Maverick spend is purchasing that bypasses the agreed process or the agreed supplier: buying off-contract when a contract exists, or outside the purchasing system when a channel exists for it.
The word carries a moral charge that the phenomenon does not deserve, and that charge is the reason most organisations treat it badly. Maverick spend looks like a compliance problem, so it attracts compliance remedies, and compliance remedies do not work on it. It is also routinely conflated with tail spend, which is a different thing entirely and needs a different response.
Maverick spend is not tail spend
The distinction is simple and worth being strict about. Tail spend is spend that has no contract. Maverick spend is spend that ignores one. A purchase can be either, both or neither:
- Neither. A catalog order placed against a framework agreement with a strategic supplier.
- Tail only. A one-off purchase from a small supplier in a category where nothing has ever been sourced, raised correctly through the purchasing system.
- Maverick only. A significant order placed with an alternative supplier while a negotiated contract for that exact category sits unused.
- Both. A small off-contract purchase from an unmanaged supplier in a category a framework already covers.
The reason to keep them apart is that the remedies diverge. Tail spend needs coverage: sourcing, or at minimum a structured way to transact where no sourcing is justified. Maverick spend needs the compliant path to be usable, because the coverage already exists and is being declined. A programme that treats them as one problem will apply the wrong remedy to half of it, and will typically try to source its way out of a usability failure.
What it actually costs
The obvious cost is the smallest one.
The negotiated price is forfeited. The organisation paid for that price with volume commitments and negotiating time, and then did not use it.
Rebates and volume tiers are missed, and the miss compounds. Spend routed away from the contracted supplier does not count towards the threshold that sets next year’s price. One period of leakage makes the following period’s rate worse, which makes leakage more attractive.
The transaction is unclassified. An off-catalog purchase arrives as free text, without a classification code or a reliable supplier reference, so it does not aggregate into the category it belongs to. This corruption is self-concealing: the category looks smaller than it is, so it ranks lower in the next sourcing plan, so it stays unmanaged. Maverick spend hides itself in the data by the same mechanism that creates it.
The supplier may be unqualified. Buying outside the contracted list means buying from a company that has not been through supplier onboarding: no insurance certificate, no sanctions screening, no data processing agreement, no confirmation of who they are beyond an invoice header.
The invoice will not match. Either there is no purchase order at all, or there is one carrying a free-text line at a guessed price. Either way it lands in exception handling, and three-way matching turns into a manual investigation that costs more to resolve than the item was worth.
The contract’s protections do not apply. Service levels, warranty terms, indemnities and liability caps were negotiated into an agreement the purchase was not made under. When something goes wrong, the buyer discovers they are relying on the supplier’s standard terms.
Why it happens
Maverick spend is nearly always a system failure rather than a discipline failure, and the evidence is that the same people behave differently when the system changes rather than when the policy does.
The recurring causes are few and specific:
The catalog does not contain the item. The requester looked, found nothing, and had a job to do. Coverage gaps are the single largest source, and they are largest in exactly the categories nobody has got round to enabling.
The item is there but cannot be found. Supplier descriptions written for a warehouse system are not searchable by the people doing the buying. Someone searching for “nitrile gloves large” will not match a line reading “GLV NTRL PWDRFREE BL L 100PK”, and a failed search is indistinguishable from an absent item. This is the most under-diagnosed cause of the lot, and catalog management is where it gets fixed.
The approval path is slower than the operational need. A three-stage approval chain running over several days is a reasonable control on a capital purchase and an obstacle on a forty euro part needed before a shift starts. The requester is accountable for the shift, not for the control.
The contracted supplier cannot deliver in time. The framework supplier has a five day lead time and the need is tomorrow. No amount of policy changes the lead time.
Nobody told the requester a contract exists. The agreement lives in a contract repository that only the procurement team opens. The person raising the requisition has no way of knowing which categories are covered, and no prompt at the point where knowing would matter.
In every one of these the requester is behaving sensibly. They are optimising for the outcome they are measured on, using the fastest route that produces it. Genuine preference-driven buying does exist, usually loyalty to a familiar sales representative, but it is the minority case, and it survives mainly where the compliant path offers nothing better in exchange.
