Blog
Expense Intelligence 2025: What Has Remained the Same—and What Will Improve in 2026

Author

Carim Chenna

Date Published

Edi Team

In the end, 2025 was not the year of revolution, but of disillusionment. Expense Intelligence delivered on what the numbers promised—but not always on what organizations expected. The algorithms became more precise, automation rates rose, and the business cases became more robust. And yet, a striking contradiction persisted in many companies: technologically, much was possible; operationally, surprisingly little had actually been fully implemented.

That is the real takeaway from 2025. Not that Expense Intelligence had failed. On the contrary. The market has proven that automated expense processes deliver measurable financial benefits today. But it has also shown that the bottleneck no longer lies in OCR quality or receipt recognition. The bottleneck lies in data models, responsibilities, integrations, and governance.

Anyone who views 2025 merely as just another year of digitization is missing the point. It was the year in which an efficiency issue became a governance issue.

2025 provided the business case

For a long time, the central question on CFOs’ minds was simple: Does Expense Intelligence really save money—or does it just shift costs between employees, shared services, and software budgets?

Based on all the benchmarks and operational data available in 2025, the answer is clearer than it was two years ago. Yes, the financial impact is real. In mature implementations, the market ROI reached up to 376 percent. Automation rates of around 90 percent were achievable in standardized processes—though under conditions that were often overlooked: clean master data, clear travel policies, defined approval workflows, and robust ERP integrations.

This is precisely where, in 2025, the robust cases diverged from the marketing slides.

The best implementations did more than just process receipts faster. They improved three things simultaneously: processing time, error rates, and policy compliance. That is crucial. Because efficiency without governance is of limited value to finance functions. An automated expense reporting system that passes through incorrect cost centers or leaves VAT opportunities untapped is not progress, but a faster way to make mistakes.

Nevertheless, the productivity gains were significant. Companies that integrated the entire end-to-end process—from submission and matching to posting and reimbursement—were able to drastically reduce manual processing efforts. This was particularly true in the travel sector, where credit card transactions, hotel bills, taxi receipts, and meal expense policies had previously been notoriously error-prone. By 2025, it became clear that Expense Intelligence is more than just OCR plus workflow. It became a control system for a cost category that is managed with surprising lack of transparency in many companies.

One aspect that is often overlooked in technology coverage was particularly relevant: the time saved in business units. When sales representatives, consultants, or project managers no longer spend 20 or 30 minutes per trip on follow-up work, corrections, and clarifications, the resulting productivity gain is often underestimated in traditional IT ROI calculations. 2025 has made this effect visible.

The Big Mistake: Automation Does Not Equate to Adoption

And yet, part of the promise remained unfulfilled. Not because of the technology, but because of how it was integrated.

In 2025, many companies learned that high automation rates in pilot phases do not automatically translate into high adoption rates in day-to-day operations. The problem was rarely the app or the recognition technology. The problem was the reality of the processes.

There were, for example, travel expense policies that had evolved over the years but had never been systematically standardized. There were organizations with multiple credit card programs, regional exceptions, and heterogeneous ERP landscapes. And there were employees who, while welcoming mobile submission, still had to manually rework every other exception. The result: friction. And friction is the silent enemy of any financial automation.

Added to this was a cultural factor that many providers underestimated. Expense management isn’t a glamorous topic, but it is a sensitive one. As soon as reimbursements are delayed, receipts are rejected, or rule violations become more visible, process optimization quickly turns into a matter of trust. 2025 has shown that Expense Intelligence gains acceptance when companies create transparency: What rules apply? Which decisions does the system make automatically? Where can a human intervene? Where is an issue escalated?

Where these answers were missing, the software remained a technical layer over old conflicts.

Regulation has made the market more mature—and more complicated

2025 was also a turning point because regulation shifted priorities. With the expansion of CSRD requirements, the travel and expense process suddenly became a key component of sustainability reporting. It was no longer just a footnote.

Because as soon as companies are required to reliably report which emissions result from business travel, hotel choices, or modes of transportation, traditional expense-tracking logic is no longer sufficient. What’s needed then is structured, analyzable data—ideally directly from the booking and not just from a PDF receipt. By 2025, it became clear just how many companies were still operating on a surprisingly thin data foundation. Those who merely track costs but not contextualized travel data can neither report reliably nor manage intelligently.

In the DACH region, a second aspect came into play: data protection and data localization remained real procurement factors. With an eye on the nDSG in Switzerland and the ongoing sensitivity regarding GDPR-compliant processing, architectural questions in 2025 were not an afterthought but part of the procurement review. This did not slow down the market, but it did professionalize it. Vendors had to demonstrate where data is stored, how models are trained, which subprocessors are involved, and what audit capabilities are actually in place.

This made sense. However, it also revealed that part of the market was more advanced technologically than it was in terms of regulatory compliance.

