# How Real-Time ERP Data Cuts Scrap Surprises



> Late ERP postings hide scrap until the order closes. Count every kilogram, trace it to a coil or heat, and set shift alerts that stop drift in time.
> 
> **URL:** https://www.gosmarter.ai/blog/real-time-erp-data-cuts-scrap-surprises/

**Date:** 2026-10-07
**Author:** BlogSmarter AI

**Categories:** blog

**Tags:** artificial-intelligence, data-strategy, digital-transformation, manufacturing, quality

## 


**You need scrap warnings before the shift ends, not an Enterprise Resource Planning (ERP) report on yesterday's losses.** Weights linked to work orders let you spot drift while you can still act.

Late postings leave you chasing paperwork while scrap eats your margin.

[GoSmarter Metals Manager](https://www.gosmarter.ai/features/metals-manager/), built by [Nightingale HQ](https://www.gosmarter.ai/nightingale-hq/), sits alongside your ERP, spreadsheets and email. There is no rip-and-replace. It keeps stock and heat links in one record, so you can trace a scrap spike to its source. [Verify the data timing](https://www.gosmarter.ai/blog/real-time-data-vs-manual-tracking-manufacturing/) before you call anything _real-time_.

Check these first:

-   **[Calculate your scrap rate](https://www.gosmarter.ai/blog/how-to-calculate-scrap-rate/):** 5% scrap against a 3% target is a gap of two percentage points.
-   **Follow the material:** Link each weight to its work order and source coil or heat.
-   **Give someone the alert:** Name an owner who can inspect material or hold output.
-   **Test one order:** Check the weights, links and response before you add another shift.

<u>A dashboard cannot stop scrap. Your team can.</u> Here's how to give them the numbers in time.

## Calculate Scrap Rate and Yield

{{< image src="6ac59086ef6c0279d8508424-1791334983629.jpg" alt="ERP Scrap Rate and Yield: Count Every Kilogram" >}}

### Separate Scrap, Offcuts and Rework

**Scrap rate is scrap weight as a percentage of input weight.** To keep it honest, give each kilogram one status. Accepted output meets spec. Keep rework as work in progress until you accept or scrap it. Retained offcuts remain reusable. Scrap allows neither reuse nor rework.

Sort the categories first. Then calculate each rate from the same linked records. For quick estimates, use the free [scrap rate calculator](https://www.gosmarter.ai/features/free-tools/).

| Metric | Formula | Counting rule |
| --- | --- | --- |
| Scrap rate | Scrap weight ÷ input weight × 100 | Count only material you have classed as scrap |
| Accepted output yield | Accepted output weight ÷ input weight × 100 | Exclude retained offcuts and rework |
| Scrap variance | Actual scrap rate − target scrap rate | Express the difference in percentage points |

Check the balance: **input = accepted output + retained offcuts + scrap + work in progress**.

Keep rework in work in progress until you accept it. Then move its weight once. **Don't count it twice.** Investigate any unexplained gap before you trust the rate.

### Calculate Scrap for a 1,000 kg Run

> Take a 1,000 kg run with 920 kg accepted output, 30 kg retained offcuts and 50 kg scrap. No work in progress remains. Yield is **92%**. Scrap rate is **5%**. Against a 3% target, scrap variance is **+2 percentage points**. The weights reconcile: 920 + 30 + 50 = 1,000 kg. Example only.

Group these metrics by work order, coil, heat number, machine and shift. Use each group's input weight.

For totals, **divide total scrap or accepted output by total input. Do not average percentages.**

> The shift example shows why. Take 10 kg scrap from 100 kg input, plus 50 kg from 1,000 kg. That gives 60 ÷ 1,100 × 100 = **5.45% combined scrap**, not the 7.5% you get by averaging the two rates.

Use the grouped rates to spot which coil, heat or shift starts drifting first.

## Trace Scrap to a Coil or Heat

**A scrap record without a material link leaves the investigation incomplete.** Link each rejection to its source material, process steps and final disposition. Check it against machine, tooling and shift records. Traceability tells you where to investigate. It **does not prove causation**.

When a shift shows scrap drift, trace the affected output to the coil or heat behind the spike.

### Keep Material Links Through Each Process Step

Use the rate signal to follow the material path. Don't guess the cause.

Keep this chain intact: **heat number → coil or handling unit → stock transaction → work order → operation → final disposition**.

When you split or slit material, give each child record its own identifier and keep its parent link. When you combine material, keep every contributing identifier.

