Every supply chain improvement, from better reorder points to new software or AI, depends on data. If that data is incomplete or inconsistent, the results will be too. The good news is that you do not need perfect data to start. You need to know where the gaps are and how much they matter. This checklist helps you find out.
Why it matters
A planning tool cannot tell that a lead time of zero days is wrong, or that two item numbers are really the same part. It will simply calculate. Data problems therefore turn into bad recommendations, and bad recommendations erode the team’s trust in any new process or tool.
The checklist
1. Item master
- Does each item have a clear, consistent description?
- Are units of measure correct, and are purchase and stock units converted properly?
- Are there duplicate items for the same physical part?
- Are obsolete items marked as inactive?
- Is there a category or commodity code for each item?
2. Lead times
- Does each purchased item have a lead time?
- When was it last updated?
- Does it reflect actual recent deliveries, or a number set years ago?
- Is it in consistent units, such as calendar days?
3. Usage and demand history
- Do you have at least twelve months of usage or sales history?
- Are unusual events, such as a one-time large order, identifiable?
- Is usage recorded when it happens, or in batches at month-end?
4. Inventory records
- How accurate are on-hand quantities? When were they last counted?
- Are locations recorded correctly?
- Are open purchase orders up to date, with old or cancelled orders closed?
5. Supplier data
- Is each purchased item linked to its active supplier or suppliers?
- Are minimum order quantities and pack sizes recorded?
- Do you have country of origin and HTS codes for imported items?
6. Costs
- Are standard or average costs current?
- Are freight and duty captured anywhere, or only the purchase price? See how to calculate landed cost.
Quick tests to run
You can check much of this in an afternoon with an ERP export and a spreadsheet:
- Count items with a blank or zero lead time.
- Sort lead times and look at the extremes for obvious errors.
- Search descriptions for near-duplicates.
- Count open purchase orders with due dates in the past.
- Compare a sample of on-hand records with a physical count.
- Count imported items without a country of origin or HTS code.
Each count gives you a simple measure of how ready a data area is.
Fixing gaps without stopping the business
- Prioritize: fix the data for your most important items first. ABC/XYZ classification helps you choose.
- Fix at the source: correct the ERP, not just the spreadsheet, so the fix lasts.
- Assign owners: make one person responsible for each data area.
- Make it routine: add a short data check to your monthly review.
- Label assumptions: where you must estimate, mark the value as an estimate so it can be improved later.
When data is ready enough
Data is ready enough when the gaps are known, the important items are clean and someone owns ongoing quality. At that point, you can update planning rules, try a planning add-on, test practical AI uses or start a monthly purchased-parts review.
How we help
Data review is the first step in nearly every engagement we run. We check your exports, report the gaps and help you fix the ones that matter most. Free call first, scope agreed before we start.
For the full software guide, see supply chain software for small manufacturers.
A simple data quality scorecard
Turning the quick tests into a short scorecard makes progress visible. For each data area, record one or two counts each month:
- Item master: number of suspected duplicate items; number of active items without a category
- Lead times: number of purchased items with a blank or zero lead time; number not updated in the last year
- Usage: number of months of history available
- Inventory: share of counted items where the record matched the physical count
- Suppliers: number of purchased items without an active supplier; number of imported items without origin or HTS code
- Purchase orders: number of open orders with due dates in the past
Keep the scorecard to one page and review it in your monthly meeting. The counts should fall over time. If one stops improving, look at the process that creates the data, not only the records themselves.
Common causes of bad data
- No owner: fields that belong to everyone are maintained by no one.
- Workarounds: when the ERP is hard to use, people keep the real information in spreadsheets or email.
- Rushed setup: new items are created quickly with placeholder values that are never corrected.
- Old migrations: data moved from a previous system often carries old errors with it.
Fixing these causes keeps the data clean after the first cleanup is done.
Frequently asked questions
How clean does data need to be?
Clean enough that the important items are reliable and the gaps are known. Perfect data is not a realistic goal for most businesses.
Who should own data quality?
Assign owners by area: purchasing for supplier and lead time data, engineering for item and bill of materials data and the warehouse for inventory records.
How long does a data review take?
A first pass with the quick tests above can be done in an afternoon. A deeper review as part of an engagement usually takes days rather than weeks.