From identified to realized: Turning indirect spend savings into P&L impact

By Timothy Derr, Norman Firchau, Shantanu Verma and Maxwell Gustafson

How AI-enabled spend controls help organizations capture savings, prevent cost creep, and change the decisions that drive spend.

In our experience, identifying savings opportunities in indirect spend is rarely the biggest challenge. The harder part is making sure those savings actually show up in the P&L and stay there. Savings are announced, programs are declared successful and financial targets appear to be met. Over time, however, costs often creep back as the organization returns to its old spending habits.

We often treat this gap between identified savings and P&L impact as an execution problem. But the causes run deeper. Indirect spend reflects how an organization operates: who creates demand, how decisions are made and where complexity generates cost by default.

AI-enabled spend controls can help address those causes. Their value extends beyond finding opportunities or automating reporting. They can connect validated savings to financial plans, protect the resulting spending baseline and influence the everyday decisions that would otherwise recreate demand.

In an engagement with a large industrial manufacturer, we identified, validated and secured more than $150 million in savings opportunities spanning approximately 5,000 cost centers and more than 100 initiatives. The opportunities were supported by supplier commitments or agreement with business stakeholders on the achievability of the reductions. Yet only approximately $30 million to $40 million ultimately flowed through to the P&L. The distinction is critical: securing a saving is not the same as realizing it. Spend leakage and reinvestment absorbed much of the expected financial benefit.

Why indirect spend keeps coming back

The problem is that sourcing savings don't necessarily change the behaviors that created the spend in the first place. Business units continue to make purchasing decisions independently, demand goes largely unquestioned, and unnecessary complexity remains embedded in day-to-day operations.

Indirect spend grows where:

· Ownership of demand is unclear.

· Decisions are made locally and justified in isolation.

Organizations add vendors, systems, reports and processes without consistently questioning whether they are necessary. Adding vendors, tools, reports and external partners isn’t inherently a problem. The problem arises when these additions create unnecessary redundancy, increase complexity or introduce costs without delivering value. These costs often reflect deeper organizational problems. Overlapping responsibilities lead to redundant purchases. Unclear priorities generate unnecessary analysis, while fragmented decision-making drives excessive specifications and services.

Traditional sourcing programs can reduce the price of this activity while leaving its underlying demand intact. Unless the organization changes how that demand is created, the savings remain vulnerable.

Identifying savings is only part of the challenge

Finding savings opportunities is important, but identifying them doesn't guarantee financial results. Organizations need to translate those opportunities into actual reductions in spend and ensure those reductions are sustained over time.

A saving must be validated against an agreed baseline, reflected in budgets and forecasts, and sustained in actual expenditure. In indirect spend, that is difficult because benefits are dispersed across many cost centers and depend on thousands of operational decisions. Finance must distinguish genuine reductions from timing, volume and scope effects. Cost owners experience tracking as an administrative burden. Procurement reconciles spreadsheets long after spending decisions have been made.

The result is a disconnect between reported savings and financial confidence. A negotiated price reduction may appear in a tracker while the budget remains unchanged. A manager then uses the apparent headroom for another purchase. Procurement has delivered a benefit, but the enterprise has not necessarily captured a lower cost base.

We need to distinguish productivity from reinvestment. Validated savings should reduce the relevant spending baseline. Where the business has a compelling use for the released funds, reinvestment should be an explicit subsequent decision with its own rationale and approval. Otherwise, savings become an automatic entitlement to spend somewhere else.

Why traditional controls break down

Fragmented spend exposes the limits of traditional controls. By the time procurement gets involved, many of the decisions that drive cost have already been made. There may still be opportunities to negotiate price, but changing the underlying demand, specifications or scope becomes much more difficult.

Organizations respond with more approvals, reporting and reconciliation. These mechanisms can be necessary, but they often slow decisions without changing the demand behind them. Users learn to work around the process, exceptions become routine and the additional governance creates work of its own.

