前言

        将航空票价（Fare）组织成商业上可识别的产品（Product），并提供与品牌相关的服务（Service）。通常，有多种产品可供选择（Offer），并与价格和服务层次或级别（Tier）有关。

『 品牌框架』：航空公司可以根据自己品牌战略，构建1个或多个品牌计划。

『 品牌组织』：航空公司可以按照市场细分、客户细分等Key Factors ，建立品牌计划。

『 品牌定义』：同一个品牌计划下，可定义N个品牌，通过Tier区分品牌之间级别的高低，通过并通过Fare Identification组件关联到特定的Fare。

『 Fare Identification 』：允许指定哪些Fares属应用于当前品牌，品牌计划下的每一个Brand都必须和特定的Fare关联。

『 Feature Identification 』：允许定义当前的品牌计划所包含那些 Ancillary Service(Optional Service) ，

以及当前品牌计划下的每一个品牌对于每一种服务Offered的描述 。

附加服务

        附加服务按行业标准分为5大类，其中行李细分为5类；按特点则分为18种分组；APTCO目前为止已发布标准服务为600多种，供航司选择，并定义自己的服务/产品库。e.g.  060/UP-Upgrades/ECONOMY TO BUSINESS

附加服务规则引擎框架 

        附加服务的规则框架，覆盖航司附加服务各种粒度的业务需求,为附加服务Pricing Engine提供一个强大的弹性报价的机制，组成规则模板的核心要素如下所示。

品牌矩阵

精准营销附加服务

品牌运价结构

一般步骤

成功要素

Upsell优化 – 结构因素

Upsell优化 – 竞争环境

设计及决策难题

在实际设计及实施过程中，由于无法真实或实时的对接真正的顾客实体，同时其真实诉求是不断清晰和变化的。所以将遭遇各种难题。则需要基于现有的数据及素材，使用相关的模型去模拟及度量，辅助人脑的工作。

模型优化

        往往需要运筹优化及科学模型和算法去应对数据缺失和失真，结果的验证和模拟，同时用于数据的分析，满足实时高速计算的场景要求。

北美航司曾经的案例

U.S. Airline Cases

Leo: 即使是简单的免费/收费行李托运，都必须结合市场。不一定单笔交易盈利就代表整体盈利，反而要从市场占有率和竞争对手的对比中评判。

Frontier Airlines: the first carrier to offer branded fares in the U.S.

Frontier Airlines launched branded fares in the U.S. in 2008 soon after the industry adopted bag fees. Frontier initially matched the industry bag fees – when purchased à la carte – but offered a large discount when purchased in the first upsell branded fare.  It did protect market share versus Southwest Airlines which had chosen not to assess a bag fee at all. Southwest continued to grow aggressively in Frontier’s Denver markets but Frontier didn’t lose as much share as it would have without branded fares.

JetBlue: one of the more recent advocates of branded fares

JetBlue faced a similar competitive situation with bag fees.  They launched branded fares – with a discount on bag fees and at the same time they began levying a bag charge. They sought more revenue from the bag fees, but didn’t want to lose share because of their new fee.

Delta Airlines: add a new fare to their branded fare menu, to compete more effectively against competitors

Delta’s “Spirit match” exists with Delta’s Basic Economy fare – a lower fare with fewer amenities offer in ultra-low-cost-carriers like Spirit and Frontier，with much lower fares but accompany those fares with more ancillary fees. The un-bundling is designed explicitly for market share protection in markets where Spirit/Frontier have successfully “stolen” share.

U.S. Airways Experience.

