Quick Guide · Root cause analysis

Build an Incident Timeline from Fragmented Records with Copilot

When investigating site incidents, pulling together maintenance logs, production records, and pack data into a coherent sequence often consumes critical technical time. This quick guide demonstrates how to use Microsoft Copilot Chat to consolidate exported logs into an attributed chronological timeline, ready for human verification.

13 Nov 2026 · 40 sec · Copilot Chat (Copilot Chat, no licence needed)

The scenario

Site
Ambient savoury snack manufacturing site (crisps and extruded snacks)
Who
Quality Systems Manager and investigation lead
Trigger
Customer complaint received regarding a piece of hard blue plastic (approx. 12 mm x 8 mm) found inside a 40 g bag of crisps.
Why it matters
Jumping to 'operator error' risks closing the incident without addressing the genuine contamination mechanism, leaving open product vulnerable to repeat occurrences and failing the fundamental expectations of root-cause evaluation during BRCGS audits.

BRCGS context

ClauseWhat the Standard expectsSensible practiceWhere technology helps
3.7
Corrective and preventive actions
Failures are investigated, root cause identified where needed, and actions implemented and verified.Sensible industry practice involves gathering timeline and maintenance data, constructing a multidisciplinary fishbone diagram, and testing multiple hypotheses before concluding on true root cause, rather than defaulting to operator blame.Copilot can collate messy log extracts into a chronological timeline and draft candidate categories for a fishbone diagram; it stops at data organisation, leaving hypothesis testing and physical verification to the factory investigation team.

Clause intent in plain English, not the Standard's wording. Check requirements against your own licensed copy of BRCGS Food Safety Issue 9.

The approach, step by step

  1. 01

    Upload logs and establish incident timeline

    Review the uploaded complaint, batch log, and maintenance records. Draft a chronological timeline for Line 4 between 06:00 and 13:00 on 14/05/2026, noting relevant engineering interventions.

    You check: Verify all timestamps against the signed production and engineering job sheets to confirm no shifts or missing operational events were omitted.

  2. 02

    Identify conflicting information and evidence gaps

    Compare the detection log and maintenance history against the complaint item. List what evidence is missing or unverified regarding hard blue plastic components on this line.

    You check: Cross-reference against the site's physical foreign body register to see if the blue modular belt or transfer chute guides match the fragment description.

  3. 03

    Draft structured fishbone categories and exploratory questions

    Draft an Ishikawa fishbone structure (Man, Machine, Material, Method, Milieu) with 3 exploratory questions per branch to investigate potential entry points for blue plastic.

    You check: Ensure questions avoid leading conclusions and adapt the technical prompts to reflect actual Line 4 equipment layouts and handling procedures.

  4. 04

    Prepare investigation meeting briefing sheet

    Format the timeline, identified evidence gaps, and fishbone questions into a one-page investigation meeting agenda for the production engineer and hygiene lead.

    You check: Confirm the briefing material contains only anonymised production data and accurately states known physical facts versus hypotheses.

Takeaways

  • Use AI strictly to organise timelines and highlight data gaps, never to determine incident root cause autonomously.
  • Competent engineering and technical personnel must physically inspect line components and verify timestamps against primary records.
  • BRCGS clause 3.7 requires genuine root-cause analysis; defaulting immediately to operator retraining without factual backing does not meet expectations.

Limitations and risks

  • Copilot cannot inspect or compare physical material samples, density, polymer types, or colour swatches.
  • Copilot cannot access factory SCADA, CMMS, ERP, or detector logs directly; files must be manually exported, checked, and uploaded.
  • Copilot Chat lacks grounding in the full tenant document library unless using paid M365 Copilot with Microsoft Graph, so only uploaded files are considered.
  • The model cannot determine whether an engineering change conformed to technical hygiene sign-off standards without written records being provided.
  • Pasting proprietary line layouts, customer names, or identifiable worker records into unauthorised public tools; work-account enterprise data protection must be verified.
  • Accepting Copilot's plausible timeline draft without checking source paperwork for off-shift engineering modifications or missed logbook entries.
  • Treating drafted fishbone branches as a completed root cause analysis rather than starting points for physical verification on the packing line.

Sources

TFCD content is general guidance, not professional advice for your site. Always check requirements against your own licensed copy of the BRCGS Standard and your customers' codes of practice. Narrated with an AI voice clone of the presenter; Copilot screens are illustrative recreations with dummy data.