Organizations often struggle to balance financial discipline with operational flexibility. Adding approval layers can slow decisions and frustrate stakeholders, while removing controls without addressing underlying process weaknesses can allow costs to creep back. A more useful question is how to reduce low-value demand before it becomes a transaction requiring approval.

What needs to change

Closing the realization gap requires a system that continuously translates savings intent into financial outcomes. That system needs shared visibility across spend, clear ownership of each measure and a reliable connection between validated savings, budgets, forecasts and actuals.

It also needs a protected baseline. Once finance validates a saving, the relevant budget or forecast should reflect it, with timing and implementation costs accounted for. Subsequent exceptions should be visible, owned and approved. A control that can be routinely overridden without accountability cannot protect lasting impact.

Tracking savings across multiple cost centers typically requires finance and procurement teams to reconcile spend data, validate savings against agreed baselines, investigate variances and manually update forecasts. This becomes increasingly difficult as the number of transactions and stakeholders grows. AI can reduce some of this work while bringing guidance closer to the decisions that create demand.

How AI-enabled spend controls reduce manual effort and improve oversight

AI-enabled spend controls can make fragmented spend more manageable without adding equivalent layers of manual governance. Automated cleansing, classification and matching can help finance, procurement and cost owners build a shared view of spend. Measures can then be monitored continuously, with exceptions flagged for review rather than every transaction reconciled by hand.

The quality of this view still depends on reliable data, agreed definitions and clear ownership. AI can support validation and documentation; financial accountability remains with the people responsible for the decision.

More significantly, AI can shift the point of intervention upstream. Policy context, existing contracts, utilization data and relevant benchmarks can be surfaced when a request is created, while the decision can still change.

That guidance does not always require another approval. Consider a request for new software. The workflow could identify similar functionality already available, show its utilization and offer the existing solution as the default. A professional-services request could show how its scope and cost compare with genuinely similar engagements. A discretionary purchase could display the revenue needed to offset its cost, using an appropriate contribution margin.

One opportunity worth exploring is the use of behavioral nudges within purchasing workflows. Rather than relying on additional approvals, organizations could present relevant information at the point of decision, such as existing software licenses, available capacity or lower-cost alternatives. The objective would be to help stakeholders make better purchasing decisions without adding unnecessary complexity.

Mandatory policy requirements still need controls. The proposed nudge approach would complement those controls for discretionary decisions; it has not yet been demonstrated in our engagements.

Enforcing savings where they are created

The challenge is determining when intervention is necessary and what information will actually influence a decision. Not every purchase requires another approval or automated prompt. Controls should focus on decisions where there is a meaningful opportunity to avoid unnecessary spend.

Nudges must be specific enough to earn attention. A generic warning that costs matter will quickly become background noise. A peer comparison is useful only when it accounts for differences in scale, geography, responsibilities and service requirements. An apparent duplicate is worth flagging only when the existing solution can meet the actual need.

The system should select interventions based on the category, transaction and decision context. The appropriate intervention depends on the decision. It could be a default option, information about existing capacity or a comparison with lower-cost alternatives. In some cases, no intervention is necessary.

This creates a practical feedback loop: intervene before commitment, observe the decision, and adjust the intervention. Savings protection becomes part of daily work, supported by targeted human review where judgment is required.

Combining nudges, accountability and positive incentives

Controls work best when they influence the decision before a commitment is made, establish consequences for disregarding agreed budgets, and give teams a reason to release funds they no longer need. We would organize the approach around three mutually reinforcing mechanisms: nudge, compliance, and reward.

First, targeted nudges at the requisition stage can surface existing capacity, overlapping subscriptions, relevant peer comparisons or less costly alternatives. They should inform decisions, not shame requesters or create an indiscriminate stream of warnings.

Second, budget compliance needs clear accountability. In a proposed design, material noncompliance could affect performance assessments or compensation, subject to fair baselines, transparent rules and legitimate exceptions. This is a potential policy choice, not a claim about a client implementation.

