If you don’t track setup, run, clean time, and schedule adherence, your delivery promises are guesses — and your customers are paying for them.

A bank CEO recently asked a manufacturing owner a simple question: “How long does it actually take you to make a product?” The answer was “roughly.” Setup could be 30 minutes or two hours. Downtime gets written down Thursday for a breakdown on Tuesday. On-time delivery runs 85 to 90 percent, and the misses are called “the cost of doing business.”

That phrase hides real money. Below are the specific costs of not measuring the four numbers every schedule depends on.

1. Setup time: the capacity you can’t see

Untracked setup time is the single biggest reason schedules look feasible on paper and fail on the floor. When setup “could be 30 minutes or two hours,” the planner plugs in an average — and the average is wrong almost every time.

• Sequence-dependent setups are invisible. Going from light to dark color, small to large size, or allergen-free to allergen-containing changes setup dramatically. Without history by from/to product, you can’t sequence jobs to minimize changeovers.
• You overstate capacity. Ten changeovers a week at an extra 45 minutes each is 7.5 hours of lost production nobody planned for.
• You can’t run a SMED or quick-changeover program. You can’t reduce what you don’t measure, and you can’t prove the improvement when you do.
• Lot-size decisions are guesswork. The right run length balances setup cost against inventory carrying cost. Without real setup data, you either run too short (lost capacity) or too long (excess inventory).
• Quotes are wrong. Short-run and custom jobs are priced with setup costs you’ve never actually measured.

2. Run time: standards that drift from reality

Most ERP routings carry run rates that were set years ago, by an engineer, on a good day. If nobody compares them to actuals, every downstream calculation inherits the error.

• Every due date is built on a bad standard. If the routing says 1,000 units an hour and the line really averages 850, a 40-hour job takes 47 hours. Multiply that across a week’s schedule and promises slip by days.
• MRP orders material at the wrong time. Inaccurate run times push or pull material requirements, creating shortages on one line and excess on another.
• Product costing is distorted. Labor and machine-hour costs come from the routing. Wrong run rates mean some products look profitable when they aren’t, and vice versa.
• You can’t spot degradation. A machine slowly losing speed, a material lot that runs poorly, or a crew that needs training all show up first as run-rate variance — if you’re measuring it.
• Capacity planning is fiction. Decisions about overtime, a second shift, or a new machine rest on run rates you haven’t validated.

3. Clean time: the hidden shift nobody scheduled

In food, beverage, pharma, cosmetics, and medical device plants, cleaning can consume more line time than setup. It is also the activity most often left out of the schedule entirely.

• Lines look available when they aren’t. A full CIP or washdown after an allergen run can take a shift. If it isn’t in the schedule, the next job’s start date is wrong before it begins.
• Compliance risk rises. When cleaning isn’t planned, it gets squeezed. Rushed or skipped sanitation leads to failed swabs, holds, recalls, and audit findings.
• You can’t optimize campaign length. The trade-off between running longer campaigns and cleaning less often only works if you know what each clean actually costs in hours.
• Shared resources collide. Cleaning often needs the same tanks, CIP skids, or sanitation crews used by other lines. Untracked, those conflicts surface as surprise delays.
• Validation data is incomplete. Regulated plants need to show cleaning happened as specified. Paper logs filled in later are weak evidence.

4. Schedule adherence: a plan nobody follows is just a wish

Schedule adherence measures whether the floor ran what was planned, in the sequence planned, when planned. Without it, you can’t tell a bad plan from bad execution.

• You never learn why you missed. Was the schedule unrealistic, or did the floor deviate from it? Without adherence data, every miss turns into a debate instead of a fix.
• Supervisors re-sequence by feel. Operators pull the easy jobs, combine similar setups, or chase whoever called last. Some of that is smart — but untracked, it quietly breaks promises made to other customers.
• The schedule loses credibility. When the plan and the floor diverge, people stop trusting the plan. Then they stop following it, and the cycle gets worse.
• Expediting becomes the system. Hot lists, phone calls, and Friday-afternoon overtime replace planning. Expediting is expensive and it always bumps someone else.
• Late-reported events corrupt everything. A four-hour breakdown on Tuesday that is recorded Thursday means two days of decisions were made against a schedule that was already wrong.

