask the harder question, >> which is >> what did our operating system make impossible to see, impossible to say, or impossible to change? Because by the time the dashboard shows the smoke, >> the wiring may already be failing inside the walls. >> And the real discipline is not pretending you can predict every fire. >> It's building an organization that knows how to respond when the alarm goes off. [Music] Thank you for tuning in to this episode of The Focal Point. [Music] Be sure to subscribe on your favorite platform. [Music] I am your host Filiberto De La Cruz. Until next [Music] time, God bless you and yours and everyone. [Music] >> [Music] [Music] [Music]
