In short: Sales and operations execution owns the weeks inside the time fence at item and location level, while the monthly cycle owns the months outside it. The design problem is decision rights rather than the agenda, so what the weekly forum may reallocate, re-sequence or approve on its own authority has to be written down before it is needed. Freezing an interval buys a large reduction in schedule instability for a modest penalty in plan cost, which Sridharan, Berry and Udayabhanu tested under rolling planning horizons in Management Science in 1987. The failure to watch for is the weekly forum quietly rewriting the monthly plan, and sorting a month of logged decisions into those that executed the plan and those that changed it measures whether that has happened.
The monthly cycle closed on the twenty-fourth. Demand review, supply review, reconciliation, executive sign-off, a published supply plan with capacity committed against it. Then it is the following Tuesday and a supplier confirms sixty percent of a purchase order, a customer moves a promotion forward by two weeks, and a filler goes down for eleven hours on the night shift.
Someone has to decide what happens to the next three weeks. In a lot of businesses that decision gets made by whoever sends the most urgent email, and the only record of it is a reply chain. Four weeks later the monthly meeting reconvenes and spends its first half hour reconstructing what the business actually did, and why it stopped resembling the plan that was agreed.
Sales and operations execution is the process that owns that gap. The hard part of designing one has very little to do with the agenda and almost everything to do with authority.
The horizon split, and where the seam sits
S&OP owns months. It works in monthly buckets, at product family level, over a rolling twelve to eighteen months, and it produces volume and capacity commitments. S&OE owns weeks. It works in weekly or daily buckets, at item and location level, over roughly the next four to eight weeks, and it produces instructions: move this, allocate that, expedite the other, run these orders in this order.
The seam between them is the frozen zone. Inside the time fence the master schedule is closed for renegotiation, because material has been committed, capacity has been booked, and changing the sequence costs more than the change is worth. Outside it the plan is provisional and the monthly cycle is the right forum for changing it. S&OE lives entirely inside the fence, and that containment is what stops it turning into a competing planning process.
Freezing part of the horizon has been studied since MRP systems started producing schedules that changed every time they ran. Steele described the pattern in Production and Inventory Management in 1975 as nervousness: a small change at the top of the bill cascading into a large number of changed order dates underneath it. Blackburn, Kropp and Millen compared damping strategies in Management Science in 1986, and Sridharan, Berry and Udayabhanu tested schedule freezing under rolling planning horizons in Management Science in 1987. The consistent finding across that work is that freezing an interval buys a large reduction in instability for a modest penalty in plan cost, and that the two parameters worth tuning are how far out the fence sits and how often you replan behind it.
Which means the fence is a real decision with a real answer, and it should be set per line or per supply route rather than as one number for the whole business. A line with a two hour changeover running on local raw material can be re-sequenced this week. A line fed by an eight week ocean lead time cannot, whatever gets decided in a meeting.
Decision rights are the whole design problem
Every S&OE process that works has answered one question explicitly: what can the weekly forum change on its own authority, and what must it escalate.
Every one that fails has left that question implicit, and the failure takes one of two shapes. Either the meeting has no authority, in which case it produces a list of issues that get forwarded to the same people who were already dealing with them, and attendance decays inside two months. Or it has unbounded authority, in which case it quietly rewrites the monthly plan every week, the monthly cycle becomes a review of decisions already taken, and the financial commitment stops connecting to anything.
The fix is to write the authority down as bands, before you need them.
Decide in the room. Changes that stay inside the envelope the plan already agreed. Reallocating finished stock between distribution centres. Choosing which orders ship from which location. Re-sequencing production inside a week where family volume and material call-off are unchanged. Approving expedites below a stated cost threshold.
Decide in the room, then log it. Changes that consume something the monthly plan committed without changing its totals. Using overtime already sitting in the flex budget. Pulling a production run forward from week four to week two. Substituting a pack format the customer has agreed to accept. These get made immediately and recorded, so that the cumulative consumption of flexibility is visible before it runs out.
Escalate. Anything that moves the monthly totals, the service commitment, the financial outcome for the period, or a decision another function owns. Cutting an account below its contracted minimum. Spending above the threshold. Concluding that a launch date will move.
The specific thresholds matter less than the fact that they exist and that everyone in the room knows them without having to ask. A meeting that can approve two thousand pounds of airfreight without a phone call, and knows it cannot approve twenty thousand, handles the first and escalates the second inside the same twenty minutes.
