Polymr for make-to-order and engineer-to-order shops.

Built for the make to order shape.

A customer PDF becomes a typed BOM with make-vs-buy per line.

The raw input is exactly what an MTO shop gets today: a customer email with a delivery date, an attached PDF that combines a cover letter, geometric drawings, a material callout, a finish callout, and an ISO 3601 o-ring groove spec. The output is a typed BOM where every line carries item code, quantity, unit of measure, make-vs-buy classification, target vendor, and a unit-cost number drawn from the live vendor stream rather than last quarter's spreadsheet.

Quote, customer sign, work order. One conversion strip.

The customer drops a PDF + STEP + spec sheet at 09:12. Polymr derives the BOM, splits make-vs-buy against current shop capacity, generates per-vendor RFQ packets for the buy lines, stubs a routing for the make lines, and renders a draft cost-up against the live markup ladder. External writes (RFQ sends, quote PDF to the customer) wait for the estimator to approve. The four-business-day quote loop becomes a single coffee.

Seven failure modes that make a five-day quote a five-day quote.

Each of these is what an estimating lead names when asked why RFQ-2218 took five business days to turn around. PMR-* / HSG-* / STD-* are internal placeholders; V-### are vendor codes.

• Spec change mid-quote
  Customer changes a dimension on a Tuesday call and the quote rebuild takes four days

  RFQ-2218 arrives with a 220mm housing dimension. On the Tuesday clarification call the customer revises it to 235mm. The original quote BOM was hand-entered from the PDF, so every downstream artifact. make-buy split, vendor RFQ packets to V-244 and V-301, the cost-up roll-up. has to be rebuilt from scratch. The estimator gets back to the customer Friday at 16:30. Two other RFQs in the pipeline lose attention for those four days.

• RFQ packet missing a callout
  An RFQ packet to V-218 leaves out the finish callout and the supplier quotes the wrong base price

  The customer-supplied PDF specifies anodize black per Type II Class 2. The buyer copies the parts list into the RFQ packet but the finish callout sits in a separate sheet of the PDF that does not get copied. V-218 quotes raw aluminum at $11.40 per part. The estimator builds the cost-up against that number. When the actual PO goes out with the finish callout, V-218 revises to $19.60 and the quote is already 9% under-margin at the customer.

• Make-buy decision in the wrong head
  A make-buy decision sits with a buyer who has no view of next week's shop capacity

  HSG-2218-A is borderline make-versus-buy. Make is cheaper per unit if the 3-axis mill at WC-204 has capacity. The buyer has the cost-up but no view of the load on WC-204 for week 31. The shop scheduler has the load but is not in the quote conversation. The default is "buy" because it is the safer answer when nobody has the full picture, and a $480 margin per unit moves to V-301.

• Substitute suggestion never surfaced
  V-244 suggests a substitute that adds three days of lead time but nobody surfaces it to the estimator

  V-244 replies to an RFQ on STD-OR-128 noting that STD-OR-128B is in stock today at the same price, with the same Buna-N spec, but adds three days because it ships from a different DC. The reply lands in the buyer's inbox. The buyer flags it for follow-up and gets pulled into a different job. The original lead-time number stays in the quote, the job ships three days late, and the customer absorbs the slip as "normal" for this vendor.

• Stale markup ladder
  The quote markup ladder lives in a spreadsheet that was last updated before the Q1 material moves

  The markup ladder applies a tiered margin to landed cost (40% under $500, 32% $500–$2K, 28% over $2K). The spreadsheet was last revised in Q4 when aluminum 6061-T6 was $4.10/lb. Aluminum is now $5.20/lb. Every quote that includes machined aluminum has been priced against the stale ladder for the last 11 weeks, and the operations VP only sees the variance at month-end through the gross-margin report.

• Award decision across four threads
  An award decision crosses four emails, two spreadsheets, and a phone call before the job is released

  The estimator drafts the quote. The operations VP approves the markup on email thread one. The customer accepts on email thread two. Engineering signs off on the BOM on email thread three. Purchasing confirms vendor commitments on email thread four. Two spreadsheets carry the cost-up and the make-buy split. A 14-minute call with the customer locks the delivery date. By the time the job is "released," half the inputs are 72 hours stale and the buyer rebuilds at least two of the vendor RFQs.

• Cost-up against a moving baseline
  The cost-up holds today, but vendor spot prices on STD-INS-M6 move 6% in the week before award

  The quote uses last-week's STD-INS-M6 unit price from V-301 at $0.62. By the time the customer accepts, V-301's spot price is $0.66 and the quote-versus-actual variance on a 48-unit BOM is now meaningfully outside the +/-8% band. There is no re-pricing trigger between quote-send and award, so the estimator finds out at PO release that the job is already under-margin.

What this looked like at a custom industrial fabricator.

Where the operational shape is close enough to share playbook.

• Discrete manufacturing

  Discrete is the parent shape. make-to-order is discrete manufacturing where every job is a fresh BOM rather than a repeat. The revision-and-RFQ shape is the same, just amplified one-off.

• Industrial / multi-site

  Multi-plant make-to-order operations need consistent make-buy decisioning across sites with central shop-capacity visibility. otherwise plants compete for the same job or default to buy when one site could make.

• Furniture / configure-to-order

  Custom furniture and casework runs the same configure-to-order loop. a sales-side configurator produces a per-order spec that needs the same one-off BOM derivation and vendor-RFQ packet shape.