← The journalMANUFACTURING / FIELD GUIDE

Production planning from real orders and forecasts

How to turn firm sales orders and a sales forecast into a master production schedule, work orders and purchase requests you can defend.

Icelandic coastline with directional camera-motion blur.MANUFACTURING / SUNDIAL
Photography: Claudio Büttler / Unsplash ↗ · Sundial motion treatment
THE PRACTICAL TAKEAWAY

Keep firm orders and forecast apart, explode both through the current bills of materials, check the plan against capacity, and re-run it every week instead of defending last month's version.

Start from demand you can name

Most production plans fail before anyone opens a spreadsheet. They start from a number that nobody can trace: a figure the sales head mentioned in a meeting, last year's volume plus a margin, or whatever the supervisor thinks the line can run. When the plan is wrong, nobody can say which input was wrong, so nobody fixes it.

A plan you can defend starts from two kinds of demand, kept separate. Firm demand is the open sales orders: a customer, an item, a quantity and a promised date. Expected demand is the forecast: what you believe customers will order but have not ordered yet. Both are legitimate inputs. Mixing them into one number is what makes the plan impossible to question.

In Sundial, firm demand is simply the open sales orders and material requests already in the system. Nobody re-keys them into a planning sheet, which removes the most common source of error: a plan built on a copy of the order book that went stale on Tuesday.

Forecasts and the master production schedule

For anything you make ahead of orders, record a sales forecast by item and period. Keep it modest and honest. A forecast is a statement of intent that someone owns, not a promise. The useful discipline is to review it monthly against what was actually ordered and adjust, rather than letting it drift unread.

The master production schedule (MPS) is where firm orders and forecast meet. For each finished item and each period it answers one question: how many will we make, and when? Sundial's Manufacturing module includes sales forecasts and a master production schedule, so the MPS reads open orders and the forecast together and shows you which part of each quantity is backed by a real customer.

Two rules keep an MPS useful. First, consume the forecast with orders: when a real order arrives for a period you forecast, it replaces part of the forecast rather than adding to it, or you will plan double. Second, freeze the near term. Changing what the floor makes this week costs far more than changing what it makes next month, so treat the first week or two as fixed unless something is genuinely urgent.

Exploding the plan into materials

A production plan in Sundial takes the items and quantities you have decided to make and works down through each bill of materials. Sub-assemblies become their own planned work orders. Raw materials become a list of required quantities, which the plan compares with stock on hand, stock already on order and stock already reserved for other work.

The output is two sets of documents: work orders for what you will make, and material requests for what you need to buy or transfer. Because both are linked back to the plan, anyone can follow a purchase request back to the finished item and customer order that caused it. That link is what lets a buyer push back on a request instead of placing it blindly.

This only works if the bills of materials are current. If engineering changed a component last month and the BOM still lists the old one, the plan will order the wrong part with great confidence. Make BOM changes a controlled step, with a new BOM version that becomes the default on a known date, not an edit someone makes on the shop floor.

Capacity is the constraint people forget

Material planning tells you what you need. It says nothing about whether you can make it. A plan that needs twelve days of time on a press you have for five is not a plan; it is a list of future late orders.

Each routing operation in Sundial runs on a workstation with working hours and an hourly cost. When the plan creates work orders, the operations land on those workstations as job cards. Looking at the load per workstation for the coming weeks shows you where the plan is over-committed while there is still time to act: move work to an alternate workstation, add a shift, subcontract an operation, or tell sales the date before the customer finds out.

Be honest about what the capacity view knows. It knows the hours you recorded and the routing times you entered. If routing times are wishful, the capacity view will be wishful too. Compare planned operation time with the actual time logged on job cards every few weeks and correct the routing where the two disagree.

Run the plan every week

The best production plan is boring and frequent. Pick a fixed day. Pull in new orders, refresh the forecast if it has changed, re-run the plan, and review three things with the people who own them: shortages that will stop a work order, workstations that are overloaded, and orders whose promised date is now at risk.

Keep the meeting about exceptions. If the plan is mostly right, nobody needs to read every line; they need the short list of items where demand, material or capacity do not line up. In the redesigned Sundial workspace those exceptions are reports the team can open on a laptop in the meeting or on a phone on the floor.

This is also where AI earns its place in planning, and where we are building carefully. Sundial is designed so that an agent can read the plan, draft the purchase requests and flag the orders at risk, then wait for a planner to approve. The judgment about which customer gets the scarce material stays with a person. The tedious part, checking every line against stock and capacity, should not.