Market consolidation was not just a passing trend

A third finding from 2025 concerns the market itself. Consolidation was no longer something that could be overlooked. Partnerships, platform strategies, and acquisitions have reshaped the question of what companies are actually purchasing: an expense management tool, a travel platform with financial capabilities, or a financial platform with travel functionality.

The closer integration of travel and expense was only logical. The market has realized that true efficiency does not lie in downstream data capture, but in the connection between booking, payment, policy, and reimbursement. This logic became particularly evident in the collaboration between TravelPerk and Yokoy: Integrating travel booking, card flow, and expense control reduces data silos and generates better data.

But here, too, the rule applies: integration does not automatically mean control. Many companies learned in 2025 that suite strategies create new dependencies. While those who commit strongly to a single provider gain speed, they may lose flexibility in ERP, card systems, or regional specialized processes. Consequently, especially among small and medium-sized enterprises, there was no blind enthusiasm for platforms, but rather a sober assessment: Where is depth worthwhile, and where does openness remain more important?

This assessment will become even more relevant in 2026.

The real turning points lay deeper

Four developments brought about structural changes in 2025.

First: the NDC transition in the airline industry. What initially seemed to many like a sales issue for airlines had tangible consequences for travel and expense data. Fare logic, ancillary services, rebookings, and fare families became more nuanced but also more complex. As a result, companies reliant on accurate travel expense analysis needed better data models and more robust interfaces. The quality of Expense Intelligence suddenly depended more heavily on upstream booking data.

Second: AI integration shifted from a feature to an infrastructure issue. 2025 was not the year that “AI” reinvented Expense. It was the year it became clear where it actually adds value: in exception detection, policy mapping, duplicate checking, VAT classification, and anomaly detection. Quality improved noticeably in these areas. Less convincing were many offerings marketed as generative assistance but lacking operational substance. CFOs saw through this quickly.

Third: Embedded finance has gained prominence. Virtual cards, automatic matching, and preconfigured payment flows have elegantly circumvented part of the old expense problem by 2025: Payments made directly in the right context do not need to be painstakingly reconstructed later. The best receipt is often the transaction itself, enriched with the right metadata. This is not a minor technical detail, but likely the most important lever for 2026.

Fourth: The CSRD requirement has brought expense management out of the back office. Since 2025 at the latest, it has been clear that travel and expense data is not only relevant for billing but also for management—for costs, compliance, and sustainability alike.

What will be better in 2026?

The short answer is: less data entry, more control.

2026 won’t be the year expenses disappear entirely. But it will be the year the core manual process shrinks further. The next step forward is called Autonomous Expense—not as a grandiose vision, but as a pragmatic evolution. This refers to a process in which expenses are classified, verified, allocated, and largely finalized at the moment they occur, rather than being reconstructed later.

Three trends support this.

First, data maturity is increasing. By 2025, companies have learned that reliable automation is impossible without well-maintained supplier, card, cost center, and travel policy data. In 2026, the best teams will therefore invest not primarily in more features, but in cleaner models and stricter master data management. That may sound unspectacular, but it is actually the biggest driver of performance.

Secondly, integrations are becoming more advanced. The critical improvement lies not in even faster receipt recognition, but in more stable connections between Booking, Payment, HRIS, ERP, and Reporting. The more closed this data cycle becomes, the less an individual employee has to "prove" what is already known to the system. This simultaneously reduces both the administrative effort and the risk of misuse.

Thirdly, intelligence is becoming more useful because it is more tightly guided. By 2026, the more successful systems will present themselves less as open assistants and more as specialized control mechanisms: they will detect anomalies, suggest corrections, document decisions in an audit-proof manner, and improve policy enforcement—all without theatrical AI rhetoric. This is precisely where the value lies for finance departments.

The regulatory context is also becoming more pragmatic in 2026. Companies will increasingly use travel and expense data for CSRD-related reporting, but with more realistic expectations. Not every emission figure will be perfect at the touch of a button. The decisive factor is that data depth is increasing and the methodology remains transparent and traceable.

The conclusion for CFOs is uncomfortably simple.

Expense Intelligence is no longer a matter of tools. It is an operating model.

Anyone who wants better results in 2026 should ask three questions. First: Is our spending data generated where the expenditure happens — or only weeks later during rework? Second: Are our rules clear enough that a system can reliably apply them? Third: Do we have the courage to actively reduce exceptions instead of just managing them more conveniently via digital means?

2025 has shown what is possible: up to 376 percent ROI, around 90 percent automation, significantly less manual processing, and better compliance. But 2025 has also shown what these figures hide: without discipline in data, processes, and governance, Expense Intelligence remains a good-looking frontend for operational disorder.

The good news for 2026 is therefore not technological. It is organizational. The tools are ready. Now, companies must decide whether they want to continue documenting expenses — or finally start controlling them.


Subscribe to the Edi Newsletter in German

The German Newsletter keeps you up to date about Edi: simply register using your e-mail address and never miss an article. Of course, you can unsubscribe at any time.


Expense Intelligence 2026: What has remained the same and what is set to improve