Business case · September 2026

Packaging cost intelligence for SAP-run beverage manufacturers

A read-only layer over SAP that gives the true, current, auditable cost of every empty bottle and package you make — and the model to test a supplier, spec or price change before you commit.

Prepared from public financial disclosures Read-only Deterministic

Sector context, this quarter

~45%¹

of manufacturing COGS — Carlsberg’s own “rule of thumb,” the only such figure any major beverage manufacturer discloses.

~6%²

of beverage COGS is metal cans alone — third-largest expense, behind wages and PET.

The problem

Eight quarters of input-cost pressure

Every major bottler and brewer is running a cost-out programme against it — and every one that quantifies aluminium exposure ties it to the 2025 US Section 232 tariff escalation (25% → 50%).

“For the full year, we expect Midwest Premium inflation to be in excess of $130 million.”
Molson Coors CFO · Q2 2026 call

“If aluminum cans become more expensive, we can put more emphasis on PET bottles.”
Coca-Cola CEO · Q4 2024 call

“We’re pushing every cost lever that is available.”
PepsiCo CFO · Q2 2025 call

Heineken EverGreen
€400–500m/yr
Diageo Accelerate + new plan
~$625m + ~$1bn
Molson Coors, 3-yr
up to $450m
Carlsberg, Britvic synergies
£110m
PepsiCo Productivity Plan
to 2030

The blind spot

Packaging is a top-tier cost — and it is managed nearly blind

Why it’s hard to see: the inputs are scattered across SAP — bill of materials, purchasing prices & info records, POs, receipts, invoices, credit notes, freight & duty conditions, scrap & conversion (production / CO), and FX rates.

Bill of materialsPurchasing prices & info records POs, receipts, invoices, credit notesFreight & duty conditions Scrap & conversion (CO)FX rates

Packaging as a share of COGS

Only one company discloses the number this business case leans on

Carlsberg’s ~45% is the single disclosed figure in the sector. Everywhere else — including in the Diageo case worked out below — the model falls back to an undisclosed 30% default. Hover a row to see exactly what’s known, assumed, or missing for that company.

Disclosed figure Undisclosed — model default (30%) ± uncertainty Qualitative only, no number given

One disclosed data point in the whole sector. Every other bar in this business case that assumes a packaging-of-COGS % is built on the 30% default shown here — treat it as an editable assumption, not a benchmark.

Positioning

Why SAP is the enabler, not the answer

SAP is the system of record. It just doesn’t answer this question on its own — and the S/4HANA programmes now underway make a clean, standardised core to connect to.

Standard SAP cost reports (CO-PC)	What the question actually needs
Standard / rolled-up cost, not current actual landed	Current actual landed cost — freight, duty, scrap, conversion, FX
Needs a costing specialist to run and read	Bottom-up from the live BOM in SAP
Slow; weak at what-ifs	An answer in seconds, self-serve
No vendor comparison on total applicable cost	Every number traced back to its SAP object

The product

What PackCost AI is

Read-only

No capability to change SAP master or transaction data. Never writes back.

Deterministic

Same inputs, same cost to the last decimal, every run. No AI in the arithmetic — the language model only reads the question.

Auditable

For any cost it renders the exact sequence of SAP reads — object, filter, fields, rows.

Answers in seconds, in plain language

From a normalised copy of your SAP data, refreshed on a schedule you set. Folds in freight, duty, credit notes, FX, scrap and conversion — true landed cost, not the price on a purchasing record.

Auditable

“Show me the path”

Reads across these object types to cost one bottle — each names the real SAP object it came from. An auditor can point at every number.

Step	Filter / keys	On SAP S/4HANA
Material master	material = FG-500ML-BOTTLE, plant	API_PRODUCT_SRV / A_Product
BOM header + version valid on the date	product, plant, as-of date	API_BILL_OF_MATERIAL_SRV / MaterialBOM
Purchasing price per vendor	material, as-of date	API_INFORECORD_PROCESS_SRV / A_PurInfoRecdPrcgCndn
Landed-cost elements (trailing 365 d)	material, date window	API_PURCHASEORDER_PROCESS_SRV / A_PurchaseOrderItem
FX rates for foreign-currency prices	—	CO_FNDEI_EXCHANGERATE_RL

Corrected in this pass: the FX-rate row previously cited a service name (API_EXCHANGE_RATE_SRV) that could not be verified against SAP’s API Business Hub. It now points to SAP Market Rate Management (CO_FNDEI_EXCHANGERATE_RL), which is documented.

Proof

One 500 ml PET bottle

This example runs on PackCost’s bundled reference dataset — not a live customer’s SAP data. Currency and values are illustrative.

