Auditors Couldn’t Find the Data Behind a Program’s Savings Claims

← Back · ACT III · Chapter 17: VIGOR: Governance and Realization

In September 2025 the Government Accountability Office published its review of category management at the Department of Veterans Affairs. Category management is the federal effort to make agencies buy like one enterprise instead of as hundreds of separate offices. The report is titled around leadership accountability and training, and that is how the trade press read it. The finding I keep returning to sits in the fifth recommendation, and it turns on a single word.

GAO asked VA to have its category managers “assess the availability of, or feasibility of making available, item-level data on prices paid, utilization, and cost avoidance.”

The word is availability. GAO was not yet asking whether the data was accurate, current or in use. Before any of that, the auditor could not assume the data existed.

What the Numbers Said Before That

By its own numbers the program was working. In fiscal year 2024, 91.7 percent of VA’s 67.2 billion dollars in contract obligations sat on contracts considered managed according to category management principles, against a goal of 90 percent. Government-wide, the General Services Administration reported 58.3 billion dollars in cost avoidance from fiscal years 2021 through 2024, and VA reportedly accounted for 14.3 billion of that.

Two further sentences come from the same report. VA reported savings for government-wide contracts that accounted for 7.6 percent of its fiscal year 2024 contract obligations. And category leads told GAO that they “did not set or manage toward category-specific savings goals.”

So 91.7 percent of the spend sat inside the governance regime, 7.6 percent of it carried a reported saving, and the people running the ten spending categories were not managing toward a savings number at all. None of those figures contradicts another. They measure different things, and only one of them is the thing the program exists to produce.

Three Words, Three Different Rungs

Read the fifth recommendation again: prices paid, utilization, and cost avoidance. They sit at three separate points on the same chain, and each is the evidence that converts the point before it into the point after.

Cost avoidance is what was negotiated: a price that did not rise, a term that moved. Prices paid is what the organization actually bought, at what volumes, on the contract and off it. Utilization is what was used out of what was bought. A negotiated figure becomes a purchased figure only if the buying followed the contract, and a purchased figure becomes a realized one only if the thing was consumed.

The 14.3 billion dollars lives on the first rung. The two rungs that would carry it to the last are the ones GAO had to ask whether VA could produce at all. The report does not say the figure is wrong, and neither do I. It says the figure is one kind of number, reported without the others, by managers who had not been asked to manage toward the others. VA concurred with all six recommendations, and as of October 2026 GAO’s page lists all six as open.

Why This Is Not a Federal Problem

A version of the same meeting happens inside almost every large buyer. Procurement reports a reduction on a major category. Finance opens the ledger and shows that spending on the category went up. Someone senior asks which number is right. The models that produce the first number get built in business value engineering, which is where I built them.

The usual answer is reconciliation. Baselines get adjusted, methodologies get argued, a memo explains why the two figures cannot be compared directly, and the following year everyone agrees to use a blended one. That makes the conflict go away and leaves its cause exactly where it was.

The gap has a location. If the negotiated figure was 100 and the realized figure was 60, the missing 40 did not evaporate. It left at a rung. Maybe volumes rose faster than the price fell, or a division kept buying from the old supplier. Perhaps half the seats were never provisioned, or an index clause reset the price in month nine. Each of those leaves its own record: a purchase order, a usage log, an invoice line. The disagreement between the two functions is telling you which rung nobody was watching.

What a Standard Puts Before the Purchase

The Value Standard puts the question at the point of purchase rather than at year end. Before the money moves, the buyer declares which measure is the claim of record for this decision, what evidence converts each rung to the next, when each conversion gets measured, and where that evidence is kept. The declaration is the buyer’s, and it is structural: it names the record rather than a person to blame when the record turns out to be empty. A claim without a record is still only a claim, whichever rung it was made on.

A purchase governed that way still produces disagreements between Procurement and Finance. What changes is what the disagreement means. It stops being an argument about whose number wins and becomes a pointer to the rung where the evidence stopped, which is the only place the value can be recovered.

A supplier that delivered gets something out of this too. When the records at each rung exist, its savings claim can be checked and stands on its own, instead of being averaged away in next year’s reconciliation.

The VA report is useful because an auditor with full access to the records went looking for those rungs and had to begin by asking whether they were recorded anywhere. Most organizations have never had anyone ask.

A coverage number says how much spend passed through the regime. On its own it says nothing about whether the regime produced anything. Joining the two takes item-level records at each rung, and the place to settle who keeps them is before the purchase, not in the meeting where the two numbers are read out.

Reference: U.S. Government Accountability Office, “VA Acquisitions: Leadership Accountability and Savings Goals Needed to Improve Purchasing Efficiency,” GAO-25-107398, published September 2, 2025, gao.gov/products/gao-25-107398. Figures and quoted phrases are taken from the report’s Highlights and Recommendations pages; recommendation status is as shown on that page in October 2026. The report has been publicly noticed elsewhere as an accountability and training story, and as a contracting-opportunity story; the reading of recommendation 5 as a measurement chain is this site’s.

Keep reading: ACT III · Chapter 17: VIGOR: Governance and Realization