Parsio finds the packs that no longer earn their place.
It reads your drawings and specification sheets, rebuilds what each pack should cost, and shows where three to fifteen per cent of the spend is sitting.
Book a demoFrom thousands of drawings to €18M in savings opportunities.
Packaging isn’t a category. It’s a combinatorial problem.
65% of all SAP part numbers. In a global pharma group, 65% of all part numbers are packaging, for about a fifth of a €1.1B scope. All of the complexity, a fraction of the money, none of the resourcing.
130 markets, 130 variants of the same carton. The same carton exists once per language. A comma added to a leaflet, a new legal entity: each one spawns a part number that someone prices by hand. Touch the pack and the artwork re-opens, market by market, for a year or more.
15 real specifications out of fifty. A buyer preparing a tender faces fifty specifications on a case. About fifteen are real. The rest are inherited fields nobody dares delete, and no tool says which is which.
| Field | Value | Source | Effect on price |
|---|---|---|---|
| Style | FEFCO 0201 | drawing, title block | Cost driversets the blank |
| Internal L × W × H | 380 × 260 × 210 mm | drawing, zone 2 | Cost driversets the board area |
| Flute | B, 3.0 mm | drawing, detail A | Cost driversets the board price |
| Board grade | not on drawing | missing | Missingflagged before pricing |
| Edge crush (ECT) | 4.5 kN/m | spec sheet, p. 2 | Paid, not used3.5 stacks the same pallet |
| 2 colours, flexo | artwork | Paid, not usedon a shipper nobody sees | |
| Closure | Tape, 50 mm | drawing, note 3 | Line constraintset by the packing line |
| Pallet code | PAL-7 | inherited, 2011 | Inheritedprices nothing |
| Former supplier ref. | CX-1187 | inherited, 2009 | Inheritedprices nothing |
| Grammage (legacy) | empty | inherited | Inheritedprices nothing |
| Sample approval | 14.03.2009 | inherited | Inheritedprices nothing |
Six plays, one model, and a price computed twice.
- 01
Read what you already have
Drawings, scans from before CAD, artworks, specification exports, ERP lines. Parsio reads them where they are, and shows the fill rate per property before it shows a single saving. That first pass takes a week.
- 02
Build the model, on your rules
One ontology per family, every value traced back to its file and its zone. Your buyers write the comparison rules in plain language, dated and versioned, so the rule outlives the buyer. Market data is priced in and kept current.
- 03
Price twice, then act
A bottom-up should-cost next to a predicted price regressed on your own portfolio. Where the two diverge, an agent opens a case carrying the comparison, the spend at stake and the constraint it would have to clear.
Typical savings levers
| 01 | Collapse duplicate references | See the portfolio by specification, not by supplier. Cluster references within your own tolerance, and price the consolidation before anyone opens a negotiation. |
|---|---|---|
| 02 | Define what the spec should be | Harmonising what exists is one thing. Telling you what your specs should be is another. Feed in the requirements (line constraints, norms, resistance, marketing) and two hundred references fold onto the three families they should have been. |
| 03 | Price every reference twice | A bottom-up should-cost with a confidence score, next to a predicted price regressed across your own portfolio. The gap between the two is the signal no other tool computes. |
| 04 | Compare plants side by side | The same case specified by eight sites, laid out in one view. Which differences are filling-line constraints, and which are just history. |
| 05 | Challenge specs with evidence | A duplicate never becomes an RFQ. It becomes a case for whoever owns the spec, carrying the comparison, the spend at stake, and what the constraint would have to be. |
| 06 | Search before you design | A new pack brief searches the existing portfolio first, then costs the new design. Nothing gets created that already exists. |
250 cartons, read in a week.
Thirty years of accumulation in one plant: 250 references, 12 suppliers, five million units a year, no view of the portfolio by specification. Board thickness was missing on more than half the drawings, and nobody knew which references were the same case under two numbers.
Parsio digitised every drawing, including the ones from the seventies, extracted 31 specifications per reference, and rebuilt a bottom-up cost from board, waste, conversion, print and freight. The first family paid for the engagement.
| Specifications read per reference | 31 |
|---|---|
| References found interchangeable | 1 in 4 |
| Portfolio coverage after one week | 25% to 97% |
| Recovered on the first family | €40K |
Modelled with packaging teams. Not customer results.
Folding cartons and leaflets. 18,000 codes, about 400 real formats. The multiplier is language, not design; Parsio extracts the format behind every code and collapses the portfolio onto its true format set.
Blisters and lidding foil. 2,400 references, four specifications carrying the cost. Parsio reads the gauge off every technical sheet and prices the move from 250 to 200 microns, line-speed constraints included.
Three per cent is the floor, and the first family usually pays for the engagement. Bring your biggest one: within a week you know what it holds.
- Our drawings are old and incomplete. Does that break it?
- No, but it changes what you see first. Parsio shows the fill rate per property and lists what it couldn’t read, before it shows a single saving. On a typical corrugated portfolio, board thickness is missing on more than half the drawings. You’ll see that on screen, not in a footnote.
- Do we have to replace Ariba, Archlet or SAP?
- No. Parsio is the layer before you decide an event is worth running. It produces the costed pack (specs, volumes, baseline, should-cost) and exports it into whatever you already use.
- Who owns the cost model?
- You do. The ontology, the rules and the model for your family are the deliverable. The software is how you use them.
- Our specifications are owned by the plants. How does that work?
- That’s exactly why a duplicate never becomes an RFQ. It becomes a specification case for the site or for central engineering, and you arrive with evidence rather than an opinion.
- Global pharma
- Consumer health & OTC
- Cosmetics
- Industrial manufacturing