The 2025 Roadmap: Capabilities to Prioritize Now

1. Why This Roadmap Matters in 2025

In 2025 many airlines are in the middle of moving from old “file fares + booking class” processes toward “Offer & Order” thinking. Customers expect clearer products, faster answers, and consistent prices. Internal teams want less friction. Regulators and partners want audit trails. You cannot modernize everything at once. This guide helps you decide what to build next with limited time and budget.

2. Core Problem in Plain Words

You need to show a trip, give a fair price, and confirm it without surprises. The trouble: different systems (cache, host / PSS, dynamic pricer, tax engine, payment, revenue accounting) each hold a different “truth” for a few milliseconds to a few minutes. If they drift you get “phantom seats”, price mismatches, angry customers, and manual fixes that burn staff hours.

3. Guiding Principles (Keep You Out of Trouble)

• Single Source of Truth per Object: Know which system is master for seat counts, prices, tax rules, order state.
• Everything Stamped: Every offer carries version numbers (inventoryVersion, pricingModelVersion, taxTableVersion).
• Fail Closed, Not Random: If validation fails, clearly reprice or explain; do not auto-silent-adjust.
• Measure Before Replace: Shadow new logic before turning off the old logic.
• Short Feedback Loop: Daily dashboard that shows “off” signals fast (divergence, reprice rate, fallbacks).

4. Top 12 Capabilities to Prioritize

5. Detailed Capability Playbook

5.1 Consistent Search & Availability

Problem: Cache says 4 seats left, host says 2. Customer gets error at payment.

Actions:

• Add an inventoryVersion (increment on each seat decrement or release).
• Search response includes: flightId, seatsRemaining (banded or exact per policy), inventoryVersion, freshnessTimestamp.
• Commit call validates the version; if mismatch, re-fetch availability then reprice (explain to user).
• Have a “stale threshold” (e.g. 60 seconds). If older, force refresh from host.

Metric: Availability divergence rate (mismatches / 1000 searches).

5.2 Offer Token & Provenance

Goal: Each offer can be traced: how built, by which model, under what rules.

Token Fields (example): offerId, anchorFareId (if any), pricingMethod=FILED|DYNAMIC|CONTINUOUS, modelVersion, inventoryVersion, expiryUTC, guardrailPolicyId, taxTableVersion, signature.

Commit Flow: Submit token → verify signature + expiry → revalidate version numbers → accept or reprice.

5.3 Dynamic / Continuous Price Band

Why: Close the gap between forecast “best” price and nearest filed fare. Lufthansa Group has publicly discussed “continuous pricing” to fill steps between classical booking classes [1].

Steps:

1. Measure current price gap: average absolute (sold fare - model bid price).
2. Define safe band: min=anchor - X%, max=anchor + Y% (with regulatory & tax checks).
3. Introduce smoothing (do not change more than N times per day for same flight/brand to protect trust).
4. Log every computed price with anchor, adjustments, final.

Metric: Gap reduction (%). Guardrail breach rate.

5.4 Product / Attribute Catalog

Why: Needed for clarity (baggage, changes, seat pitch, lounge) and to move away from only letter codes (RBDs). Qantas and other IATA participants have engaged in Offer & Order pilots that rely on attribute expression [2].

Actions:

• List current brands → break into atomic attributes.
• Create table: attributeCode, description, capacityImpact(Y/N), pricingImpact(Y/N).
• Map current RBD → attribute set (versioned mapping). Store mappingVersion on offers.
• Expose a /catalog API for consistent use across web, NDC, mobile.

Metric: Attribute coverage (% of offers built using catalog vs static text).

5.5 Event Stream (Inventory & Orders)

Why: Near-real-time updates reduce stale caches and manual reconciliation.

Events: SeatDecrement, SeatRelease, OrderCreated, OrderModified, RefundIssued.

Actions:

• Implement ordered (sequence) event IDs per flightLeg.
• Gap detection (missing sequence) triggers snapshot pull.
• Store lastAppliedSequence in projection; expose via status endpoint for monitoring.

5.6 Observability & Correlation IDs

Actions:

• Generate a root x-correlation-id at first request; pass through all services.
• Attach correlation ID to log lines, metrics, traces.
• Dashboards: p50/p95 latency; error causes by stage (pricing, tax, payment).

Metric: Mean Time To Identify (MTTI) incident root cause (minutes).

5.7 Fallback & Degrade Modes

Example: If dynamic pricing API latency > 400 ms, fall back to cached discrete fare and flag “fallbackUsed=true” for analytics.

Actions:

• Define performance budgets per component.
• Configure circuit breaker thresholds (error %, consecutive timeouts).
• Provide safe static price anchor list for emergency mode.

5.8 Servicing Automation (Change / Refund)

Why: Manual change and refund work loads call centers and delays revenue clarity. Airlines like American Airlines and United Airlines have publicly emphasized digital self-service expansion [3][4].

Actions:

• Expose /order/{id}/price-delta endpoint to simulate change before commit.
• Store penalty and residual value logic in a single rules service (not duplicated in UI).
• Emit OrderModified events with before/after components for accounting.

Metric: % of changes fully self-service; average handling time saved.

