S16.4 · Corporate & Business Travel Management

Corporate Booking & Expense Software

A sub-segment of Corporate & Business Travel Management. Market structure, and how AI is reshaping it.

S16.4.2

What is on this page. Market structure, and how AI is reshaping this segment. Ownership, buyer universes, transaction comparables and deal-timing analysis are maintained privately by El Dorado Capital and are not published.

Corporate Booking & Expense Software is the application layer of corporate travel and expense: online booking tools, policy and approval engines, expense capture, reconciliation and reporting, and integrated T&E suites. SAP Concur, Navan's software product, TravelPerk, Coupa Travel & Expense, Expensify, Emburse, the T&E modules inside Brex and Ramp, Center, Fyle, Mesh, Pleo, Payhawk, Soldo and the booking tools embedded in TMC platforms all sit inside the boundary; human agent servicing sold on a contract, traveler risk services, and the underlying card-issuing and interchange rail sit outside it. The segment owns the application layer on the payment rail, not the rail itself. It also owns the booking action and the expense-capture moment directly — which is exactly why it is more exposed than most enterprise software to an AI agent bypassing the interface entirely, though who captures the commission on an agent-executed booking remains an open industry question.

Market structure

T&E software market-size estimates diverge sharply by scope, and neither of the leading figures holds up as a ceiling. Research and Markets puts the base at $3.77bn in 2025, rising to $6.1bn by 2030 on a 9.7% CAGR; Fortune Business Insights puts it at $4.08bn in 2025, rising to $12.98bn by 2034 on a 13.8% CAGR. Summing disclosed revenue for just five participants — Navan's $765m TTM, Ramp's roughly $1.0-1.5bn run-rate, Expensify's $138m, plus Brex and Coupa's T&E slice — already approaches or exceeds the smaller estimate. The likely explanation is classification: most market-sizing houses count card-led players as "fintech" rather than "T&E software" and undercount accordingly.

Geographically the segment is mature and developed-market-weighted: North America 38.9%, Europe 27.1%, APAC 16.3%, Latin America 11.9%, Middle East/Africa 5.8% (Research and Markets, 2026). SAP Concur remains the largest single incumbent by enterprise seat count, but SAP discloses no standalone Concur revenue or customer count today — Concur is folded into a broader "Intelligent Spend and Business Network" segment, a genuine information gap rather than an estimate.

The split that matters is subscription versus usage-based revenue, because the two carry materially different margins and AI exposure. Navan is the sharpest disclosed example: 90% of FY2025-26 revenue is usage-based, mostly a 7% take rate on travel bookings, against only 10% from SaaS subscription — and the SaaS line is growing faster (52% year over year) than the usage line (32%). Gross margin reached 71% LTM, up from 60% in FY2024, still short of pure-SaaS norms above 80% because usage revenue carries network and processing cost; net revenue retention runs above 110%. Switching cost bifurcates by tier: SAP, Oracle and Workday ERP/HRIS integration locks in the enterprise base, while the SMB tier switches cheaply and shows thin CAC discipline — Expensify's multi-year revenue stagnation is the evidence.

How AI is reshaping this segment

Adjacency traffic runs in both directions, and neither direction is a defensible moat by itself. Card issuers are moving into T&E software to own the transaction-capture point and deny it to competitors — "card and expense management should be deeply linked," as one platform's CEO put it (Center CEO, March 2025) — while software platforms expand the other way, into cards and fintech, to capture take-rate economics they would otherwise leak to card-led competitors.

Start with what AI does not threaten. Per-employee seat pricing — used by Concur, Expensify, Pleo and Soldo among others — is a headcount proxy, not a work-volume proxy: a seat is issued to every employee who might travel or expense, independent of how many expense reports AI generates on their behalf. Because the buyer is finance or T&E administration, not the line employee, AI automating receipt capture and reconciliation is a straightforward cost-side gain the buyer keeps; it does not by itself shrink the licensed seat base. The seat becomes unnecessary only if an agent bypasses the booking and expense interface entirely — a separate and more consequential risk.

The two revenue types then read genuinely differently. Navan's usage-based revenue is mostly travel-booking commission — a travel-agency-style take rate that requires no regulated position to earn, which is exactly why automation can compete for it. True interchange revenue fires on a purchase happening rather than on work being done, so its exposure is AI-driven travel-volume compression rather than labor automation — and interchange models are increasingly being absorbed into regulated depository institutions, folding that exposure into a genuinely regulated position rather than a purely competitive one.

At the regulated enterprise tier, the compliance case for a human-governed system wins decisively for now. SOX, audit and expense-substantiation requirements mean CFOs will not let an ungoverned agent book or file outside a system that produces an approvable record, and incumbents are racing to become that policy layer — Ramp has built an AI-agent corporate card, and TripGain has extended a Model Context Protocol server into agentic booking approvals (August 2026). On undifferentiated point-to-point bookings, though, commission capture is genuinely unresolved: who gets paid on an agentic booking is a live, unanswered industry question (Hospitality Net, 2026). That makes the booking-commission line, not the subscription or controls line, the segment's real point of AI exposure. The compliance stack itself mostly favors automation rather than blocking it — SOX audit trails, IRS accountable-plan substantiation and VAT reclaim documentation all require a record to exist, not a human signature, and receipt OCR is already table stakes. The one human checkpoint that survives, European payment-authentication rules requiring human-initiated confirmation at payment execution, sits on the payment rail rather than this segment's core software licensing. On balance, the compliance stack does not protect the T&E software seat from AI bypass.

Elsewhere in Corporate & Business Travel Management