Why enforcement-first responses fail
The standard response is a policy memo, a blocked supplier, an escalation path, and a compliance figure reported to the executive committee. Each of these addresses the symptom while leaving the cause untouched, which produces a predictable result.
Blocking a non-compliant route when the compliant route cannot meet the need does not create compliance. It creates displacement. The purchase reappears as a corporate card transaction, an expense reimbursement, a favour from another department’s budget, a free sample that arrives with an invoice later, or a retrospective purchase order raised to legitimise an invoice that already exists.
That last one deserves attention, because it is common and it looks like success. A retrospective purchase order produces a matched invoice and a clean compliance metric while the buying behaviour is entirely unchanged. Any organisation reporting a sharp improvement in on-contract spend should check what share of its purchase orders were created after the invoice date.
Enforcement also spends credibility. Procurement teams that are experienced primarily as an obstacle lose the informal influence that actually moves behaviour, and they stop hearing about problems early enough to fix them.
What works
The principle is to make the compliant path the fastest path, so that following it is the route of least resistance rather than an act of discipline.
Close the coverage gaps, and use the free-text requisitions to find them. Every free-text line is a requester telling you, in their own words, what the catalog should have contained. It is the best available backlog for catalog enablement and it costs nothing to produce.
Write descriptions in the language buyers use. Normalise supplier content into searchable terms, including the common trade names and the words people actually type. Search success rate matters more to adoption than catalog size.
Set approval thresholds proportionate to value. Controls that cost more than the exposure they mitigate are not controls. Low-value purchases from contracted suppliers at contracted prices have already been approved in substance, and re-approving them individually adds risk rather than removing it.
Use PunchOut where hosting cannot keep up. For assortments too large or too volatile to maintain as a hosted file, PunchOut puts the supplier’s live catalog in front of the requester while still returning a structured, priced, contract-referenced line to the purchasing system.
Plan for urgency instead of pretending it away. If a category genuinely has same-day needs, contract a second supplier for that scenario. An unsupported emergency route guarantees leakage in precisely the moments when it is least likely to be reported.
Show the contract at the point of need. The requester should see that a contract covers what they are buying inside the requisition screen, not inside a repository they have no reason to open.
The measurement problem
Acting on maverick spend requires detecting it, and detecting it requires knowing, per transaction, what was bought, from whom, and whether an agreement covered it. That means classified spend data with contract references attached.
Maverick spend is precisely the spend that arrives unclassified and unreferenced. The thing being measured destroys the instrument that would measure it. Two consequences follow: any figure for maverick spend is an understatement, and the understatement is biased, because the least visible purchases are the ones furthest outside the process.
The practical answer is to measure the visible proxies instead. The share of requisition lines sourced from a catalog rather than typed as free text. The share of invoices arriving against a pre-existing purchase order, with the order date checked against the invoice date. The number of distinct active suppliers in categories where a contract exists. None of these is maverick spend exactly. All of them move in the right direction when the underlying problem improves, all are measurable from the first day, and none can be improved by writing a memo.
Where programmes go wrong
Compliance is measured on spend value rather than transaction count. A handful of large on-contract orders can carry the value-weighted figure to a healthy number while most individual transactions bypass the process entirely. The reported percentage improves and the operational reality does not.
Retrospective purchase orders count as compliant. Covered above, and worth auditing before any other number is trusted.
Catalog coverage is measured by line count. Two hundred thousand lines loaded is not coverage if the twenty most frequent searches return nothing. Measure hit rate against real search logs, not volume.
Blocking precedes coverage. Enforcement applied before the compliant path can absorb the demand converts visible non-compliance into invisible non-compliance, which is strictly worse.
The measure becomes the goal. When compliance is enforced hard and coverage is poor, requesters learn to buy the nearest catalog line rather than the item they needed. The compliance metric is satisfied, the wrong thing is purchased, and the cost surfaces somewhere no procurement report will ever show it.
Maverick spend is the visible symptom of a catalog that does not cover what people buy, does not describe it in language they recognise, or cannot be reached quickly enough. That is a content and integration problem before it is a policy one. SupplierForge addresses that side of it: ingesting supplier content through whatever channel a supplier can manage, validating and classifying it centrally, and delivering it into the buying systems through native PunchOut and cXML, so the compliant path is also the quickest one in front of the requester.