Automation ROI Is a Scheduling Bet: Why Capex Cases Underperform by Year Two

Opening Insight

Capex review at a CPG contract manufacturer. The proposal: 4.2 million dollars for a new case-packing cell on Line 3. The ROI case is clean, fourteen productive hours per day, 82 percent OEE, twenty-two month payback. The committee likes the math.

The plant manager asks one question: whose schedule did we model?

The room pauses. The case was built against the scheduling pattern Line 3 ran last year. Line 3 runs fourteen different brand contracts at any given time. Each brand owner sets their own demand pattern, their own SKU variety, their own seasonality. The scheduler does not own the schedule. The contracts do.

Last year's schedule averaged 11.2 productive hours per day, not 14. Next year's schedule depends on which brands renew, which SKUs get added, which seasonality shifts. The 14-hour assumption is not a property of the case packer. It is a property of a schedule that may not exist when the machine ships.

The machine will run as specified. The question is whether the schedule will feed it as modeled.

The pattern is identical across food and CPG. Automation ROI is a scheduling bet. Most mid-market capex cases are betting on a schedule that will never quite exist again.

System Context

Automation capex gets evaluated by finance against utilization assumptions operations made. The utilization assumptions are functions of the schedule. The schedule is controlled by factors outside the capex-approval window: customer mix, product mix, labor availability, upstream constraints.

No one in the capex committee owns the question: is the schedule we modeled the schedule we will actually run? The assumption lives in the appendix, if it lives anywhere.

Mechanism

Automation payback math is written as if utilization is a property of the machine. Utilization is a property of the schedule that feeds the machine. Change the schedule, change the payback.

Run the aggregate math.

A 4 million dollar case packer running 14 productive hours per day at 82 percent OEE delivers about 11.5 effective run-hours per day. At a throughput value of 450 dollars per run-hour, the case packer generates roughly 1.9 million per year in gross throughput value. Subtract labor savings, maintenance, and cost of capital. The 22-month payback pencils.

Change one input: actual run hours are 10.5 instead of 14. At 82 percent OEE that is 8.6 effective hours. The same machine at the same throughput rate generates 1.4 million per year, not 1.9 million. Payback stretches from 22 months to about 34.

The machine is the same in every scenario. The payback is not.

Layer two is the hidden part. Why does the plant run 10.5 hours instead of 14?

Three drivers appear consistently:

The scheduler sequences against customer orders that do not align with case-packer optimization. Format changes cluster when the customer mix rotates weekly. Case packers are fine with long runs of one format and slow on frequent changes. Customer-driven schedules force frequent changes.

Labor availability caps effective run time. Qualified operators for the case packer are a subset of total headcount. On shifts where that subset is short, the machine runs at reduced speeds or idles while a line operator covers the case packer along with two other positions.

Upstream feed gaps create micro-stops that do not register as equipment failure but subtract from effective hours. A 15-minute upstream gap reads as "running" on the machine log and "not producing" on the P&L.

None of these are case-packer problems. All three are schedule-coupling problems the capex case did not model.

System Interaction

Automation sits inside the same Four Decisions as the other three. It inherits whatever dysfunction lives in the other decisions around it.

Scheduling interaction: if the scheduler sequences heavier format changes on Line 3, effective run-hours fall below the capex assumption. The more fragmented the schedule, the worse the automation economics.

Labor interaction: the automation capex usually assumes labor savings at current efficiency. If the labor plan is stale or the product mix has drifted, labor efficiency is below the model's baseline. The modeled savings inflate against a weaker actual baseline.

Packaging sourcing interaction: a procurement change that shifts film or case dimensions changes the case packer's per-unit throughput and changeover profile. If the procurement change lands after the capex approval, the machine runs at a rate different from the one modeled.

The Four Decisions share a schedule assumption. When the schedule moves, all four decisions' approved math moves with it. Automation is where the movement shows up as financial underperformance.

Economic Consequence

Approved on clean math. Running on messy reality.

Capex cases on 20-to-30 month payback typically realize 36-to-48 months when the actual schedule variance exceeds 15 percent of the assumed utilization. The gap does not come from equipment issues. It comes from unmodeled schedule drift.

The components:

• Direct utilization gap: 15 to 25 percent below modeled utilization costs 6 to 14 months of payback
• Labor savings inflation: if labor efficiency is below baseline, the modeled savings do not materialize at the modeled rate
• Opportunity cost: the capex consumed capital that another decision (often a scheduling investment) could have deployed at higher ROI

For a 4 million dollar case packer running at 10.5 hours instead of 14, the three-year gap against the approved case is between 800 thousand and 1.2 million of unrealized throughput value. The P&L reads as a payback that keeps slipping. The root cause reads as equipment or labor.

The harder cost is compounding. The next capex request is sized against the actual (lower) performance of the previous one. The plant approves another capacity investment to cover the gap the first one was supposed to close.

Diagnostic

The test is a fifteen-minute exercise.

Pull the last capex case approved for the line in question. Find the utilization assumption: daily productive hours, weekly changeover count, OEE target.

Pull the last 12 weeks of actual data for the same line.

Compare.

If the gap between assumed and actual utilization is under 10 percent, the automation will approximate the approved payback.

If the gap is 10 to 25 percent, payback will stretch by 40 to 60 percent.

If the gap exceeds 25 percent, the capex should not have been approved at the modeled payback. The schedule was not modeled, only assumed.

Decision Output

• Decision type: Automation capex gate
• Trigger: Any automation capex case over one million dollars on a line with variable product or customer mix
• Action: Require a schedule simulation as part of the capex approval. Size utilization against the 12-week rolling average, not the target. If the simulation shows utilization more than 10 percent below the capex assumption, the case must re-price or defer.
• Tradeoff: Schedule simulation adds 3 to 6 weeks to the approval cycle. It also reveals which capex cases are actually scheduling problems wearing automation clothing.
• Evidence: Mid-market manufacturers that gate automation capex on schedule simulation approve 30 to 50 percent fewer automation projects. The ones they approve hit payback within 10 percent of the modeled case.

Framework Connection

The Monument Was Never the Monument. When automation underperforms, the natural instinct is to blame the machine, the vendor, or the maintenance program. The underperformance almost always lives in the schedule that feeds the machine, not in the machine itself.

This is why Automation is one of the Four Decisions, not a standalone capex category. Automation's economic return is coupled to Scheduling, Labor, and sometimes Packaging sourcing. Treating it as independent is how the Silo Tax accumulates on the capex line.

The fix is not better machine selection. It is a capex gate that requires the schedule to be modeled before the machine is approved. The model does not guarantee the schedule will hold. It does guarantee the committee sees the schedule risk before the capital commits.

Strategic Perspective

Automation ROI is a scheduling bet. Most mid-market manufacturers make the bet unknowingly. The capex case looks like a machine case. It is actually a schedule case.

The highest-leverage question in any automation review is whose schedule you modeled.

For contract manufacturers, the answer is almost never "ours." For food manufacturers running customer-driven mixes, the answer is frequently no. For industrial mid-market with stable long-run customer contracts, the answer might be yes, but only if the schedule has been modeled as a primary variable rather than assumed in the appendix.

Factories that think model the schedule before signing the capex.