Leo:在特定航线有极高的市场份额和利润。大量的策略分割或防止商务旅行者购买低票价的休闲旅行票。当市场和限制消失后，区区5美金，LCC就能够抢夺巨大的休息消费群体，以及甚至抢夺原本的商务旅客

US Airways: With the impact of an LCC and restriction free pricing

Previously, US Airways enjoyed an extremely high market share and profit on Washington DC and New York to Florida markets. The legacy formula of a 21, 14, 7, 3 day advance purchase, round trip requirement, Saturday night stay and a single, restriction free, one way fare worked very well at segmenting, or preventing, business travelers from purchasing the lower fares in place for leisure travel

The introduction of LCCs, with one way fares, no restrictions - and no frills like seat assignments, lounges, meals, F class, free bags and distribution only through web sites designed to support the add-on pricing of such options left US Airways unable to compete

The markets were highly leisure with enough high demand business travel - as soon as restrictions were gone the leisure travel all moved quickly to the LCCs

Experimented and found that leisure travelers would shift carriers for less than $5 difference.  With no mechanisms on internal web sites, and especially none in the GDS that were highly dependent on, US Airways had to lower the fully loaded product to LCC prices.  It took years to get the technology in place to allow for unbundling and the fare families that the major airlines have in place today - and most can compete quite effectively with the LCCs and simultaneously provide the type of product business travelers demand.

Advices from U.S. Airline Implementations.

Leo:充分考虑分销策略，渠道，细分结果，技术等等因素。评估和确信内部能力足够支持品牌策略。周全照顾由于附加服务的绑定变化造成的其他影响。往往客户分群，市场细分会得出较多的分群，但是真正高价值或高消费群体的就较少的3/4个。简单明了化，不仅仅是产品，更关键需要考虑参与和使用的代理，机场，以及航司内部人员。

Think carefully before take action:

Recommended distribution strategy based on segmentation results, distribution mix, distribution channel/partner, and technical capabilities - all points of sale, internal and intermediaries

Assess internal distribution (reservations, ATO/CTO, brand web sites, kiosks) booking/pricing logic to ensure technical and business processes support branded fare strategy, including change recommendations

Customer Service / customer facing functions process review to identify impacts of branded fare implementation (eg. if seat assignments not included how are families with small children handled at the gate, policy/procedure/remediation for equipment change, agent/in-flight service awareness and enforcement of carry-on bag policy/fee collection, etc.)

Based on working experiences from Carlson & Accenture, fare groupings of 3 or 4 is likely the most needed to serve the major customer segments and would support what you're proposing.

Some specific cases from the US market: United Airlines recent roll-back of their Basic Economy highlights two risks of not planning the implementation well enough.  Their product was early to market and the confusion across their fare families drove business travelers away as agencies booked to competitors that were easy to understand and manage through the GDS.  To make things worse, the operational execution made things difficult for flight attendants and gate agents to enforce extreme carry-on bag restrictions and limited seat assignments.

Delivery Reference

Leo: 客户分群首先可采用历史行程或PNR数据。基因算法的本质。选择简单市场的休闲旅行，建立模型后投放到其他市场，然后不断优化。multinomial logit （行为预测的模型，用于分类的机器学习模型）

The input to the customer segmentation will be historical trips / PNRs. We can divide them up, using data in them and heuristics, to determine which are likely leisure travelers / business travelers, as well as long-haul / short-haul, etc.  While we’re doing the product design, especially if we start with surveys and hand-built models, this will keep it simple.

Can try a couple of different segmentation to create the branded fares, but would not try to generate a lot of different options. The idea here is that we can learn about short-haul leisure travel preferences, and then use it to design a branded fare product across many markets. One market may be 20% leisure, another might be 60%, and we just adjust the mix. When we have logs and actual data, after implementation, we can then use machine learning at an O&D level and will have less worries about segmentation - or might not even use it.

Modeling the customer purchase probability

Prefer to use multinomial logit choice, which has a fairly straightforward derivation. Each feature of the brand has a parameter to describe the attractiveness, and than look at the overall probability of purchasing each brand. meanwhile the parameters are the logarithm of the odds ratio, which can be described to experts or even calibrated on expert or survey input. i.e. ask questions like, “What price premium will result in 50/50 upset to get an assigned seat?”

Similar to the Frat5 curves used by PROS and others to look at upset.

总结

        Merchandising中关键的一个环节，在目标市场及客户群体细分的基础之上，优势化自身航司的产品和服务，关键在于产品的差异性和菜单服务方式投放，但是并不是差异性的投放和不一致报价。