56.8%
of Division 1 shares a description verbatim with another part.
Proposal · August 2026
Turn part data into a working asset.
A fixed fee, model first cleanup of The Griff Network's Epicor Kinetic part master, built to make every record searchable, structured, traceable, and ready to plan.
Explore proposal
01 · The challenge
The catalogue is carrying business logic it was never designed to hold.
Your verified replica, measured July 29, 2026, shows a precise pattern: identity, size, quantity, status, and customer context are trapped inside identifiers and free text.
One description covers 606 parts
48
parts have structured dimensions populated.
54/71
purchasing unit codes embed a quantity in the code itself.
7,265
active records across both lots have never transacted or have not transacted since before 2023.
I / 01
Search depends on tribal knowledge
When hundreds of parts share an identical description, quotation, order entry, and purchasing depend on memory. Wrong selection in converting can mean scrap, rerun, and freight.
I / 02
Planning and costing lose their foundation
MRP suggestions, min and max planning, and cost rollups can only be as sound as the part master. Dimensions in part numbers cannot drive the system.
I / 03
Purchasing cannot compare
Units describing a roll of a specific footage do not convert to pounds or feet. Price per pound comparison across suppliers becomes impossible by construction.
I / 04
Reporting reflects catalogue archaeology
Dormant and duplicate records inflate reports and counts. Each new joiner must learn years of naming habits before they can navigate the catalogue safely.
02 · Root cause
This is not a data entry problem. It is a missing data model.
The part number has been doing the database's work. Webisoft will construct the missing middle layer first, sign it off with your experts, then require every transformed record to satisfy it.
What the business handles
- Materials and forms
- Sizes and specifications
- Customers and suppliers
- Units and true conversions
→
Governed data model
- Material to item relationships
- Unique material, form, and size
- Customer and supplier cross references
- Generated descriptions and UOM catalogue
→
Where it becomes operational
- Part master fields and attributes
- Part cross references
- UOM classes and DMT loads
- BAQs, reports, MRP, and costing
Where the model meets a business decision
- Fifty four of seventy one in use purchasing codes carry a quantity inside the unit itself. The footage varies by part, so no shared conversion factor exists.
- Path 1: move quantity to a part level conversion and reduce the code to a true unit.
- Path 2: keep compound codes and accept manual conversion on the 131 parts affected.
- Path 3: split into true units plus a packaging layer.
- Griff selects the purchasing path during days 1 to 3. Unresolved codes ship as a documented decision list, never a silent gap.
One thing the cleanup does not fix
- Collapsing customer specific codes can detach pricing, packaging specifications, or customer price lists keyed to those codes.
- The crosswalk maps every retired code to its replacement, making cutover survivable.
- Reconciliation in quoting and order entry systems sits outside this engagement.
For the largest family, the approximately 600 entry fan out resolves to one material, approximately 270 items, and customer relationships as cross references.
03 · Phase 1 · Cleanup
From raw extract to governed load files in 15 business days.
Both lots are included in one fixed fee deliverable. Confirmed standalone: item master only, fixed fee, no platform components. Webisoft delivers data. You load and cut over.
PhasePhase 1 · Cleanup
Fixed fee$10,000
ScopeBoth lots, priced as one fee, scoped and reported separately
Timeline15 business days
- Division 1: fan out, buried specifications, bent units, and identifier collisions.
- Division 2: dormant catalogue pruning and description normalisation against small clusters.
- Shared approach: one model and pipeline, with findings, scorecards, and audit results reported separately.
Phase 1 stands alone: you receive everything your brief requests whether or not Phase 2 proceeds.
What makes the timeline work
Fifteen days is achievable because the preparation is real. Four things from your side keep it on track.
Extracts and rules on day zero
Full item master extracts for both lots, plus the naming habit parsing rules your analysis produced.
20 hours of product expert time
Concentrated in days 1 to 3 for model signoff and UOM decisions, then distributed across days 9 to 13 for adjudication. Decisions are clustered, so one answer resolves many records.
A nonproduction Epicor company
For the day 7 test load. This is what removes cutover surprises.
DMT licence and a single approver
Griff holds the DMT licence for the load. Scott acts as single point of approval, per Mike Trapp's designation.
01
Days 1 to 3
Define the target state
Draft the material, item, and cross reference model, description grammar, and UOM catalogue from your data. Review against real parts with your experts and settle the UOM path.
Result
Signed target state specification, the standard against which everything is measured.
