Billing-first sequencing at a hybrid hardware, support, and usage business
Northwind Instruments sells laboratory hardware (average order $48,000), a three-year support subscription bundled with each unit, and a usage-based data plan billed monthly in arrears. The company sells through three entities (US, Germany, Japan), quotes in 11 currencies, and both sales teams still build quotes in spreadsheets. Finance takes 12 days to close revenue and has been told to reach five. Volume is roughly 700 orders per quarter, 4,200 active subscription lines, and three metering sources pushing about 90 usage files a month. The steering committee proposes going live with Billing first and adding Products and Quotes in a later phase, because billing is where the pain is. Sales has asked to keep spreadsheets for one more year.
1. Evaluate the steering committee's billing-first sequencing proposal. What would you recommend instead, and why?
Read the worked answer
Billing is the last stage that can safely go live, not the first. Billing consumes three upstream artifacts: catalog data (what was sold, at what price, in which currency), order/asset data (who owns it, from what date), and contract data (term, co-termination, uplift). A billing-first go-live therefore forces the team to populate the catalog and asset records anyway, but does it as a side effect of billing configuration rather than as a governed catalog. The result is hardcoded product lists and price logic buried in billing schedules - the spreadsheet problem reproduced inside a new system, and far harder to unwind.
The usage stream makes this worse. Rating requires a metering-source-to-product mapping and a price per unit that lives in a price book; there are 90 files a month from three sources, so errors compound quickly.
Recommended sequencing: (1) catalog plus quote-to-order for the two standard motions - hardware with its support bundle; (2) usage rating and billing; (3) revenue and reporting. The hardest governance decision is the spreadsheet cutoff. Dual-running for a year guarantees a reconciliation burden and two versions of truth at close. Push for a fixed retirement date per motion rather than a blanket one-year extension.
2. Which of the three revenue streams is the worst candidate for phase one, and what makes it so?
Read the worked answer
The usage-based data plan. Three factors combine:
- Integration surface: 90 files a month from three sources means late, duplicated, and malformed files. This is a data-operations problem that will dominate the project regardless of how well the revenue model is configured.
- Variable consideration: usage revenue must be estimated or recognized as consumed, and disputes over metered quantities are common. Getting this wrong at go-live is a restatement risk, not a reporting annoyance.
- Low transferability of learning: solving usage rating teaches the team little about the hardware and support motions, which carry most of the order volume.
The hardware-plus-support bundle, by contrast, is a standard point-in-time plus ratable pattern with observable standalone selling prices. It exercises the catalog, quoting, contracting, billing-schedule, and revenue path end to end with the lowest variance. That is the motion that should prove the model.
3. Name the governance decisions that must be settled before any build starts, and say what breaks if each is deferred.
Read the worked answer
- Product identifier ownership. If item masters live in an ERP and the catalog is a copy, every new SKU needs a sync rule and an exception queue. Deferring this produces duplicate products and orphaned billing schedules.
- Currency and price book strategy for 11 currencies. Decide whether local prices are entered per currency or converted from a base with a rate source and refresh cadence. Deferring it means prices are entered ad hoc, and someone will discover a rounding or repricing problem after invoices go out.
- Customer master. Which system owns the account, and how are the three entities and their tax registrations represented? Invoicing and revenue both key off this.
- Metering normalization. One canonical usage record shape, with source, period, and quantity, before any rating build.
- Revenue policy ownership. Who signs off on standalone selling price, allocation, and catch-up versus prospective treatment for changes? Without a named owner, mid-project changes reopen closed designs.
Each of these is cheap to decide in week one and expensive in month six.