Trace rejected output through parent-child links to the source coil or heat. Then follow those links forward to every connected order, accepted output, offcut and scrap record. Record disposition changes against the same links. That keeps the material history intact.

| Traceability field | Source record | Validation check |
| --- | --- | --- |
| Heat number | Mill test certificate (MTC) PDF | Check the extracted identifier against the PDF |
| Coil or handling unit | Goods-in receipt and stock transaction | Confirm the heat link and transaction timestamp |
| Parent-child material links | Split, slit or merge records | Check that child records keep all contributing material identifiers |
| Work order / operation | ERP records and production log | Match material consumption, operator, shift and machine records |
| Outcome | Quality inspection, scrap and yield records | Confirm the weight and trace back to the parent coil or heat |

Once you have the material path, check the [mill certificate or material test report (MTR)](https://www.gosmarter.ai/blog/material-test-report-vs-mill-test-certificate/) that supports it.

### Find Mill Certificates by Heat Number

[MillCert Reader](https://www.gosmarter.ai/features/mill-certificate-reader/) extracts heat numbers from [mill certificates](https://www.gosmarter.ai/hubs/mill-cert-automation/). It connects each certificate to the stock items it covers. You then search by heat number to retrieve the certificate. One heat-number spine feeds MillCert Reader, Metals Manager and Cutting Plans, so you keep one record for every tool. On the stated inputs behind our [pricing](https://www.gosmarter.ai/pricing/), MillCert Reader pays back in about 11 months.

**Check extracted identifiers against the original PDF before you use them.** A readable certificate means little if its heat number doesn't match the material record.

Check the certificate link against the goods-in and stock records. Then review the PDF alongside the operation history. That gives your team an investigation path, not a verdict on the material.

## Set Shift Alerts and Assess Margin Impact

Once traceability identifies the coil or heat, use shift alerts to act **before the order closes**.

GoSmarter does not send scrap-threshold alerts today. It does not take live weights from machines or sensors either. You record or import the weights, and your ERP, a spreadsheet or a messaging channel carries the alert. GoSmarter does keep a scrap log that you can upload and export as CSV, so the records stay linked.

Metals Manager can flag stock, certificates and orders against rules you set, such as low stock. Use that for the material side. Run the scrap thresholds below through your own alert route.

The thresholds are examples. Set yours from your baseline. Give each alert one owner who can inspect material or hold output before losses spread. Route each alert using the linked work order and material ID.

Record the work order, material ID, trigger value, ERP target, last update and deadline. Include the action, escalation contact and acknowledgement log.

### Set Alert Thresholds and Assign Owners

Compare equivalent products and operations. Set minimum input weights or run progress thresholds. That stops alerts firing during setup or on incomplete runs.

| Trigger | Threshold | Minimum progress | Deadline and responsible owner |
| --- | --- | --- | --- |
| Scrap above ERP target | Scrap rate more than 2 percentage points over the ERP target | 200 kg input | 15 minutes; shift supervisor inspects material and holds affected output |
| Repeated defects | 3 consecutive defective units | No minimum | 10 minutes; quality lead corrects the process and quarantines the batch |
| Abnormal coil loss | Unexpected loss of more than 15 kg | 1 coil | 20 minutes; shift supervisor checks machine calibration |
| Input/output mismatch | Variance greater than 2% of input | 500 kg input | 30 minutes; production manager audits input vs output records |
| Missing records or stale feed | No update for more than 10 minutes | No minimum | 5 minutes; IT or operations lead checks the ERP feed or weight import |

A stale feed must show **"data unavailable"**. Escalate to the named contact when an owner misses the acknowledgement deadline. Carry unresolved holds, actions and ownership into the next shift.

When an alert fires, count only the loss above target.

### Calculate Above-Target Scrap Costs

Calculate gross scrap cost as scrap weight × £/kg. For drift, count **only the weight above the ERP allowance**.

> In the 1,000 kg example, 50 kg scrap against a 30 kg target (3%) leaves 20 kg extra. At an assumed £0.80/kg, that is £16.00 before recovery or handling.

| Financial input | Calculation or validation | Responsible owner |
| --- | --- | --- |
| Actual versus target scrap | 50 kg − 30 kg = 20 kg above target | Production lead |
| Above-target material exposure | 20 kg × £0.80/kg (assumed price) = £16.00 | Procurement manager |
| Rework and handling | Add only the costs you can attribute to the extra loss | Operations manager |
| Recovery value | Extra recoverable weight × recovery £/kg | Commercial/finance |
| Incremental margin exposure | Above-target material exposure + additional costs − recovery | Finance |

Keep avoidable costs separate from allocated overhead. Don't count the above-target loss twice. Don't charge rework twice through overlapping rates. Don't deduct recovery you have already credited elsewhere.

