Integrated Business Planning Was Never Integrated

Every engine in your planning stack is doing its job. That is the problem.

IBP was sold as the next generation of S&OP, and the promise was right: one plan across the enterprise, operations and finance in the same language, scenarios instead of a single guess, executives who could finally see the whole business and decide. The ambition has not aged a day.

What arrived was something else.

Not because the engines failed. Driver-based demand sensing genuinely beats naive extrapolation. Multi-echelon inventory optimization positions stock better than any planner with a spreadsheet ever did. Supply optimizers solve material and capacity constrained problems no MRP could solve as effectively. Serious mathematics, working exactly as designed.

Each one optimizes its own domain. Not one of them optimizes the enterprise.

The damage happens at the boundary

Finance mandates a working-capital reduction. Perfectly rational. The inventory engine executes it faithfully: safety stocks come down, schedules move toward just-in-time. What the solve cannot see is clear-to-build on the shop floor, where one missing component derails a production schedule, and the working-capital saving is repaid several times over in expedites, idle lines, and a missed shipment to the customer who mattered most.

Or the supply optimizer re-runs and churns the entire buy plan. Mathematically better against its own objective. It has no view of what it just did to the supplier base: how many firm orders now need rescheduling, how much goodwill that burns with the vendors you will need most in the next shortage, no ability to price the risk to resilience.

Neither engine did anything wrong. They simply could not see the P&L they were spending.

Each engine is correct within its own boundary. The damage happens at the boundary. And the boundary belongs to nobody, except the meetings. Demand review, supply review, reconciliation, executive review: capable people doing enormous work, almost all of it reconciling outputs that were produced separately. Each function optimizes, freezes its answer, and hands it downstream for the next function to optimize around.

And when a genuine trade-off surfaces, nobody in the room can say whether thirty basis points of margin is worth more than eight days of cash, because basis points and days have no exchange rate. So it gets settled by seniority, by precedent, by whoever is most articulate before lunch.

Nobody is failing. The architecture leaves them nowhere else to stand.

What "integrated" turned out to mean

The plan never reached the edges. Your customer already knows what they will pull: their forecast, their backlog, their point of sale, your share in their bill of materials, their MRP procurement plan on what you supply.

Your sales team already knows the deal is slipping a quarter. A pipeline modeled properly, with stage velocity, churn, and conversion by segment, is a probability-weighted demand signal sitting between firm orders and statistical forecast. It reaches planning as anecdote.

Your production network already knows its real OEE. The MES knows. The plan runs on nameplate capacity.

Your critical suppliers already know which commitments will slip.Nobody needs to integrate three hundred supplier systems to fix this. You need supplier collaboration on the few critical buys where a slip puts revenue at risk, and the engine can tell you which those are. For those, it should be a structured exchange, not a buyer chasing a spreadsheet by email.

The plan never priced the order line. Just applied standard cost, a freight allocation, an overhead rate. Nothing for the changeover the rush order caused, the air freight that rescued it, the capital consumed by ninety-day terms, or the tariff that just quietly inverted a sourcing decision. So the plan knows what an order is worth and never what it costs to serve, which means it cannot tell you whether to accept it, what to charge for it, or which plant should make it. Meanwhile price, the single largest lever over demand, is set elsewhere, on another cycle, by another organization. So is promotion. The two biggest optimization levers over the IBP plan sit outside it.

Finance was integrated by reconciliation. Operations builds a plan; finance values it, compares it to budget, and reports the gap; somebody is sent to find volume. Finance gets a report, never a lever. The operating plan and the P&L go on living in different languages, the exact condition IBP was sold to cure.

Uncertainty was answered with three more guesses. A base case, an upside, a downside: hand-built, argued over, then quietly set aside in favor of the base case. Three single futures is not uncertainty. It is three more forecasts. The real distribution is already sitting in your ERP: every promise date against every actual, every supplier slip, every yield excursion. Fit distributions to your own transaction history and you have hundreds of plausible futures instead of three imagined ones. Then model risk events on top in hundreds or even thousands of system generated scenarios. Then solve against them with both kinds of constraint inside: not just the physical ones that make a plan feasible, capacity, lead times, MOQs, but the promised ones that make it acceptable: the working-capital range you gave the street, the margin floor, the service commitment to your largest account. A plan that satisfies every physical constraint and breaks every promise is feasible. It is not acceptable.

And no number could explain itself. Margin is off ninety basis points. Why? Today that question starts a three-week investigation and ends in a narrative. It should be a traversal: which customers, which products, which order lines, which decisions. Hold that thought, because it is not a reporting nicety. It is the reason economic metrics keep dying inside companies.

The diagnosis

Every failure above is the same failure in different clothes. There is no object that can hold demand, supply, price, promotion, terms, cost, tariff, capital, carbon, and uncertainty at once, and solve them together, against one economic number, under both kinds of constraint. So the joint optimization that IBP correctly identified as necessary got handed to people, meeting by meeting, stitching seams the architecture was never designed to close.

We bought silo optimization engines, plural. The enterprise optimization engine, singular, was never built.

The company that ran on one number for twenty-five years

Many companies that adopted EVA — economic value added — in its 1990s heyday let it fade within a few years. My read is that they let it fade for exactly the fifth reason above: EVA arrived as an annual, opaque, finance-owned number that nobody on a shop floor could trace to anything they did on a Tuesday. A metric nobody can traverse is a metric people correctly ignore. They optimize the number they can see, every time.

Godrej, a conglomerate spanning consumer goods, appliances, chemicals and real estate, adopted EVA in 2001 and never let it fade. Not as a year-end metric, but as the currency that trade-offs get argued and settled in: pricing, inventory, capacity, terms. For nearly twenty-five years. The difference was not better math. It was that the number became the company’s language, backed by incentive design that stopped anyone inflating a single year and walking away from it.

That part, making one number a language, no software has ever solved. But the engine that makes the language speakable at every level of the company can now be built. That is what IBP as a process always needed and never had. That's what VYAN is: an integrated, resilient, autonomous engine designed to solve for the enterprise, simultaneously across all the silos.

On July 14 I am hosting a fireside with Dr. Rakesh Sinha, who spent thirty-nine years at Godrej: before the number, through the transition, and running on it for the two decades since. What it took, what nearly broke it, and what made it hold.

Then I will show what VYAN's enterprise optimization engine looks like, built against the five gaps above. Cost-to-serve computed down to the order line, capital charge included. Customer and supplier signal read live instead of chased in spreadsheets. Uncertainty fitted from your own ERP history, not three imagined cases. Physical and promised constraints inside one solve, with the resilient plan sitting among hundreds of futures rather than one. And economics that climb into the financial statements and walk back down to the decision that moved them, so the ninety-basis-point question takes four steps instead of three weeks.

Register here: vyan.ai/webinars/eva-fireside-rakesh-sinha

IBP asked the right questions. It just never built the engine that could answer them.