After-Hours and Disruption Servicing: What It Costs a TMC to Run or Outsource
TL;DR: Every TMC covers nights, weekends and IRROPS in one of three ways: an owned 24/7 rota, an outsourced after-hours desk, or a hybrid with AI first-line triage. Fixed cost favors outsourcing at low volume; cost-per-case favors ownership at high volume; hybrid wins when peakiness and language coverage both climb. NDC reissue skill shifts the math again.
Drawing from 8+ years building AI-powered corporate travel platforms, the pattern that holds up across TMCs of every size is this: after-hours and disruption servicing is the hardest unit-economics problem in the business. The ticket itself clears for a few dollars; the exception that pays for the agent's whole night runs five hundred. The three models — in-house rota, outsourced desk, hybrid with AI triage — each break down at a different volume threshold, and most TMCs cross those thresholds without re-pricing their service levels. This article lays out the mechanics so operations leaders can price their own situation.
Travel Code works with TMCs as a supplier and servicing partner (US hotel content, ARC-side servicing, overnight disruption cover sitting under an existing brand), not as a replacement booking tool. The TMC partner page describes that scope. The analysis below is model-agnostic and useful whether you keep servicing in-house, buy it, or blend.
The Three Operating Models
Owned rota. The TMC staffs its own 24/7 desk, typically two or three agents on a graveyard shift plus an on-call escalation path. This model keeps IATA/ARC accountability tightly in-house (per IATA Resolution 850 and ARC Accreditation Minimum Standards, the accredited agent is responsible for every PNR touched on its stock — see our primer on ARC accreditation for TMCs) and preserves customer experience: the same voices who service accounts by day answer the phone at 02:00. It stops making sense below roughly 60,000 transactions a year, because the fixed cost of three graveyard FTE plus a supervisor spreads over too few cases.
Outsourced desk. The TMC contracts a specialist after-hours provider on a per-call or hybrid retainer-plus-case basis. Fixed cost drops; cost-per-case rises; quality control depends almost entirely on the SLA and the escalation discipline. This is the default for TMCs under about 100,000 transactions and the right answer for anyone whose night volume is low but spiky. It stops making sense when corporate accounts start asking who actually answers the phone at 03:00 and the honest answer hurts brand delivery.
Hybrid with AI first line. An AI agent (chat, voice, or both) handles first contact: identifying the traveler, pulling the PNR, classifying the case (seat change, re-accommodation on an IRROPS, lost bag, true emergency), and resolving the categories that can be resolved deterministically. The human desk — owned, outsourced, or both — handles everything above the AI's confidence threshold. Travel Code's work on AI agents for corporate travel points to the categories where deflection is honest: PNR lookup, seat map changes, policy questions, confirmation reissues. Reaccommodation on an IRROPS where the airline hasn't posted a protect schedule still goes to a human.
Cost Drivers That Actually Move the Price
Four variables set after-hours cost structure. First, coverage window: nights-only (18:00-08:00 local) is cheaper than true 24/7/365 because the day team handles weekday peaks. Second, call volume peakiness: a desk sized for the median shift drowns on a major weather night, when industry case rates spike 5-10x above baseline per US DOT Air Travel Consumer Reports. Third, language coverage: every added language roughly doubles the staffing math because specialist agents cannot be shared across shifts the way English agents can. Fourth, first-contact resolution rate (FCR): a desk with 70% FCR costs materially less per case than one at 45% because unresolved cases round-trip to day shift. The GBTA 2025 Business Travel Index puts US business-travel spend at $1.6 trillion with recovery still volatile, so volume forecasts carry wider error bars than they used to, and under-sized desks fail precisely on the nights that matter most.
How NDC Changes the Servicing Equation
NDC changes the servicing equation in two concrete ways. First, reissue and refund mechanics differ by airline because each carrier controls its own NDC implementation; IATA's Modern Airline Retailing roadmap acknowledges that servicing standardization lags offer-and-order progress. A GDS-era reissue could be solved by a competent agent from pattern memory in six to ten minutes. An NDC reissue on a non-standard carrier can take 25-40 minutes and often requires the agent to work in the airline's own servicing portal rather than the GDS. Second, the skill floor rises: refund rules under US DOT's 2024 automatic-refund final rule (effective October 2024) obligate carriers to refund within specific windows, but the agent still has to initiate the request through the correct channel. After-hours desks staffed for a GDS-only workload will under-perform on an NDC-heavy portfolio until retrained, and the retraining cost belongs in the budget.
The SLA Terms That Actually Matter
The SLA terms that actually matter in after-hours contracts are fewer than most RFPs list. Response time (seconds to first live human) is table stakes but only tells you about staffing. First-contact resolution rate is the real quality metric and should be measured by category, not blended, because an IRROPS reaccommodation and a seat change are different problems. Reissue time under disruption (minutes from inbound call to confirmed new PNR) is the operational truth. Escalation path (named roles, phone numbers, maximum hold at each tier) stops a 03:00 case from stalling. Overnight case reporting (a dashboard delivered before the day team logs in, with every opened and unresolved case) prevents handoff loss. Credit and penalty clauses should name specific breach levels. ARC Accreditation Minimum Standards hold the accredited agent accountable for every servicing decision taken on its stock, so SLA scope must match that liability.
