Before you buy an e-procurement system, count your paper

Somewhere this quarter, a procurement director is sitting in a demo. The platform on screen promises to cut cycle times by three-quarters and administrative handling by most of what it costs today. The numbers are real enough, in the sense that somebody measured something somewhere. And nobody in the room can say what the current process costs.

That gap is the single most common failure point in public procurement modernization, and it has nothing to do with software. A government that cannot state its cost per solicitation, its cost per supplier registration or its cycle time per contract action has no baseline. Without a baseline it can't scope the project, can't sequence it, can't defend the spend to a finance committee and can't tell two years later whether the thing worked.

So before the demos, do the boring part.

What You're Not Measuring

The waste in public procurement isn't in what governments buy. It's in the paper wrapped around how they buy it, and none of it appears as a budget line.

It's the bid packet that gets printed, couriered out and mailed back. It's the contract that circulates for three weeks collecting wet-ink signatures from five approvers, one desk at a time. It's the registration form a vendor fills out by hand so a staffer can key it into a system a second time, adding a second chance for error and duplicate vendor records that haunt master data for years.

These costs are invisible because they're distributed. Nobody in a county or a state agency owns “the cost of paper.” It shows up as labor spread across dozens of desks, as cycle time nobody clocks, as an early-payment discount that expired while a contract waited on a fifth signature.

Governments do eventually price this pattern, usually when a regulator forces the issue. In March, the Centers for Medicare & Medicaid Services finalized a rule replacing fax and mail for claims attachments with standardized electronic exchange and electronic signatures. Projected savings: roughly $782 million a year. That figure existed the whole time. Someone finally counted it.

Why the Vendor Benchmarks Won't Do This For You

You'll be offered numbers. A document costs about $8 to handle. Supplier onboarding runs several hundred dollars. E-signature cuts administrative handling by three-quarters.

Treat all of it as order-of-magnitude. These figures come largely from cross-industry vendor and analyst research, not controlled studies with published government samples, and public procurement differs from the private-sector processes they're drawn from. Sealed-bid rules, public posting, protest windows and records retention all change the arithmetic. Statutory windows in particular set a legal floor on how fast an award can move that no routing engine can repeal.

None of that means the direction is wrong. It means the magnitude is yours to establish, not theirs.

Four Numbers Worth Having Before the RFP

Cost per procurement document. Take a representative solicitation and add up the labor: drafting, printing, mailing, logging, opening, tabulating, filing. You'll want that detail anyway when someone asks you to justify the project.

Cost per supplier registration. Time a staffer re-keying a mailed vendor packet, then add the downstream cost of fixing what gets keyed wrong. The rework is usually the larger number, and it's the one that gets left out.

Cycle time per contract action. Not the average, the distribution. The tail is where the money is: contracts that sat for six weeks because an approver was on leave.

Duplicate and stale vendor records. Pull the count. It's a proxy for how much manual data entry your process depends on, and it predicts whether a new platform will inherit your problems or leave them behind.

None of these require a consultant. All of them can be assembled in a few weeks by staff who already do the work.

What the Numbers Will Probably Tell You

In most governments the same three processes carry most of the recoverable cost, and they're worth attacking in roughly this order.

Contract signatures come first, because it's the cleanest win and the least controversial. A routed e-signature envelope replaces the physical relay, and the tamper-evident audit trail it produces is stronger evidence than ink on a page. Under the federal ESIGN Act, the signature is fully enforceable.

Supplier registration comes second, and it's usually the biggest number. Self-service registration flips ownership of the data to the vendor, ending the re-keying and closing the fraud surface created by hand-entered banking details.

Sealed bidding comes third, because it draws the most objections and needs the most care. The concern is legitimate: pricing has to stay confidential until a public opening. Access controls handle this better than an envelope does, keeping pricing sealed until the designated time and releasing it to everyone at once with a full record of who did what and when.

Two Things Measurement Won't Catch

A cost baseline won't tell you whether people can actually use what you build. A public portal has to meet Section 508 and ADA standards, or some employees and vendors simply can't transact with you. And small, rural and first-time bidders get shut out when registration and bidding move online without training, help-desk support and low-bandwidth options, which narrows competition rather than widening it.

Neither argues for keeping the paper. Both argue for funding adoption and supplier outreach as line items rather than afterthoughts, and both are far easier to fund when you can show the finance committee what the paper costs.

The direction is settled: NASPO's 54 state and territory chief procurement officers ranked modernizing the procurement process their number-one priority for 2026, with eProcurement second. But McKinsey's research with the University of Oxford found large IT projects run 45% over budget on average while delivering 56% less value than predicted, and procurement modernizations sit squarely in that population.

The difference between the programs that deliver and the ones that don't usually isn't the platform. It's whether anyone counted the paper first.

Animesh Dhole is an ERP application analyst for the Arkansas Department of Shared Administrative Services, where he works on the state's SAP-based procurement modernization across sourcing, contracts and supplier management. He holds an M.S. in computer science from Indiana University Bloomington.

Disclosure: The author is employed by the State of Arkansas, an SAP customer, and received no compensation from any vendor for this article. The views expressed are his own and do not represent those of the State of Arkansas or any of its agencies.