Product data quality
Years of vendor imports and manual overrides leave AV integrators with duplicate parts, inconsistent manufacturer names, unflagged discontinued items, and wrong item types. The effects often show up during inventory counts, margin reporting, or a merger of two Q360 instances.
A few bad entries per import can add up to tens of thousands of records over a decade, slowing quotes and distorting reports.
Cleanup checks
We check the Q360 masterno database for five things: duplicates, part-number collisions, inconsistent manufacturer names, missing end-of-life flags, and incorrect item types.
1. Deduplication
The same part can end up with multiple Q360 masternos after repeat imports with different formatting or when vendors list it under different numbers.
- Symbol variations: hyphens, underscores, slashes, spaces all get normalized before comparison
- Manufacturer dictionary lookups: we resolve "Crestron", "Crestron Electronics", "CRES", and "Crestron Inc" to a single canonical value before matching
- Vendor part numbers: cross-reference vendor-supplied part numbers to catch cases where different customers use different internal numbers for the same product
- UPC comparison: use shared UPCs, where available, to identify the same part across records
2. Collision prevention
When two manufacturers use the same part number, imports can overwrite a part or make quoting ambiguous. We add a 3-letter manufacturer prefix to each part, such as "CRE-MC-125" and "EXT-MC-125", so both stay distinct.
3. Manufacturer normalization
Customer-facing documents and reports need consistent manufacturer names. We maintain a master dictionary of canonical manufacturer spellings (with customer-specific preferences where they matter: some customers want "Samsung Electronics America," others just "Samsung"). Every import is normalized against the dictionary on the way in.
4. End-of-life management
Discontinued items shouldn't just disappear from the master, as some customers still have stock on the shelf or open projects using them. We flag EOL items in a dedicated tracking field that carries:
- When the manufacturer announced EOL
- Current stock level across all warehouses
- Suggested replacement part (if the manufacturer published one)
- Date after which the item should be automatically deactivated (usually when stock hits zero)
5. Item types (A / Q / K)
Q360 uses item types to distinguish serialized equipment (A-type), non-serialized stocked goods (Q-type), and kit/bundle items (K-type). Bulk imports frequently mis-type items because the vendor didn't mark them: a rackmount switch imported as Q-type instead of A-type creates problems at project delivery when the install tech needs the serial-number field that only exists on A-type items.
We use a workflow bucket concept during bulk imports: items go into a pending bucket first, get auto-classified by rules, and then get human-reviewed before committing to the master. Catches 99% of typing errors before they hit production data.
Cleanup actions
Implementation
- Data pull. We extract the current product master (usually hundreds of thousands of rows) along with the transaction history (POs, quotes, invoices) that references each item.
- Rules run. Our cleanup engine applies deduplication, collision detection, manufacturer normalization, EOL flagging, and item-type validation. Produces a report of every proposed change.
- Human review. Your Q360 admin reviews the proposed changes, usually focused on high-impact items and ambiguous cases, and approves, edits, or rejects each one.
- Staged execution. We apply approved changes against a Q360 staging copy first, validate no transactions broke, then replay against production during a quiet window.
- Post-cleanup report. You get a full log of what was merged, renamed, or retyped, with before/after counts and a list of items that couldn't be auto-resolved (need manual decisions).
FAQ
A 50,000-item master with moderate duplication can take 2–3 weeks. A 500,000-item master after an acquisition can take 2–3 months, largely because edge cases need human review.
No. Every proposed change is validated against open transactions before execution. If a merge would break an active quote, we either defer it until the quote closes or route it through a manual-review queue.
Both. Most clients start with a one-time cleanup, then switch to ongoing maintenance: we process incoming vendor pricelists through the same normalization pipeline so new imports don't re-introduce the problems we just fixed.