5.9 Security (Signed Offer / Anti-Replay)

Actions:

• Sign offer token payload (HMAC or asymmetric). Include expiry (e.g. 15 minutes).
• On commit, check: signature valid, not expired, versions unchanged (or reprice).
• Store minimal hash of token to detect replay attempts.

Metric: Rejected replay attempts (should be near zero).

5.10 Interline / Partner Bridge

Why: Partners still expect filed fare + RBD. Air France-KLM, Lufthansa Group, and others continue to use hybrid models while adopting more dynamic elements [1][5].

Actions:

• Offer includes: anchorFareCode + adjustmentAmount (positive or negative).
• Sector proration: allocate adjustment proportionally by distance or revenue share rule; log method used.
• If partner cannot consume token, send filed anchor fallback within allowed tolerance.

Metric: Interline rejection / dispute rate.

5.11 Data Quality & Governance

Actions:

• Nightly job: rebuild “expected state” vs actual (inventory counts, order totals).
• Publish “data quality score” (e.g. 99.7%) to leadership dashboard.
• Define single owner per dataset (no shared orphan data).

5.12 Experimentation / A/B Harness

Actions:

• Route a % of eligible searches to treatment (e.g. dynamic band). Control group remains legacy pricing.
• Measure conversion, revenue per seat, refund rate difference.
• Stop test early only if clear statistical signal (avoid false wins).

Metric: Experiments concluded per quarter; average time to decision.

6. Example End-to-End Flow (Search → Commit)

7. Real Airline Public Examples (With References)

• Lufthansa Group: Has publicly described using “continuous pricing” to reduce gaps between filed fare steps and to broaden offer granularity in NDC channels [1].
• Air France-KLM: Has discussed dynamic and continuous pricing experimentation in industry forums (e.g. ATPCO / conference sessions) [5].
• American Airlines: Publicly pushed increased NDC adoption and emphasized modern retail offer distribution changes in 2023–2024, shifting some content earlier to modern channels [3].
• United Airlines: Public communications on expanding digital self-service for changes / refunds (reducing call volume) support the “servicing automation” priority [4].
• Qantas & Other IATA Pilots: Engaged in IATA Offer & Order pilots (ONE Order / modernization), highlighting attribute-based retail direction [2].

All examples reference public statements and conference or press materials. See References section for source domains. Verify any claim before internal reporting.

8. First 90 Days Action Plan

9. KPIs & Simple Formulas

• Availability Divergence % = (mismatched cases / total availability checks) * 100.
• Price Gap = avg |soldPrice - modelBidPrice| (per seat).
• Reprice Rate = (commits requiring reprice / total commits) * 100.
• Fallback Rate = (requests hitting legacy fallback / total requests) * 100.
• Self-Service Change % = (self-service changes / all changes) * 100.
• Experiment Uplift = (treatment revenue per seat - control) / control * 100.
• Data Quality Score = 100 - ((drift incidents * penalty) / normalization factor).

10. Key Risks & Lightweight Mitigations

11. Quick Checklist (Print & Stick)

• Master list: who owns capacity, price, taxes, order state.
• Offer token includes versions + expiry + signature.
• inventoryVersion validated at commit (no silent auto-adjust).
• Dynamic band guardrails live + logged.
• Attribute catalog + RBD mapping versioned.
• SeatDecrement and OrderCreated events sequenced and monitored.
• Fallback thresholds documented and tested (game day).
• Change / refund delta quote API working.
• Dashboard: divergence, gap, reprice, fallback, yield variance.
• A/B harness capturing uplift with control holdout.
• Signed token verification pipeline (reject replay).
• Risk register reviewed monthly.

12. Plain Glossary

• Availability Divergence: When cached view and host view differ beyond tolerance.
• Attribute Catalog: Structured list of product features (bags, changes, seat pitch).
• Band (Pricing): Allowed range around an anchor fare where dynamic price can move.
• Guardrail: Rule preventing unsafe or non-compliant price output.
• Offer Token: Signed package describing how a price was built and its validity window.
• Provenance: Recorded details of “who/what/when” for data creation.
• Replay Attack: Attempt to reuse a past valid token after it should be invalid.
• Shadow Mode: Run new system silently to compare with live system.

13. References & Notes

• [1] Lufthansa Group public conference and ATPCO / industry presentations referencing continuous pricing (e.g. ATPCO Elevate events, public panel discussions). Source domains: atpco.net, lufthansa-group.com.
• [2] IATA Offer & Order / ONE Order pilot participant communications (IATA public program materials). Source domain: iata.org.
• [3] American Airlines public newsroom / distribution updates discussing NDC adoption and channel strategy (2023–2024). Source domain: news.aa.com.
• [4] United Airlines public digital self-service and customer experience improvements (press releases / investor communications). Source domain: united.com / ir.united.com.
• [5] Air France-KLM mentions of dynamic / continuous pricing in public industry events (e.g., ATPCO, conference sessions). Domains: corporate.airfrance.com, klm.com, atpco.net.
• [6] ATPCO public material on dynamic pricing concepts and fare filing constructs. Domain: atpco.net.
• [7] IATA documentation on Offer & Order architecture (NDC, ONE Order). Domain: iata.org.

Back to all blogs