Technical Scoping and Sequencing
Discovery produced a long wish list: all 40 carriers, every kind of charge, a review screen, writing results back into the TMS, monthly reports and, as Meera's late addition, warehouse invoices. The pilot has eight weeks. You cannot do all of it, and the order you work in matters as much as the list itself.
The rule
Thin slice end to end first; riskiest and slowest things earliest.
- A thin slice is one narrow path through every layer of the system: mailbox → reading the invoice → matching it to a shipment → checking the rate → review screen. Do it for one carrier and one or two kinds of charge. Building layer by layer (perfect invoice reading for all 40 carriers before any matching) hides integration surprises until week seven.
- Riskiest first. Ask what could sink the pilot. At Harbourline it is this: can an invoice be matched to its shipment at all? If invoices do not reliably carry the lorry receipt (LR) number, the consignment number that also appears in the TMS, the whole plan changes. Test it in week one on 200 real invoices.
- Slowest first. Some work is mostly waiting: database credentials that need a security review, a VPN account, a firewall change. These take calendar time you do not control, so request them on day one.
- Write down what is out. A scope with an explicit "not in this pilot" list gives everyone something to point back to when new ideas arrive.
The shortcut
Sort every item with two questions: "Could this sink us?" and "Does it wait on someone else's calendar?" A yes to either moves it to week one. Everything else is ordered by value along the thin slice.
List the work
Database access, invoice-to-shipment matching, reading rate cards, review screen, TMS write-back, monthly reports, warehouse invoices.
The trap is polishing one layer while the others stay untested. A perfect invoice reader is worth little if week seven reveals that invoices cannot be matched to shipments.
Drills
Answer fast — trust the rule.
Which plan is a thin slice?
Why request TMS database access on day one?
Which is the riskiest assumption to test in week one?
In week four, Meera asks to add warehouse invoices, which is on the "not in this pilot" list. What does the list let you do?
Remember
- Thin slice end to end first; riskiest and slowest things earliest.
- Shortcut: "Could this sink us?" or "Does it wait on someone else's calendar?" A yes means week one.
- Trap: polishing one layer while the rest of the pipeline stays untested.