Component	Eff. qty	Unit	Extended
PET preform 24 g	1.020	1.400000	1.428000
Closure 28 mm	1.005	0.340000	0.341700
BOPP label · USD at 83.00	1.010	0.182600	0.184426
Manufacturing conversion	1	0.071000	0.071000
Empty package cost			₹2.025126

14 reads, one calculation, re-run → identical to six decimal places

Vendor comparison · PET preform		Ranked by total applicable cost
Preform Supplier A · approved	1	₹1.400000 — recommended
Preform Supplier B · unapproved	—	₹1.290000 — no landed basis, not ranked

Evidence panel

CalculationCALC-48745F04CB80

Versionsengine 0.4.0 · policy poc-1

Selected BOMB1 v1 — valid on 2025-08-15

Policiesscrap: bom-then-production · landed: freight + duty + insurance + handling · conversion: configured plant rate · FX: configured

Reproducibilitystatus: complete

Engine deterministic and unit-tested against a fixed reference dataset; the S/4HANA connector is verified end-to-end against representative SAP responses — a customer’s environment is validated in the POC.

The value case

Four value drivers

The model applies these to a company’s disclosed revenue and COGS, every rate an editable Low / Base / High assumption.

Total-landed-cost sourcing

Buy on total applicable cost — unit + landed — not the quoted price, across addressable packaging spend.

Spec & lightweighting validation

Model a gram or spec change for cost impact before the plant trial — while these companies are actively re-cutting specs.

Decision latency removed

Analyst time no longer spent hand-pulling BOM-cost, vendor and what-if answers out of SAP.

Audit & close efficiency

FP&A hours saved on the quarterly packaging-cost bridge and reconciliation back to SAP.

The value, ranked

Read the range, not any single number

Even the cautious case clears its cost well inside a budget cycle, because packaging spend is in the billions. Hover or tap a bar for the assumptions behind it.

Cautious Base Optimistic

Bar height uses a log scale so the cautious and optimistic ends both stay readable — labels show the real dollar value.

Illustrative estimates from public financial data — not a quotation. Carlsberg and Diageo are modelled because their data has been worked through; others are a straightforward addition. Payback is measured against a placeholder PackCost fee pending a pricing decision — it will move once pricing is set.

Worked example

Carlsberg

Disclosed

Revenue DKK 89.1bn · COGS DKK 48.9bn · packaging ≈ 45% of manufacturing COGS

Packaging spend

≈ $3.2bn (COGS in USD × 45%)

Addressable

≈ $1.8bn (primary rigid + closures + labels)

Payback

Base ~1 month; cautious ~11 months — against a placeholder fee; not yet priced.

Every rate on this path is an editable assumption in the model. Change what you don’t believe and the number moves.

The value case

Sustainability targets are spec changes

Each pulled-back target is a live cost / supply trade-off — a BOM change PackCost models before the trial, not an aspiration.

Coca-Cola · Dec 2024

Cut recycled-content and collection goals; removed the refillable target.

Diageo · Aug 2025

Cut recycled content 60% → 50%, citing glass cullet availability; retired the 10% packaging-weight-reduction target.

PepsiCo · May 2025

pep+ reset — recycled-content and reuse targets pulled back.

The offer

The proof of concept

Weeks, not quarters — gated mainly by how fast the extract and the reconciliation report arrive.

STEP 01

Start from a flat extract

CSV / Parquet of the BOM, prices, vendors, freight and FX for one plant and one packaging family. No SAP access required.

STEP 02

Reconcile against one existing cost report

PackCost’s output compared line-by-line to a report you already trust — every difference explained.

STEP 03

Switch to the live SAP connector later

Same mapping format, same results. IT security review runs in parallel, not on the critical path.

Security & IT

A narrow review, by design

Read-only

No capability in the system to write back to SAP.

Secrets externalised

Config files hold only environment-variable names and secret-manager references — never secrets.

Single-tenant POC

A dedicated instance; the normalised copy of your cost data is deleted at the end unless the engagement continues.

No process change

ERP users are not retrained; nothing in SAP changes.

Objections, answered

What buyers push back on first

“We already have SAP costing.”

CO-PC gives standard / rolled-up cost and needs a specialist. PackCost gives current actual landed by component and date, vendor comparison on total applicable cost, and scenarios — self-serve, traced to source. It complements CO-PC.

“Our data isn’t clean enough.”

Missing inputs are flagged partial, never zeroed — so problems surface. The POC reconciliation is, in effect, a data-quality audit with a scorecard.

“Another tool to run.”

Read-only, no process change, no retraining. A faster question interface for the people who already ask these questions.

The ask

A scoped proof of concept

Scope — one plant, one packaging family, e.g. PET bottles at a single site.
From you — one flat extract, and one existing cost report to reconcile against.
From us — a dedicated instance, the build, and the reconciliation, every difference explained.
Timeline — a few weeks to a reconciled result.