As of 13 August 2026, AI can analyse your quality control data.

What to actually do

1. Hand it to a person

   The route this page recommends

   A person who owns the outcome does this end to end, worth it when the failure is dear.

2. Use a tool built for this

   Second choice

3. Do it yourself

   The distant third

   A chat interface, power-user skill, and roughly 30 minutes until you can act on the result.

   How to actually do it

   1. Open the source quality records and export a copy as CSV or XLSX, keeping record identifiers, dates, batches, measurements, units, specification limits, pass or fail results, supplier, shift, machine, and operator fields where available.
   2. Remove unnecessary personal information, then check that the exported columns have consistent names, units, date formats, and pass or fail labels before sharing the data.
   3. Gather the current specification limits, acceptance criteria, sampling method, and a short list of relevant process changes or incidents from your quality documents.
   4. Open a chatbot, paste the prompt, replace each bracketed context slot, and paste the cleaned table or upload the file if the service supports it.
   5. Ask the model to show the calculations behind every total, percentage, trend, outlier, and comparison, then save the response with the source-file name and date.
   6. Compare each reported count and flagged record with the source data, reproduce key calculations in your spreadsheet, and have a quality colleague assess the findings before any release, rejection, or corrective action.

   Prompt

   Analyse the quality control data below. Treat the data as observational evidence, not as proof of cause. Do not invent missing values, specifications, units, dates, batches, or explanations.
   
   Context:
   - Product or process: [describe it]
   - Measurement names and units: [list them]
   - Required specification limits or acceptance criteria: [paste them]
   - Relevant process changes, suppliers, shifts, equipment, or incidents: [list them]
   - What a pass or fail means: [describe it]
   
   Data:
   [paste a CSV or table here]
   
   Produce:
   1. A short description of the dataset, including its date range, row count, fields, missing values, and possible duplicate records.
   2. Data-quality problems that could affect the analysis.
   3. Counts and percentages of passes, failures, and missing results, showing the denominator used.
   4. Trends over time and differences by batch, product, supplier, shift, machine, or operator where those fields exist.
   5. Outliers and unusual runs, quoting the relevant record identifiers and values.
   6. Comparisons with the supplied specification limits. Do not create limits where none are supplied.
   7. A clear separation between observed findings, plausible hypotheses, and claims that cannot be supported by this data.
   8. The exact calculations or simple reproducible steps used for every important figure.
   9. A short list of follow-up checks for a qualified quality colleague, including which raw records or process conditions to inspect.
   
   Do not recommend releasing or rejecting product, changing a process, or closing a corrective action. Flag any result that needs review by the responsible quality professional.

   Open it prefilled in ChatGPT or Claude, or copy it into Gemini, which takes no prefill link.