Six Months to a Target Architecture at a Three-Unit Freight Operator
Hexin Logistics is a 6,000-employee regional freight operator with three business units - Linehaul, Warehousing and Last-Mile - each running its own IT team, budget and vendor contracts. In five years the units built 14 overlapping systems, including three transport management systems, four customer portals and two billing engines. Month-end reconciliation now takes 11 days and 40 hours of manual spreadsheet work; quote-to-invoice averages nine days against a 24-hour market norm. The board has approved a CNY 60 million three-year envelope but will release nothing further until it sees a target architecture. The new CIO has six months before the budget round. Unit heads argue consolidation will starve their local projects. There is no architecture function, no application inventory beyond a spreadsheet, and no capability model. A vendor offers a 'digital platform' replacing all three TMS products at 30% discount if signed within 90 days.
1. Should Hexin sign the 90-day vendor offer? Set out the analysis you must complete before deciding, and state the conditions under which signing would be the right call.
Read the worked answer
Do not sign on the 90-day clock - but do not reject the platform either. The deadline is a commercial tactic; the decision is an architecture decision, and the two are currently tangled.
Analysis to complete first:
- Capability test. Build a level-2 capability map for the three units. Ask whether the three TMS products differ because the businesses genuinely differ (linehaul load planning versus last-mile route sequencing) or only because of history. If two of the three capabilities are commodities, a single platform fits two units and not the third.
- Quantify the offer. A 30% licence discount is not the price. Model three-year TCO: licences, migration of master data out of 14 systems, integration rebuild, dual-running, retraining, and - critically - the cost of freezing the roadmap for 90 days plus the exit cost if the platform underperforms. If TMS licence spend is roughly CNY 20M of the 60M envelope, 30% is about 6M; a failed migration of the billing and reconciliation chain costs multiples of that.
- Contract terms before signature. Data portability, escrow, termination assistance, price protection at renewal, and phased acceptance tied to the 11-day close actually improving.
Sign if the platform demonstrably covers two units' capabilities, exit terms are acceptable, migration is separately funded, and architecture principles are already approved. Otherwise the vendor is setting your target architecture. Counter the deadline by asking for a six-month option at a smaller discount; the response tells you how much the vendor needs the deal.
2. Design the six-month path to a signed target architecture: what you deliver each month, who owns it, and how you overcome unit resistance.
Read the worked answer
Sequencing. Six months buys a target architecture and a funded roadmap, not a full blueprint.
| Window | Deliverable | Owner | Gate |
|---|---|---|---|
| Month 1 | Inventory of all 14 systems (owners, cost, interfaces) and baseline KPIs | CIO office | Each unit head signs the inventory |
| Month 2 | Level-1/2 capability map; duplication heat map | Enterprise architect | Capability owners named |
| Month 3–4 | Three target-state options with cost and risk | CIO + architecture board | Board paper |
| Month 5 | Business cases and funding gates for wave 1 | CFO + unit heads | Envelope allocation |
| Month 6 | Three-year roadmap, principles, ARB charter | Board | Approval |
Breaking resistance. The 11-day reconciliation is a shared wound - publish it per unit so the pain is attributed to the landscape rather than to a rival unit. Convert each unit head from budget owner to capability owner: they keep decision rights over what the capability must do and lose them over which product delivers it. Fund one visible local quick win per unit in month 2 (for example automated reconciliation for Warehousing) so consolidation is not purely subtractive. Use the vendor's 90-day clock as an external forcing event in the board paper - a deadline the units did not create is easier to align against than one the CIO imposes.
Decision rights. The architecture board must have binding authority over exceptions, or the target architecture is a document rather than a control. Publish exception expiry dates from day one.
3. The board asks how it will know the plan is working. Which measures would you put in the roadmap, and which would you refuse to report?
Read the worked answer
Measure the chain from architecture to business outcome, not the architecture itself.
| Measure | Baseline | 12-month target | Why |
|---|---|---|---|
| Month-end reconciliation elapsed time | 11 days | ≤3 days | The pain the board actually feels |
| Quote-to-invoice cycle | 9 days | ≤2 days | Market norm is 24 hours; 2 days is a credible first step |
| Core finance/logistics applications | 14 | ≤8 | Portfolio outcome, not a document count |
| Manual reconciliation effort | 40 h/month | ≤10 h | Direct labour release |
| Run versus change spend | unknown - measure in month 1 | 70/30 → 60/40 | Shows whether savings are reinvested or absorbed |
Refuse to measure: number of architecture artefacts produced, 'EA maturity level' before a baseline exists, and interface counts without a cost attached. All three reward activity and are trivially gamed.
Leading indicators to add: share of new projects that passed architecture review before funding, and the number of live exceptions to principles (should fall, and each should carry an expiry date).
Report outcome measures quarterly to the board and leading measures monthly to the architecture board. Mixing them in one dashboard makes the board manage activity instead of results.