← Back to Exelon case study

Exelon · Deep dive

The Material-Request Black Hole

Why could completed work still have unresolved material—and who should own the remaining decision?

Cross-role ResearchRoot-Cause AnalysisService Ownership Design

A job could be physically finished while its material request stayed open. The first explanation was simple: Stores had issued the material, so Stores should close the request.

I tested that explanation against both sides of the handoff: Operations, who saw the open request, and Stores, who handled the material.

01 /

Two sides of one complaint

Both accounts were accurate. Neither was complete.

The first explanation, that Stores simply had not closed the request, fit what Operations could see. It was a reasonable hypothesis, and it described only one side of the service.

What Operations experienced

  • The job was physically complete
  • The material had already been used
  • The request was still open
  • Supervisors followed up, sometimes walking a printed copy to Stores

What Stores experienced

  • It knew what it had issued and staged
  • Lines were still open on the request
  • Nothing in the request showed whether the job was finished or the lines were still needed

A Stores participant explained that Stores knew what material had been issued, not whether the job itself was finished, and described the resulting responsibility as unclear.

02 /

How a request moved

Physical state was clear the whole way. Job intent was not.

A request described what was planned. The field decided what was actually needed, and Stores correctly charged only what left the building. That left a gap between planned and actual material.

Current-state material request flow

  1. Owner: Operations side

    Request created

    A planner or job owner creates the request, which also creates demand against inventory.

  2. Owner: Operations side to Stores

    Paper handoff identifies the exact request

    The crew or job owner brings a paper copy so Stores can find the exact request.

    Why paper persisted

    A Stores participant said there were too many digital requests to find the right one. Paper was doing real work:

    • Exact-request selector
    • Handoff artifact
    • Shared physical reference

    It also signaled need and priority, and triggered staging.

    Do not digitize the artifact. Digitize the function it performs.

  3. Owner: Stores

    Pick and stage available material

    Stores prints a pick list, pulls what is available, and stages it. Lines the system shows as unavailable are left off the list.

  4. Owner: Stores

    Issue what is actually taken

    Stores charges only what physically leaves, which can be less than requested.

  5. Owner: System

    Close if every line is resolved

    The request can close on its own when every line is resolved.

  6. Owner: Ownership unclear

    A remaining line needs interpretation

    If lines remain, the request stays open and someone must decide what they mean.

    Physical state is known

    • What was issued
    • What was staged

    Job intent is not clear

    • Whether the job is complete
    • Whether the line is still needed

    A remaining line could mean

    • Still needed
    • Not needed
    • Sourced elsewhere
    • Deferred
    • Canceled

    Nothing in the request says which meaning applies. A Stores participant said it was not clear who was supposed to do what.

Reconstructed for portfolio presentation based on original project research and surviving notes.

Based on interviews with Stores and Operations participants.

03 /

What made it harder

Four things complicated an already ambiguous handoff.

Physical vs system availability

Material could be on the shelf but shown as unavailable because it was reserved to other work. A Stores participant described the effect, not the rule.

Cross-center sourcing and ghost demand

When a crew sourced material from another location, the original request could stay open, and special-order material could still arrive for finished work.

Stale staging

Stores could see a cancellation more readily than a reschedule, so staged material could sit until someone asked.

Emergent borrowing

Urgent work could borrow material from a later job, creating a backfill obligation tracked by email.

04 /

Who should own each decision

Assign each decision to the role that actually knows the answer.

Stores knows what is physically true. Operations knows what is true about the job. Neither can safely infer the other's state, so the service has to connect them.

Ownership model

  • Stores

    Physical truth

    • What was picked and staged
    • What was issued
    • What is physically on hand

    Not asked to infer job intent.

  • Shared service

    Connection · state · routing

    • Keeps job, request, and material connected
    • Shows physical and job state together
    • Routes the decision to the role with authority
    • Raises an explicit exception when they disagree
  • Operations

    Job truth

    • Whether the job is complete
    • Whether the material is still needed, now or later
    • Whether the work was deferred or canceled
    • Whether another source was used

    Not asked to infer physical inventory state.

When a material line stays unresolved

  1. A material line remains unresolved.
  2. Physical state and job state are shown together.
  3. The decision routes to the role with authority.
  4. If the meaning is still unresolved, it becomes an explicit exception.

Where a meaning depends on both truths, both are shown together, not assigned to one side.

Illustrative. Synthesized from research findings; not an original project artifact.

Research-derived future-state model — not an implemented workflow.

Closeout safeguard

An Operations participant reported that finishing work could cancel a still-unresolved request. The research-derived response is to warn or stop before the final completion step if required material remains unresolved. Participant-reported and research-derived, not an implemented behavior.

PrincipleDon't make Stores guess job intent. Don't make Operations guess inventory transactions. Make the service connect the two.

05 /

What the research changed

The initial complaint was true, but incomplete.

Operations correctly saw completed work leave material behind. Stores correctly could not close material it did not know was no longer needed. The research did not choose a side. It exposed the missing service between them.

That moved the design question from "how do we make Stores close the request?" to "how do we connect work status, material status, and decision ownership so the right role can safely close the loop?"

Make material, work, and ownership converge

These are design intent, not measured, post-launch results. The engagement ended before implementation.

Intended future-state outcomes

  • Fewer open or orphaned requests
  • Fewer stranded reservations
  • Less stranded special-order material
  • Faster emergent sourcing
  • Clearer accountability
  • Better alignment between physical inventory and system state
  • Safer closeout transitions

Cross-role research can change the problem entirely

Talking to only one side would have produced a confident, wrong fix. The value of interviewing both was not extra detail. It was seeing that the missing piece was the connection between them.

The interface problem was visible in one screen. The real problem lived in the handoff between two teams.