Use cases
Your Process Is Different. The Way We Work Is Not.
A molding defect and a wafer-yield problem look nothing alike. Both need the right context, a testable explanation and a response someone owns.
The examples below show how an investigation could start in your process. They are illustrative, not claimed customer results.
Choose a production context.
Injection molding
Short shots across a product family
- Question to investigate
- Which recorded conditions differ between accepted and short-shot runs of one part?
- Evidence to connect
- Inspection results, press settings, part and mold identity, material lots, timing.
- A useful next step
- A supported comparison for the engineer; an earlier-warning test if history allows.
Rejects after a tool change
- Question to investigate
- Did rejects change with the tool or setup, and what else changed?
- Evidence to connect
- Tool and maintenance records, production logs, material context and inspection results.
- A useful next step
- Separate competing explanations and attach the evidence to the agreed engineering review.
Moisture and drying questions
- Question to investigate
- Are recorded drying conditions or material differences associated with reject changes?
- Evidence to connect
- Material lots, dryer records where available, process conditions and quality outcomes.
- A useful next step
- Identify which comparisons the records support and where checks are needed first.
Die casting
Porosity concentrated in one cavity
- Question to investigate
- Does porosity follow a cavity, shot condition, alloy lot or a combination?
- Evidence to connect
- X-ray or inspection results, cavity identity, shot parameters, die temperature, alloy lots.
- A useful next step
- Prioritize hypotheses for casting engineers to validate; no substitute for metallurgical investigation.
Cold shuts following a process change
- Question to investigate
- Which recorded changes preceded the defects, and do acceptable parts share them?
- Evidence to connect
- Furnace and shot logs, material records, die context and defect classifications.
- A useful next step
- Review likely contributors on evidence, not a cause picked from timing alone.
Quality variation between operating periods
- Question to investigate
- After accounting for product mix and conditions, what still differs between periods?
- Evidence to connect
- Aggregated shift context, machine and material records, maintenance events and inspection results.
- A useful next step
- Focus on process conditions and missing context, not conclusions about individual operators.
Electronics assembly
Solder defects on an assembly line
- Question to investigate
- Which board, material and process conditions are associated with the inspection failures?
- Evidence to connect
- Optical inspection results, reflow records, paste lots, stencil changes and board history.
- A useful next step
- Bring the evidence into the quality review; test any warning against history.
Failures that pass on retest
- Question to investigate
- What distinguishes the first failure from the later pass, including fixture changes?
- Evidence to connect
- In-circuit and functional-test records, unit identity, fixture maintenance and component-lot history.
- A useful next step
- Separate product, fixture and test-process hypotheses for engineering validation.
A defect found by the customer
- Question to investigate
- What pre-shipment inspection or test evidence existed, and what did it show?
- Evidence to connect
- Complaint details, unit or lot history, outgoing inspection, test results, dispositions.
- A useful next step
- Route follow-up to the responsible team without assuming the escape was preventable.
Semiconductor
Yield loss at a process step
- Question to investigate
- Where did wafer or unit counts change, and in what process context?
- Evidence to connect
- Lot history, step movement, quality results, tool or chamber records, recipe versions.
- A useful next step
- A connected view of the affected step and the evidence worth reviewing.
Test results that cluster by lot
- Question to investigate
- Do the results share material, equipment or processing history across comparable lots?
- Evidence to connect
- Test results, lot genealogy, equipment history and relevant maintenance records.
- A useful next step
- Narrow the search; a shared lot identifier is not proof of cause.
A process excursion noticed late
- Question to investigate
- Was there a usable signal before the quality outcome was known?
- Evidence to connect
- Inline measurements, process-state records, maintenance history and subsequent quality results.
- A useful next step
- Test an early-warning approach, including false alerts and the time to respond.
These examples illustrate questions and possible evidence. The actual data, method and scope depend on the process and the problem.
Connect the evidence that changes the decision.
Quality and inspection
Defect codes, pass/fail results, measurements, test records and inspection history show the outcome to explain.
Production and material context
Part, lot, order, equipment, material and production timing establish which conditions belong to that outcome.
Process and maintenance
Recorded parameters, tool changes, maintenance and operating context can help identify useful comparisons.
Response and ownership
Alerts, review tasks and recorded decisions keep a finding connected to the people expected to act.
You do not need every possible source, only enough relevant evidence for the question being tested.
Finding the right fit
Are these the only manufacturing processes you work with?
Do we need an advanced analytics team?
Do we need sensors on every machine?
What does a useful starting dataset contain?
Are these examples delivered customer projects?
Different Process? Bring the Question Anyway.
We will help you find a useful starting point and the evidence to test it.
Book a CallOr write to info@datansh.com