How the gaps compound

Each gap alone costs a few hours. Together they multiply: a wrong setup estimate, plus an optimistic run rate, plus an unscheduled clean, plus an unrecorded breakdown turns a one-day cushion into a three-day miss.

The business result shows up in places that rarely get traced back to the schedule:

• Missed delivery dates and the customer trust that goes with them.
• Excess safety stock carried because nobody trusts the plan.
• Overtime and expedited freight to recover from surprises.
• Lost bids to competitors who quote shorter, more reliable lead times.
• Bad capital decisions — buying a machine when the real problem was 20 percent of capacity lost to unmeasured changeovers.

Then there is the people risk. When this knowledge lives in a production manager’s head, it leaves when they retire, quit, or take two weeks off. A plant that “would have to rebuild” its scheduling after one departure doesn’t have a process. It has a person.

What it costs: an average mid-sized plant

Using average figures for a mid-sized plant, untracked time costs about $1.27 million a year — and loses roughly 3,600 hours of capacity, close to a full line’s annual output.

The example assumes 10 production lines, 2 shifts, 5 days a week, 50 weeks a year (about 4,000 hours per line), and a fully burdened cost of $250 per line-hour.

Gap	Average assumption	Hours lost per year	Annual cost
Setup time	10 changeovers per line per week, each running 20 minutes over the planned time	1,667	$416,700
Run time	Routing standards 4% faster than actual on 60 run hours per line per week	1,200	$300,000
Clean time	3 sanitation cleans per line per week, each running 30 minutes over plan	750	$187,500
Schedule adherence	1,500 orders a year at 87% on time = 195 late orders, at $1,500 each in expedited freight and overtime	—	$292,500
Excess inventory	15% extra safety stock on $2 million of inventory because nobody trusts the schedule, at 25% carrying cost	—	$75,000
Total		3,617	$1,271,700

These are averages for illustration; your numbers will differ. The total also leaves out the hardest costs to measure: customers who quietly move to a competitor, quotes priced on setup times nobody has verified, and capital spent on a new machine when the missing capacity was hiding in untracked changeovers.

How ScheduleIQ closes the gaps

ScheduleIQ is an AI execution layer that sits on top of PlanetTogether or Siemens Opcenter. Your APS builds the finite-capacity plan; ScheduleIQ puts it on the floor as a live dispatch board, tracks what actually happens, and warns you before a promise slips.

Gap	What ScheduleIQ does about it
Setup time	Captures actual setup time on every job and compares it to the standard, with changeover analytics by machine and product sequence, so you can fix sequencing and target quick-changeover work.
Run time	Captures actual run time and rate against the routing standard, so drifting standards are flagged and corrected. Queue time and lead-time distribution analytics show where jobs really wait.
Clean time	Captures actual cleaning and sanitation time separately from setup and run, so the true cost of each clean is known and campaigns can be sized to it.
Schedule adherence	Measures plan versus actual — jobs started and finished in the planned window and sequence — and records every resequence made on the board, so you can tell a bad plan from bad execution.
Late-reported events	Machine down, material late, or a hot order is logged and reacted to during the shift, not written up on Thursday.
Missed delivery dates	Predictive on-time risk scores every order from the live schedule, surfacing slips days before they happen.
Knowledge in one person’s head	A daily AI-written rundown of at-risk orders, overloaded machines, and material blockers, plus a Schedule Assistant that answers load, risk, and status questions in plain English.

ScheduleIQ reads from your scheduler and is permission-aware by role: the AI advises, your team decides, and your plan stays in your APS.

What to do about it

You don’t need a perfect data system to start. You need to begin capturing actuals and feeding them back into the plan.

1. Capture start and stop times for setup, run, and clean separately on your bottleneck resources first. A simple tablet entry or MES feed beats end-of-shift paper.
2. Record downtime when it happens, with a reason code, not two days later.
3. Compare actual to standard weekly. Update routings and setup matrices when the variance is consistent.
4. Measure schedule adherence — jobs started and completed in the planned window and sequence — and review misses as a team.
5. Put the real numbers into an Advanced Planning and Scheduling (APS) system so sequence-dependent setups, cleaning rules, and finite capacity are modeled, not guessed.

The bank CEO’s point was simple: certainty comes from visibility. Manufacturers who measure what actually happens can promise a date and keep it.

Watch: the 2-minute version