DEV Community

Ezra Duthie
Ezra Duthie

Posted on

Keep a handoff ledger before drawing a delivery map

A delivery map can look wonderfully precise while hiding the question a reporter actually needs to ask. Where did the goods wait, and who recorded that wait? A line between two places does not tell me whether a vehicle was moving, a warehouse was holding stock or someone was waiting for a corrected document.

I find a small, plain ledger more useful at the beginning. Each row describes one reported event: a consignment reference, a place, an event type, a timestamp and the source of that information. The aim is modest. I want to know what the records can support before I turn them into an explanation.

A reporter checks the record of a shipment.

Give every timestamp some context

A time without a time zone is an invitation to make a confident mistake. I keep the original value in one column and put any normalized value in another. If a source cannot explain the time zone, I mark that uncertainty rather than silently borrowing the one used by a different system.

The same care applies to the event itself. Does dispatched mean that the paperwork was completed, that a cage reached the loading area or that the vehicle left the gate? Those events might happen close together. They are still different events, and the distinction becomes important when an article describes a delay.

Let a script find questions

A dispatcher checks a consignment at a warehouse.

A short validation script can check whether required fields are empty, whether an event type is recognized and whether a timestamp can be parsed. It can flag duplicate references and apparent reversals in the event sequence. I would rather see those problems in a separate exceptions file than discover them underneath a finished chart.

I keep the checks readable and run them on a copy. A suspicious sequence should produce a question, not an automatic correction. Perhaps the event was recorded later than it happened. Perhaps two systems use different clocks. Software is good at finding an inconsistency; it cannot interview the person who understands it.

Keep the source beside the calculation

For each row, I retain a reference to the document, export or interview note that supplied it. This makes an awkward follow-up much easier. Instead of asking whether a whole dataset is reliable, I can ask what a particular event means and show the source exactly where my interpretation began.

Private customer details do not belong in a public working example. I remove names, addresses and identifiers that readers do not need. A synthetic demonstration can explain the validation method without publishing someone else's operational records. The private reporting file and the public explanation serve different purposes.

Draw only what the ledger supports

Once the exceptions have been discussed, I can choose a visual. Sometimes a simple timeline explains more than a geographic map. Sometimes the honest result is a short description of a gap in the records. That is still useful reporting.

The reward for this small software habit is less dramatic than a moving dashboard. It is a clearer question, a reproducible calculation and a better account of what remains unknown. For supply chain journalism, those are excellent reasons to begin with a ledger.

Top comments (0)