The four decisions that belong in the week
S&OE covers a smaller set of decisions than most implementations assume. In practice it is four.
Deployment. Where finished stock goes, now that you know what actually sold last week and what is actually arriving this week. This is the highest frequency decision in the process and the one that most repays automation, because it is mechanical: compare projected cover across locations against target, respect transit times and minimum shipment quantities, move surplus toward shortage. The weekly forum should be reviewing the exceptions the deployment run raised rather than deciding each move by hand.
Allocation under short supply. When there is less available than the order book requires, someone decides who goes short. Deciding that in the moment produces a bad answer with great reliability, because the loudest account wins and the loudest account is rarely the most valuable one. The policy belongs in the room in advance and gets applied when the shortage arrives: fair share against forecast, priority by contract tier, protect the promotion that has already been advertised, protect the new listing a buyer is watching. The mechanics of netting supply against a promised date is a separate discipline with its own logic.
Expedite approval. Airfreight, premium road, an unplanned changeover, a part pallet build. Each one is a decision to spend money to protect service and each one is defensible on its own. The value of routing them through a weekly forum with a cost threshold is that the running total becomes visible while there is still time to do something about it. A business that airfreights eleven times in a quarter usually has one root cause and eleven separate justifications.
Short-term production sequence. Inside the fence the volume is fixed and the order of it is open. Sequencing within the week changes total changeover time, the risk of a late order, and how easily you can absorb a pull-forward. What the weekly forum owns is the trade-off between campaign length and responsiveness, which is a business choice rather than a scheduling detail. Whether the resulting sequence is actually feasible against finite capacity is a question for the scheduling engine, and it has its own treatment.
The weekly meeting should be mechanical
The monthly cycle is deliberative. It exists to surface disagreement, test assumptions, compare scenarios and reach a commitment that several people have to live with. It takes as long as it takes, and it should.
The weekly meeting is a different kind of object. Short, same time every week, same running order, driven by a pack that was published before anyone walked in, and organised around exceptions instead of a review of everything.
A workable shape is thirty minutes and four fixed items. What changed since last week, read off a report rather than recalled from memory. The exceptions the system raised, sorted by value at risk rather than by how recently someone shouted. The decisions required today, each carrying its authority band. The escalations, each with a named owner and a date.
Two disciplines separate that from a status call. The pack has to be produced by the system rather than assembled by hand, because a pack that costs a planner four hours to build will be late, thin or quietly abandoned within a quarter. And every decision gets logged with what was decided, who decided it, and what it was expected to cost or protect. That log is the input to the monthly review, and it turns the sentence "we had a difficult month" into a list of the specific choices that made it difficult.
The limit
S&OE degenerates into a second S&OP whenever the monthly plan is not trusted, and this is the failure worth watching for, because it arrives gradually and nobody announces it.
The mechanism is simple enough. If the agreed plan is systematically wrong, the weekly forum cannot execute against it, so it starts correcting it. Correcting a plan is planning. Within a couple of quarters the weekly forum is where the real numbers get set, it grows in length and attendance, it acquires a pre-meeting of its own, and the monthly cycle carries on as a ritual producing a document nobody opens.
The tell is measurable. Take the decisions logged across a month of weekly meetings and sort them into two piles: those that allocated, sequenced or moved something inside the agreed plan, and those that changed what the plan said the month would be. If the second pile is large, redesigning the weekly agenda will not help, because the cause sits upstream of it.
Upstream means one of three things. The demand plan is biased in a direction the weekly forum has learned to correct for. The supply plan is being published against capacity or lead times that do not hold. Or the fence has been set further out than the business can genuinely commit to. Each is diagnosable on data you already have: bias by stage shows up in value add scoring, supply optimism shows up when you compare published lead times against actual receipt dates, and a fence set too far out shows up as a high rate of authorised changes inside it.
There is also a scale below which the formal process is overhead. A single site with one distribution centre and a two week lead time already has sales and operations execution, and it consists of people talking to each other. The formal version earns its cost when the decisions are spread across sites, functions or time zones, and when an uncoordinated choice costs enough that someone notices.
Take the last four weekly meetings, list every decision that came out of them, and mark each one as either executing the agreed plan or changing it. The ratio tells you inside an hour whether you have an execution layer or a monthly plan that nobody believed.