Supply chain: the real cost of silos between planning and execution.

The symptom is easy to recognize. Planners and operational teams share neither the same systems nor the same meetings. The plan is built in one tool, often supplemented with spreadsheets; execution lives in the ERP, the WMS, the TMS. Between the two: exports, rekeying, and above all a mismatch of rhythm: the plan is revised monthly, execution decides by the hour. When reality deviates, the information takes weeks to travel back up to the people who plan. And it rarely travels back intact: everyone dresses it up a little along the way.

The cost is not in the licenses

When a leadership team tries to put a figure on its silos, it looks first at what can be seen: redundant tools, interfaces to maintain, duplicate data entry. Those costs exist. They are nonetheless marginal next to three items nobody measures, because everyone regards them as the normal way the house runs.

The first is the buffer stock each link adds to protect itself from the others. The planner who no longer believes the announced lead times inflates safety stocks. The warehouse, scalded by a past stockout, keeps an unofficial reserve. Procurement, doubting the forecast, orders wide. Each cushion is individually defensible; their sum ties up working capital that no one ever decided to commit. It is the direct financial translation of mistrust between functions.

The second item is emergencies. When plan and execution do not talk to each other, execution quickly learns that the plan is negotiable: it only takes a customer or a salesperson calling loudly enough. Every order handled off-plan sets off its cascade of extra costs, express freight, overtime, other orders pushed back. That is not the worst of it. The worst is the lesson the organization draws: the official channel is the slow channel. From that moment on, urgency becomes the ordinary mode of management and the plan a fiction updated out of politeness.

The third item is more discreet: trade-offs made twice. The S&OP meeting settles a question, say between two markets to serve on constrained capacity. Then, three weeks later, a team leader settles the same question again at the loading dock, with different criteria and no knowledge of the first. The company pays twice for the same trade-off and gets the coherence of neither.

Silos do not show up in the licenses. They show up in the stock each function adds to protect itself from the others.

Why an integrated ERP does not settle the matter

One might think the problem solvable by technology: a single ERP, planning and execution modules on the same database, and the silos disappear. We wish it were true. An integrated system run by siloed governance simply produces better-interfaced silos. If the forecast still belongs to sales, the stock to logistics and the supplier lead time to procurement, each function keeps optimizing its own variable, now on shared data. The system guarantees that everyone reads the same number; it guarantees neither that everyone draws the same decision from it, nor above all that anyone answers for the complete loop, from forecast to delivery. That end-to-end accountability is not installed through configuration.

That is why we are wary of programs that begin with the tool. A new planning engine laid over fractured governance will be bypassed like its predecessor, only with better documentation.

Where to start: small, but as a complete loop

Faced with this, the temptation is the big bang: an S&OP overhaul, a new platform, a general rollout. Our advice is the opposite. Pick one flow, one product or product family, on one site, and rebuild a short loop there between plan and execution: the forecast feeds the plan, the plan drives execution, execution reports its deviations without dressing them up, and the same people look at the same numbers at a tight cadence, weekly rather than monthly. Three measures are enough to judge the loop: stock level, service rate, share of emergencies.

This reduced scope has a virtue the big programs lack: it forces the governance questions to be settled at full scale, where mistakes are still recoverable. Who freezes the plan, and over what horizon? Who has the right to declare an emergency, and who bears its cost? What do we do with a forecast error: hide it, or use it? Once these rules are proven on one flow, they extend to the next ones. Tooling then comes in its rightful place, second: an S/4HANA core carrying plan and execution on the same data, an analytics layer giving both populations the same view of the deviations. In that order, technology consolidates a discipline that already exists. In the reverse order, it waits for it indefinitely.

Our supply chain practice starts there every time: not from the choice of a tool, but from the shortest plan-execution loop that can prove itself.