Where an AI First Line Genuinely Helps — and Where It Must Hand Off
AI handles the deterministic categories well: PNR lookup, seat-map changes, policy questions, confirmation reissues, status checks on in-progress exchanges. It also handles natural-language intent routing (identifying whether the caller needs re-accommodation, a receipt reissue, or an emergency escalation) more reliably than an IVR tree. It must hand off on: IRROPS re-accommodation when the airline has not posted a protect schedule, medical or security emergencies, duty-of-care escalations (see our duty-of-care hub), ARC exchange decisions that require accredited-agent judgment, and any case where traveler affect indicates distress. The honest deflection rate for a well-tuned AI first line in corporate after-hours servicing is 30-45% of inbound contacts — not the 70-80% sometimes quoted; the balance needs a human.
Comparison: Own Rota vs Outsourced vs Hybrid
| Dimension | Own Rota | Outsourced Desk | Hybrid (AI + Human) |
|---|---|---|---|
| Fixed cost | High (3-5 FTE + supervisor + benefits) | Low (retainer or minimum) | Medium (AI license + reduced headcount) |
| Cost per case | Low at scale | Higher per call; rises on peak nights | Lowest if AI deflects 30%+ of volume |
| Control over customer experience | Full | Limited; depends on SLA | Full on AI tier; SLA-bound on human tier |
| Quality risk | Fatigue, single-shift coverage gaps | Provider staffing changes you don't see | AI over-confidence on edge cases |
| NDC readiness | Depends on in-house training | Varies sharply by provider | Depends on AI model + human tier skill |
| Where it stops making sense | Below ~60,000 transactions/yr | Above ~250,000 transactions/yr or when branded accounts demand transparency | Below ~40,000 transactions/yr (AI license overhead dominates) |
Pricing packages and thresholds described in this article are illustrative, not an industry standard. Actual costs vary by region, labor market, language mix, carrier content mix, and account profile.
Where Travel Code Fits
For TMCs that keep booking in-house but need overflow cover, Travel Code supplies ARC-side servicing, US hotel content, and after-hours capacity under a white-label overlay that sits on the TMC's existing OBT. The partner model is documented at /for-tmcs. The point is simple: a 50-FTE TMC can offer the coverage of a 500-FTE operation without hiring for trough-hour volume.
Frequently Asked Questions
What transaction volume justifies an in-house 24/7 desk?
From the operating math of ARC-accredited TMCs, roughly 60,000 serviced transactions per year is the floor. Below that, three graveyard FTE plus a supervisor spread over too few cases, and cost-per-case stays above sensible outsourced retainers. Above about 250,000 transactions, in-house usually wins on cost-per-case and best controls brand delivery.
What does an outsourced after-hours desk typically cost?
Illustrative only: per-call pricing in the US market tends to sit between $25-45 for straightforward cases and $85-150 for complex reissues, with monthly minimums ranging from $2,500 to $15,000 depending on volume commit. These numbers are not an industry standard; they vary sharply by provider, language coverage, and FCR expectations. Negotiate on category-specific SLA, not blended rates.
How does NDC change the after-hours cost model?
NDC lengthens average servicing time per case because reissue and refund mechanics differ by airline and often require the carrier's own servicing portal. Per IATA's Modern Airline Retailing roadmap, servicing standardization lags behind offer-and-order progress. Budget for retraining and expect your cost per NDC-heavy case to run 2-3x your GDS case cost for the first 6-12 months.
Can AI realistically handle the first call at 03:00?
Yes, for 30-45% of inbound volume: PNR lookups, seat changes, policy questions, status checks on in-progress exchanges. It cannot handle unposted IRROPS re-accommodation, duty-of-care escalations, or ARC exchange decisions that require accredited-agent judgment. The honest answer is deflection and triage, not full automation.
What SLA terms separate a serious outsourced desk from a weak one?
Category-specific first-contact resolution rate, reissue time under disruption, a named escalation path with phone numbers at each tier, overnight case reporting delivered before the day team logs in, and credit/penalty clauses tied to specific breach levels. Response time alone tells you about staffing, not quality.
How should we price the hybrid model internally?
Treat the AI tier as a fixed license cost and the human tier as variable per-case; track AI deflection rate weekly, segmented by case category. If deflection on reissues drops below 20% while deflection on seat changes stays above 60%, the model is working as designed — do not force a blended target.
Sources
- GBTA 2025 Business Travel Index Outlook — gbta.org
- IATA Resolution 850 (Standards for Agents) and Modern Airline Retailing program — iata.org
- ARC Accreditation Minimum Standards — arccorp.com
- US Department of Transportation Air Travel Consumer Reports and 2024 Automatic Refund Final Rule — transportation.gov
