Testing a $3 Million P2P Business Case

Rebuild the benefit stack by dependency, not by size.

1. Price improvement, contract-driven. 4% on $240 million is $9.6 million theoretical, and that assumes 100% compliance, the exact assumption that failed in 2021. Apply a realization curve of 25% / 55% / 70% across three years and restrict it to spend actually influenceable, roughly $180 million after excluding sole-source and pass-through. That gives about $1.8M, $4.0M, and $5.0M. Still large, now defensible.

2. Invoice cost, process-driven. 310,000 invoices at roughly $9-10 fully loaded for paper versus $2-3 for electronic. Moving 200,000 electronic by year two saves $1.3-1.5 million in AP effort, but only if headcount is genuinely redeployed. Keep this as a separate business case owned by the CFO.

3. Leakage and payment integrity. Duplicate vendor records are a fraud and duplicate-payment exposure, not just a reporting nuisance. Recovery audits typically find 0.1-0.3% of spend, so plan $300-700k, and note this benefit does not depend on user behavior.

4. What I discount. The vendor's 14-month payback assumes all three legs at full value simultaneously. On the adjusted numbers, three-year benefits of roughly $7.5-8.5 million against $3.0 million of cost is still a strong case with payback around 22-26 months.

Conditions in the recommendation. Fund in gates, releasing $600k for data and process work only after master data and process standards pass a test. Contract on milestones and adoption with a 15% holdback and no payment for go-live without measured usage. Name adoption, not software, as the investment thesis, and put an adoption dashboard next to the financials. Do not count AP headcount savings as procurement savings; mixing budget owners is how business cases lose credibility.

Phase 0, months 0-3: fix the data before the software.

• Vendor master cleanup: dedupe the 3,900 records, standardize legal names, validate tax IDs, and lock down bank-detail change controls, which is a fraud exposure rather than a reporting issue.
• Classify the top 80% of spend to UNSPSC; the tail can be coded later.
• Map the current process with the people who perform it, not with the vendor's reference model.

Phase 1, months 3-9: invoice and catalog. E-invoicing with the top 150 suppliers, targeted at about 60% of invoice volume. Hosted catalogs for the top 20 categories to cut free-text requisitions. A contract repository for the top 100 agreements so pricing is visible before sourcing.

Phase 2, months 9-18: e-sourcing and guided buying. Three sourcing events in high-leverage categories, run against the old process for one cycle only. Guided buying for tail spend, the only realistic way to control 38% of transactions.

Phase 3, 18+ months: analytics and supplier self-service.

Five mechanisms that differ from the CRM failure.

1. Process owners, not IT owners. AP and category leads carry adoption targets in their objectives.
2. Test with real work before design freeze. AP clerks run 20 real invoices through the prototype; buyers run a real RFQ.
3. Hard cutover where possible. Switch off the old PO form on a date with a documented exception route. Indefinite parallel running is how adoption stalls at 30%.
4. Super-user network of 8-10 people with formally allocated time.
5. Per-site adoption dashboards reviewed monthly by the CFO, escalating to the site lead after two consecutive months below threshold.

Supplier side. Segment suppliers: white-glove onboarding for the top 150, portal plus phone support for the tail, a grace period, then a paper-invoice fee. Give suppliers a visible benefit such as PO visibility and faster payment, or they will route around the system.

Position: do not commit to eight in year one. Commit to six by the end of year two, in writing, with quarterly milestones.

Reasoning.

• 78% of invoices are paper today, so the first 12 months require more AP capacity, not less: supplier calls, exception handling, PO cleanup, and rework. Cutting staff during migration removes the people whose cooperation determines whether suppliers enroll.
• Attrition gives a realistic glide path. At 12-15% annual turnover on 11 FTE, 1-2 positions turn over each year; a hiring freeze plus redeploying 3-4 people into supplier enablement reaches six fewer by month 24 without involuntary separations.
• Involuntary cuts create a resistance bloc that will tell every supplier the system is a headcount exercise, which becomes self-fulfilling.

What I would offer the CFO.

• A signed glide path: zero in year one, three by month 18, six by month 24, contingent on e-invoice penetration reaching 65%, with a quarterly report.
• A holdback: if penetration misses, the headcount commitment slips and I own that.
• Savings credited to the CFO's line, not mine. My accountability is the adoption metric that produces them.
• Full transparency on cost per invoice before and after, so the claim is measurable rather than asserted.

What I would not do. Sign up for eight in year one to win the meeting. A business case that depends on a headcount reduction that cannot be delivered in the year it is promised is exactly the pattern that produced the CRM overrun: optimistic benefit timing, no gate, and nobody accountable for the gap.