“We've tried everything. What should we change?” is a difficult question to answer, even for someone who knows the process. The missing information is usually more specific than the phrase suggests: which feature, which samples, which changes, and which result still needs explaining?
A useful injection molding troubleshooting request gives the next person enough evidence to choose a next investigation. It does not need to include every screen on the press. It needs a clear observation, a traceable comparison, the work already tried, and one explicit request.
Start with what you need from the reader
Choose the immediate decision. Do you need help classifying the defect, finding a missing measurement, reviewing a tooling concern, or planning a controlled comparison? “Please advise” leaves the recipient to infer all of that.
Make the requested outcome concrete: “Which evidence would distinguish a filling issue from a local tooling issue at feature E-3?” is a question someone can examine. “What pressure should I use?” may be unanswerable without an established process, equipment definition, material guidance, and authority to change it.
The American Society for Quality's problem-solving guidance separates defining a problem from selecting and implementing a solution. Apply that separation to the opening sentence: describe what is happening before recommending what another person should change.
Write the observation in ordinary language when the defect name is uncertain. “An incomplete rib tip on cavity 2” is useful. A confident but unverified label such as “vent blockage” can steer the whole conversation toward an assumed cause.
The first message needs enough context to locate the evidence
Put a concise summary in the message and use attachments or approved record links for detail. The recipient should be able to identify the samples and understand the outstanding decision without opening a dozen unnamed files.
| Part of the request | Include | Avoid |
|---|---|---|
| Observed problem | Feature, location, cavity, time window, affected and inspected counts, inspection reference. | A cause written as an observation. |
| Job identity | Part/revision, press and mold IDs, exact material grade and lot or approved references. | Only a resin-family name or an unexplained local nickname. |
| Comparison | Last relevant accepted record, matching or differing sample conditions, and known gaps. | “It was fine before” without evidence of what was checked. |
| Work already tried | Trial IDs, exact changes in their records, measured results, current authorized state. | “Tried everything” or a list of settings without outcomes. |
| Requested help | One specific review question, needed response time, and responsible contact. | An implied request to approve production from a photograph. |
Choose attachments that answer different questions
An overview image locates the feature on the part. A detail image shows the observed mark under the recorded conditions. A sample list connects the image to cavity and time. A trial log explains what changed. Those attachments complement each other.
Name them by record and purpose: “T-31 sample map,” “T-31 individual inspection results,” and “T-31 process comparison.” Keep the controlled originals in the approved system. If an attachment is revised, identify the revision rather than replacing it silently with another file named “final.”
Only include material you are permitted to share with the recipient. For an external request, use the approved channel and sharing rules for drawings, customer details, and process information. If a required detail cannot be shared, say so and ask what alternative evidence would support the review. Removing a label without explaining the gap can make the sample impossible to interpret.
From “nothing works” to a reviewable request
Constructed teaching example. The job, times, IDs, and counts are invented. This message has not been sent to anyone and does not report a factory case.
The initial draft says: “Short shots again. We increased pressure and speed but it still happens. Could it be the vents? Need a fix.” It names a possible cause and two changes, but gives no sample identity, outcome, or comparison.
Example request to the process owner
Subject: Review requested: incomplete rib at E-3, cavity 2, trial T-31
Observation: At 13:10–13:20, cavity 2 had an incomplete rib tip at E-3 on 4 of 20 inspected parts. Cavity 1 had 0 of 20 under the same visual reference V-4. Samples A-01–A-40 are retained in the trial area; the cavity map and image list are attached.
Context: Part Demo Cover rev B, mold M-6, press P-3. Material grade, lot, and preparation references are recorded in T-31. The last relevant accepted inspection is Q-27. Its sample conditions are documented in the comparison attachment.
Work tried: T-30 changed two controls within its approved plan. The incomplete-rib count fell in that sample, but flash appeared. The record does not isolate the contribution of either change. Current process state and product status are recorded in the shift log; no new change is requested by this message.
Unknowns: No current vent-condition finding is attached. Available machine results are linked, but there is no cavity-pressure trace.
Request: Please review whether the next useful step is a tooling-condition review or a more controlled filling comparison, and identify the evidence needed to choose between them. Process owner review requested before the next planned trial; quality retains disposition ownership.
The revised message does not diagnose the rib. It gives the process owner a specific choice to review and makes the missing vent evidence explicit. It also preserves the unfavorable result from T-30 instead of reporting only the improvement.
If the issue were urgent under the plant's procedure, the required escalation would happen through that procedure. A carefully written message should support the response, not delay it while someone gathers a perfect attachment set.
A question back is not a failed request
A useful specialist reply may ask for the full material grade, a wider image, a measurement method, or a record from the previous shift. Add the requested evidence to the same case and update the list of known gaps. Do not start an unrelated conversation that loses the earlier context.
If someone asks for a value you do not have, distinguish “not measured,” “not retained,” and “available but not yet retrieved.” They lead to different next actions. A guessed melt temperature or reconstructed setting can send the investigation in the wrong direction while appearing more helpful than an honest blank.
When the reply proposes a cause, ask what observation would support it and what would make it less likely. When it proposes a trial, clarify the response to measure and the approved plan. This keeps a suggestion from becoming a finding merely because an experienced person proposed it.
Be precise about previous trials. NIST's discussion of one-factor-at-a-time experiments explains that such experiments do not reveal interactions between factors. Conversely, several changes made together without a suitable plan do not isolate their individual effects. Your request should disclose the trial design and its limits, not label any before-and-after comparison “proof.”
Capture what the reply changes about the next action
Record the reply date, person or role, evidence reviewed, recommendation, and limits. Keep recommendation, authorization, and completed action as separate states. Advice from a supplier or colleague is not automatically the plant's permission to modify a process or release product.
If the response resolves a factual gap, update the record with its source. If it leads to a trial, link that trial ID. If the question remains open, state the next owner and review trigger. Avoid closing the request as “solved” merely because someone replied.
The printable request worksheet is a preparation aid, not a form that sends data. Use it to draft the summary, then work through your existing communication system. Carry unresolved actions into the handover record.
For supporting evidence, use the drying record, process-sheet reading check, or maintenance-change timeline as appropriate. One well-chosen attachment with clear sample identity is often more useful than another page of unsupported explanations.
Sources and scope
ASQ supports separating problem definition from solution work; NIST supports the stated limitation of one-factor-at-a-time experiments. The message structure, attachment guidance, worked request, and worksheet are original MTF editorial work. They are not copied correspondence, a verified support case, or a substitute for a company's controlled communication procedure.