A planner's Monday: from forecast to purchase order without a spreadsheet

The MRP run finished over the weekend, so on Monday morning the buyer exports its recommendations to Excel, because that is where the real work happens. She sorts by item, re-keys a handful of due dates by hand, and starts a purchase order for oat base. What she does not see is that a firm purchase order, placed three weeks ago for another reason, already covers most of next month. The sheet has no idea, and the legacy tool did not warn her either. By lunch she has doubled up on stock and tied cash in material that will sit until spring.

Nothing in that story is a skills problem. It is what happens when the forecast, the orders, the plan and the purchase order each live in a different place and the spreadsheet is the glue. The planning studio in Cairn is built so the glue is not needed.

Where the plan starts: a forecast that becomes a demand plan

The usual break is at the very first step. A forecast gets built somewhere, pasted into somewhere else, and by the time the planning run reads it, nobody can say which numbers are the forecast and which are what a planner changed afterwards.

In Cairn a forecast run, statistical or uploaded, is approved into its demand plan. That approval is the hand-off. The plan holds the lines a planner can edit, each line keeps its source, and the active plans and your firm orders are what the MRP run reads. If a plan already exists for the same scope, Cairn asks before it overwrites anything, so a Monday approval cannot quietly erase last week's hand edits. The demand planning page covers the forecasting side in more detail.

Firm orders consume their own forecast

Here is where double counting usually creeps in. Suppose the forecast expects 2,000 cases of oat milk in week 43, and a retailer has already placed a firm order for 600 cases for that same week. If the system counts both, the plan believes demand is 2,600 and recommends buying for it. The buyer sees a number that looks big and, with no way to check, buys.

Cairn nets firm orders against the forecast. Each firm order consumes its own bucket first, the span that holds its ship date. If that bucket is spent, it reaches backward and then forward through neighboring buckets, nearest first, following a window ladder you set per item and site.

The nearest weeks work differently. Inside the demand time fence the forecast is set aside altogether and only firm orders drive supply, so a forecast that is already stale cannot plan stock that close in. The planning run, the overlay and the time-phased view all read the same consumption pass, so they cannot disagree.

One table, pegging on every row

The MRP planning studio shows Recommendations, Confirm and issue, and Firmed as one table. The row keeps the same shape as it moves from stage to stage, so a planner is never translating between a recommendation grid and a document list. Every row carries pegging, which answers the question a buyer asks fifty times a day: why is this here? Pegging traces the recommendation to the demand that caused it, a sales order, a forecast bucket, or a work order that needs the component.

Open a row and a time-phased drawer shows the item at that site: projected balance by period, supply and demand against it, and a supply-chain timeline of what arrives when. Capacity sits beside it. The load view stacks the planned work against each work center's calendar, so a make that would land on a closed week or an over-booked line is visible before anyone releases it, rather than discovered on the floor.

Messages that leave a firm purchase order alone

Back to our buyer. The purchase order she missed is marked Firm, and in Cairn that switch has teeth. Reschedule in, reschedule out, reduce and cancel are never raised against a firm purchase order. It nets on its own date, and if the shortage it leaves needs cover, the plan proposes a new order rather than asking anyone to move a commitment you made to a vendor. A surplus held only by firm orders is named for what it is.

One message does still name a firm order: past due, once it has slipped past its date, because that is a fact and not a suggestion. So the Monday surprise inverts. She sees the covered quantity netted in the row, the messages on the orders that can move, and nothing proposing to touch the one that cannot.

Buy spend in the currency you report in

A vendor in one currency, a second source in another, and a catalog price in a third make the total of a release hard to state. Cairn converts buy spend into your base currency at the day's rates, in the worklist, the selection total and the export alike. A line that had to fall back to the catalog price because no rate existed that day is counted and labelled, never hidden. The buyer and finance start from the same figure, and both can see which part of it rests on a catalog price.

A gesture is a request, a release is a transaction

Last, the plumbing that decides whether a tool feels trustworthy. A studio gesture in Cairn is one request. Firming a recommendation, changing a date or releasing a selection goes to the server once, and the answer carries the fresh rows back, so the screen shows what was written rather than what was hoped. Where an assistant helps, Flint drafts for your approval and never invents a quantity or a date, since those come from the plan.

Bulk release is one transaction. Select forty recommendations and release them: either every purchase order and work order exists afterward, or none does. There is no half-released Monday where thirty-one orders went out and nobody can say which nine did not.