Margin impact also depends on current pricing and total output volume. Check whether overtime or rescheduling can recover the lost capacity.

## Test Scrap Tracking on 1 Work Order

Test the ERP link on one work order. Catch scrap drift earlier, before you add machines or shifts.

Choose **one work order you can finish within one shift**. Make sure you can reach the weight records and trace the material through a coil or heat identifier.

Use [Metals Manager](https://www.gosmarter.ai/features/metals-manager/) alongside your ERP records to hold stock and heat links. Export or import records as CSV, or use the [REST API](https://www.gosmarter.ai/features/api-integrations/). Before you start, check data transfer, the refresh interval and alert routing. **Don't call it real-time until those checks pass.**

### Define Success Criteria for the Test

Complete these checks before you release the work order:

-   **Assign an owner:** Name the test lead and whoever will respond to scrap alerts.
-   **Agree the baseline:** Confirm scrap, reusable offcut and rework categories, plus the ERP standard scrap rate.
-   **Record traceable weights:** Log the work order, coil or heat identifier, actual input weight and each output category in kg.
-   **Reconcile the balance:** Check input against good output, scrap and reusable offcuts. Then check that balance against ERP records.
-   **Check the cost:** Calculate the above-target loss and ask finance to verify it.
-   **Test the response:** Trigger an alert with a labelled test event. Confirm receipt, then record acknowledgement and action.

Judge success by **reconciled weights, [traceable material](https://www.gosmarter.ai/hubs/integrated-cert-traceability/) and action before shift end**.

Measure detection delay from event to alert. Measure response time from alert to acknowledgement and intervention.

If the test passes those checks, compare results against the ERP baseline before you extend coverage.

### Review Results Before Adding Machines or Shifts

Review missing events, duplicate postings, false alerts and unexplained balances with production, quality and finance.

Fix the cause, adjust thresholds and confirm ownership before you extend coverage. Repeat the test if you can't explain the balance.

Log the timestamp format, shift start and finish times, and refresh interval. Keep them fixed for the next test.

## Next Steps: Track and Act on Scrap Drift

With ERP data, traceability and alerts in place, tackle scrap drift in this order:

-   Link scrap records to the ERP work order and heat or coil ID.
-   Use one scrap status. Route each alert to **one named owner**.
-   Put a **£ cost on scrap above target** so your team acts on the money you lose.
-   Test one work order first. Scale only when data accuracy, material links and alert response times hold.

If the test holds, carry the same rules to the next machine or shift.

## FAQs

{{< faq question="How fast must ERP updates be to catch scrap drift?" >}}
Fast enough that someone can still act before the order closes. Batch postings and manual entry delay the numbers, so scrap hides until the shift is over.

Set the refresh interval before you trust any alert. Then show "data unavailable" when a feed goes stale. [GoSmarter](https://www.gosmarter.ai/), built by Nightingale HQ, sits alongside your ERP. You keep your system and add the heat and stock links.
{{< /faq >}}

{{< faq question="How do I trace scrap from mixed heats?" >}}
Keep a parent link on every split, slit or merged record. Keep every contributing heat number on combined material. Then trace rejected output back through those links to the source coil or heat.

[MillCert Reader](https://www.gosmarter.ai/features/mill-certificate-reader/) extracts heat numbers from mill certificates and links them to stock. Check each extracted heat number against the PDF before you rely on it.
{{< /faq >}}

{{< faq question="How do I set scrap alerts that reflect margin risk?" >}}
Alert on scrap above target, not total scrap. Price only the extra weight: 20 kg above target at an assumed £0.80/kg is £16.00. Add the rework and handling you can attribute to that extra loss. Then subtract recovery value.

Name one owner per alert and set an acknowledgement deadline. Then ask finance to check the cost.
{{< /faq >}}

{{< faq question="Does this need new hardware, and where is my data held?" >}}
No. You record or import weights, and you keep your existing ERP. Export stock, orders and certificates as CSV, or pull them through the [REST API](https://www.gosmarter.ai/features/api-integrations/).

{{< data-handling topic="hosting" style="short" >}} {{< data-handling topic="training" >}}
{{< /faq >}}