Third, positive incentives can help counter the familiar instinct to spend remaining budget rather than surrender it. Validated, released savings could flow into a centrally managed reinvestment pool. Teams demonstrating sound budget discipline could become eligible to seek funding for new projects. Budget performance would establish eligibility; the merits of competing proposals would determine allocation. Teams would not automatically receive back the money they saved.

The mechanisms reinforce one another: a nudge improves the immediate choice; compliance establishes expectations; access to future investment creates a reason to participate. But the design must guard against padded budgets, delayed necessary expenditure and apparent underspend caused by external demand changes. Whether the combined approach produces incremental, sustained savings would need to be tested rather than assumed.

Measure nudges like a product

A prompt is not effective simply because it appeared on a screen. We should measure nudges with the discipline we apply to digital products: test the intervention, observe the outcome and refine or retire it based on evidence.

Four KPIs provide a useful starting point:

Dismissal rate. Track the share of prompts dismissed without a change to the request, alongside immediate click-through rates. Persistent dismissal may indicate poor relevance, timing or excessive frequency. Some unchanged requests will be justified, so dismissal alone is not proof of failure.

Time to decision. Measure the time between displaying a prompt and the next meaningful action. Repeatedly dismissing prompts without reviewing the information presented may indicate that stakeholders are ignoring the guidance rather than considering it. Tracking dismissal patterns can help identify prompts that need to be redesigned or eliminated.

Post-nudge spend delta. Compare the original request with the resulting commitment. Track changes in quantity, specification, supplier, use of existing capacity or cancellation. Follow the transaction through completion so a delayed or resubmitted purchase is not mistaken for a saving.

Behavioral change over time. Track whether avoidable requests become less frequent among comparable users and transactions. Are duplicate purchases declining? Are teams selecting existing solutions before a prompt is needed? Normalize for changes in business activity and monitor whether spending moves into other channels.

These metrics help us understand whether nudges are influencing decisions, but they don’t necessarily demonstrate savings. Finance still needs to validate whether changes in purchasing behavior translate into sustained reductions in spend.

To establish incremental value, compare nudged requests with comparable requests that did not receive the intervention, ideally through randomized testing where practical. Test peer comparisons against financial-impact messages, or consistent prompts against message rotation. Assess decision quality and processing time alongside spend, so a lower purchase value does not conceal operational harm or unnecessary delay.

Finance should then validate the resulting benefit and connect it to the spending baseline and actuals. Avoid double counting benefits already attributed to sourcing initiatives. A cancelled request, a lower requisition value and a sustained cost reduction are different stages of evidence, not interchangeable savings claims.

The objective is continuous improvement. Retire low-value prompts, refine effective ones and watch for declining relevance. If behavior improves, fewer eligible requests should need intervention. Simply showing fewer prompts is not evidence that demand has changed.

Case example: Building confidence in realization

In the industrial manufacturer engagement described earlier, more than $150 million of opportunities had been identified, validated and secured across approximately 5,000 cost centers and over 100 initiatives. Only $30 million to $40 million ultimately translated into P&L impact. This was a realization gap, not simply an opportunity-identification gap.

Everyday demand illustrated the challenge: teams purchased overlapping data and services, learning-and-development modules had utilization below 10%, and reporting and project tools duplicated functionality across business units. Better prices alone could not remove the underlying demand.

Leakage and reinvestment must be distinguished. Leakage erodes an expected benefit; reinvestment may be a legitimate business decision, but it should be separately evaluated and approved. A validated saving should be reflected in the appropriate financial baseline, and any subsequent redeployment should be transparent.

The case demonstrates why organizations need stronger links among secured savings, budgets, forecasts and actual expenditure. The nudge, compliance and reward framework above is a proposed extension, not an intervention claimed to have been deployed or measured in this engagement.

A different way to think about savings

Identifying savings is only the starting point. The real challenge is making sure those savings translate into lower costs and don’t disappear over time. That requires clear accountability, effective controls and better decisions about how money is spent.

AI can make those controls easier to implement and maintain, but technology alone won’t deliver the results. Organizations still need to change the processes and behaviors that allow costs to creep back.