02
Days 4 to 6
Pilot the ETL
Build parsing rules and the pipeline. Run one difficult family plus a random sample end to end across both lots.
Result
Validated rules, measured auto parse rate, and pilot data reviewed against your parts.
03
Days 7 to 8
Prove the Epicor load path
Generate DMT compatible load files for the pilot and support your test load into a nonproduction company.
Result
A proven load path ahead of scale, with no cutover surprises.
04
Days 9 to 13
Iterate to full coverage
Run the full catalogue nightly. Cluster unresolved records with evidence and a proposed resolution. Encode every decision as a reusable rule.
Result
Exceptions diminish daily. One answer resolves a whole cluster, not one record at a time.
05
Days 14 to 15
Verify and hand over
Sample audit to ANSI Z1.4, AQL 1.0, per lot. Verify 100% of crosswalk and resolved conversions. Run the delta against a fresh extract.
Result
Audited deliverables, UAT signoff, and a repeatable cleanup system your team owns.
04 · Phase 1 · Deliverables
Clean data, complete traceability, and the machinery that produced both.
Cleaned item master
UTF 8 CSV plus an XLSX review copy.
Old to new crosswalk
Complete lineage in CSV and XLSX formats.
DMT load files
CSV output for each approved Epicor template.
Exception log
Filterable XLSX with explicit reason codes.
Signed model and rules
Target specification plus versioned rule catalogue.
ETL pipeline and scorecard
Rerunnable quality checks for duplicates, dimensions, UOM, classification, and dormancy.
Deactivation recommendations
Load ready XLSX per lot, with transaction evidence per record.
Acceptance is measured, not asserted.
- Baseline: your July 29, 2026 figures.
- Each lot sampled to ANSI Z1.4, general inspection level II, AQL 1.0.
- Crosswalk and resolved conversions verified at 100%.
- Results reported per lot. Any lot failing its sample audit is reworked at Webisoft's cost.
- Enablement office hours included throughout, so part creators and stewards work to the new standard at handover.
05 · Phase 2 · Optional
Make the standard permanent where data is created.
Phase 2 turns Phase 1's machinery into a system your team adopts. It is optional and scoped from Phase 1 findings.
PhasePhase 2 · Optional
Indicative fee$15,000
ScopeSystematisation per Phase 1 findings
TimelineScoped at Phase 1 close
Your timeApproximately 15 hours across the engagement
- Path B: selected when auto parse holds above 75% and new part creation is concentrated among a small number of people.
- Path A: selected when creation is distributed across many hands or parse rates run lower.
PATH A
Clean behind automation
Your people work as they do today while the Phase 1 pipeline runs continuously against the nightly replica.
- Score and correct new records by the rule catalogue
- Route genuine exceptions to adjudication
- Turn the scorecard into a live dashboard
PATH B
Constraint at entry
A lightweight translation layer enforces the model at the point where parts are created.
- Prevent nonconforming data from entering
- Show quality score and missing information before save
- Ensure new parts are born conforming
06 · Terms
Fixed fee, fixed scope, no lock in.
Payment$5,000 on signature. $5,000 on UAT signoff.
- Fixed fee against fixed scope. Any change quoted in writing before work proceeds, at $150 per hour.
- Invoices net 30.
- Griff owns all deliverables, transformation rules, and pipeline code on payment. No retained licence, no lock in.
- Either party may terminate on 10 business days notice. Griff pays for work completed and keeps all work product to date.
- Mutual NDA governs. Extracts held in an isolated environment and destroyed on request at completion, with written confirmation.
- Proposal valid 30 days from August 2026.
Not in scope
- Load execution and production cutover, which Griff handles.
- Bills of material, routings, methods, and transaction history.
- Vendor and customer master files.
- ERP evaluation, migration, or upgrade.
- Physical inventory counts.
- Downstream systems affected by consolidation: quoting, pricing, and order entry.
- Epicor, DMT, and third party licence fees.
Counts are taken from your July 29, 2026 replica measurement. If verification finds material variance from that baseline, it triggers a written scope conversation before work proceeds, never a silent overrun.
Ready for a governed catalogue
Webisoft for The Griff Network · August 2026
WEBISOFT
Technology and data transformation
Technology and data transformation
THE GRIFF NETWORK
Part Master Data Cleanup Proposal
Part Master Data Cleanup Proposal
Prepared in response to
Project Scope Brief · July 2026
Project Scope Brief · July 2026