Miller Milling / Ops Strategy
A Four-Phase Playbook / May 2026

From wheat
to truck.
A cost truth
the bag line
cannot hide.

A pack-to-order flour milling business loses 100 to 300 bps of EBITDA to changeover waste, long-tail SKUs, and channel mispricing. Miller Milling can credibly target 300 to 700 bps of EBITDA expansion over 24 to 36 months by executing four phases in order.

300–700
bps EBITDA
expansion target
100,500
cwts/day of
installed capacity
~84%
of mill input
cost is wheat
~62%
avg F&B packaging
OEE today
Briefing

What this is, in one read.

The core problem is a pack-to-order bag line with too many changeovers, in an industry where every point of yield, every percent of OEE, and every avoided changeover matters.

01

The industry is tight. US flour milling at $21.2B revenue (down 4.3% YoY) and ~7.8% net margin. Ardent Mills (the #1) at 9.2% net margin in FY25. There is no slack to give back.

02

Excessive changeovers on a pack-to-order line is the highest-ROI operations problem in bagged flour. Best-in-class CPG packaging lines run 17-minute changeovers; laggards run 50. SMED routinely cuts 30 to 75%.

03

Miller Milling's position is favorable but not durable without action. 4th-largest North American miller, owned by Tokyo's Nisshin Seifun Group since 2012. Saginaw is the 3rd-largest mill in the US and just doubled in capacity.

04

Bulk-vs-pack is the biggest strategic lever, but the latest decision. The net spread is often only $0.50 to $2.00/cwt of margin. An inefficient bag line can make pack less profitable than bulk. The cost truth determines the answer.

05

Wheat backdrop adds urgency. USDA projects 2026/27 HRW production at 515M bu, down 36% YoY, the smallest winter wheat crop since 1965/66. Margin compression risk is real.

The Four Phases

Sequence is non-negotiable. Each phase produces the truth the next one needs.

Phase 1 builds the integrated P&L. Phase 2 layers the SKU-level cost truth. Phase 3 turns that truth into integrated decisions. Phase 4 makes the channel call, only after the math supports it.

Why This Order

Each phase unlocks the next.

Skip Phase 1 and you optimize with allocated cost illusions. Skip Phase 2 and Phase 3 becomes opinion. Skip Phase 4 and you leave the biggest lever on the table, untouched and unmodeled. The order is the value.

01
P&L Truth

Build an honest wheat-to-truck cost stack by plant and channel. Reconcile to corporate.

02
SKU Cost Truth

TDABC layer exposes which SKUs, customers, and changeovers really cost what.

03
Integrated Decisions

Mix, pricing, SKU choice, scheduling. Decisions defensible because the math is.

04
Channel Allocation

Bulk-vs-pack capacity call, made on contribution per constraint-minute, not opinion.

The Opportunity Stack

Twelve levers. Filter by phase, capex, or speed.

Stacked, a well-executed program can credibly target 300 to 700 bps of EBITDA expansion at Miller Milling scale. Tap a filter to focus.

Filter
Where To Start

Pilot at Saginaw. Replicate from there.

It is the largest plant, has the cleanest 60/40 bulk/pack mix, the newest infrastructure, and a co-located Innovation and Technical Center. What proves out at Saginaw replicates to Winchester, Fresno, Los Angeles, and Oakland in weeks, not months.

Saginaw. The Pilot Site

Why here, why first.

  • Scale. 34,000 cwt/day, the 3rd-largest mill in the US after recent expansions.
  • Channel diversity. Roughly 60% bulk and 40% bagged, the cleanest test bed for channel economics.
  • Infrastructure. Newest assets, co-located Innovation and Technical Center, LLumin CMMS+ already deployed.
  • Nisshin R&D access. Premium noodle flour, premix, frozen dough, AMULEIA high-fiber wheat. A moat no US competitor matches.
  • Replication path. What works here translates to the other four plants on a known integration platform.
Implementation Roadmap

Quick wins fund the structural work.

Stand up the P&L in 90 days. Run SMED kaizens immediately. Pay for the rest of the program with the capacity you unlock.

0 to 90 Days
Quick Wins, Self-Funding
  • Integrated wheat-to-truck P&L stood up at Saginaw
  • SMED kaizens on top 3 changeover types, target 50% reduction
  • MOQ enforcement on bottom-decile complexity SKUs
  • Truck-fill discipline and freight optimization
  • Protein and spec giveaway tightening
3 to 9 Months
Cost Truth Installed
  • TDABC model live on Saginaw bag line
  • Complexity-based re-pricing on long-tail SKUs
  • Rhythm-wheel cyclic schedule piloted on highest-volume line
  • Yield uplift and energy programs underway
  • Course-correction triggers measured monthly
9 to 18 Months
Structural Shift
  • SKU rationalization wave executed (target 15 to 25% cut)
  • MIP-driven S&OP institutionalized
  • Hybrid pack-to-order / pack-to-stock policy live
  • Premium-pack portfolio launch (Nisshin R&D)
  • EBITDA bridge tracking to plus 150 bps trailing-12
18 to 36 Months
Channel and Network
  • Bulk vs pack capacity reallocation by plant
  • Customer-concentration de-risking complete
  • Indexed bulk contracts with pass-throughs
  • Network and footprint review
  • Replication across all five plants
Course-Correction Triggers

Four signals: stay on track, intervene, or escalate.

Programs of this kind die quietly when leadership stops measuring. These are the leading indicators.

OEE on Pilot Line
75%
Above 75% by month 9: expand to all lines
Below 70%: escalate engineering and SMED
Changeover Time
−40%
Down 40%+ in 6 months: on track
Down less than 20%: leadership intervention
SKU Count Reduction
15–25%
Down 15 to 25% in 12 mo, rev impact under 2%
Revenue impact above 5%: pause and rerun
EBITDA Expansion
+150 bps
Above 150 bps by month 18: continue
Below 75 bps: restructure governance
Watchouts

The five places this can go wrong.

Most operations strategy programs do not fail on analytics. They fail on sequencing, comms, and the unmodeled customer relationship.

The net spread between channels is often only $0.50 to $2.00/cwt of margin. An inefficient bag line can make pack less profitable than bulk. The Phase 2 cost truth must precede the Phase 4 answer. Anyone making the call before that is guessing.
Tellius and BCG document that consumer and customer intelligence changes the decision on 10 to 20% of SKUs that pure data flags for cutting. A low-volume SKU may be the reason a top-20 distributor takes a full truck. Cut it and you lose the truck, not just the SKU.
The fixed cost does not disappear when you cut the SKU. Run both marginal and fully-allocated views. Decide on marginal contribution per constraint-minute, not on a bookkeeping artifact.
A MIP schedule the floor cannot execute is worthless. Bring plant leadership into the rhythm-wheel design from day one. The schedule has to be operationally sane, not just mathematically optimal.
Bulk channels create concentration risk that pack channels do not. A 30%-of-plant customer loss is operationally catastrophic. Map concentration at every plant, set a ceiling, and write the 24-month plan to get there.