AI & An Airline Startup - Part Two
Entrepreneur Bob Branston is using artificial intelligence to launch Vestal Atlantic – find out how Bob & his team prepare the airline for launch
Yesterday we met Bob Branston, an imaginary entrepreneur building an airline in Canada called Vestal Atlantic.
We found out how Bob thinks that his success against incumbents Air Transat and startup jetlines, Lynx, and Flair will depend upon delighting consumers at every stage of the passenger journey. Bob reckons that using machine learning and artificial intelligence to design the airline’s commercial products will allow him to compete successfully.
His goal is simple - to earn as much revenue as possible from four aircraft serving three airports over two years. He has built his team and is ready to proceed. What does Bob do next? Read on to find out…
(Click here for the original article)
Step 4: Bob defines a data strategy
With the passenger journey clearly understood, the people on Bob’s team considering digital media, consumer analytics, commercial management and data science set out a data strategy. This had to consider every step of the passenger journey and imagine what types of data can make the people buying Bob’s tickets happy at every stage.
Some of the data that Bob uses will also be used by jetlines and Air Transat – information like how many people travel between cities historically and how much they pay for tickets. Other data, like social media and lifestyle chatter, credit card spend and mobile phone movements (see article), traveller interest, income and demographics, plus digital assistants are much more innovative.
Step 5: Bob designs a neural network
At this stage Bob starts taking decisions about which commercial models Vestal Atlantic should use.
Instead of a traditional seat-sale model they might use subscriptions (see article) or offer flights for free while earning commission on other travel products (see article).
Bob needs to decide where to offer premium economy (see this article and this article), lounges (see article) and how to price cabin upgrades (see this article and this article).
The data strategy should help Bob make his decision about the right commercial models for Vestal Atlantic. To conduct the necessary analysis, Bob’s team build a type of model called a neural network, which seeks to learn about the relationships between different sources that humans on their own cannot see to think through a solution to a problem. Like people, ‘but’ faster, smarter, and now, not later.
The fundamental question that Bob’s neural network has to answer is simple – given everything that Vestal Atlantic’s team knows about spend, interest, travel habits, location and disposable income, how likely is it that specific consumer groups will consider travelling with Vestal Atlantic at different price points?
Machine learning tools like Bob’s use deep learning – that means multiple layers of analysis combining links between different variables simultaneously. The deep learning process is not run in isolation – it needs to be trained using rules that show the machine how to use, process and pass on information.
Step 6: Bob crunches the numbers to make predictions
Vestal Atlantic’s neural network model is now up and running, giving streaming insights and predictions about consumer buying behaviour in real time.
Traditional airline processes have fares set on the basis of demand to come (see this article and this article) for a flight that has been scheduled independently of the demand forecasting process. Dynamic pricing (see this article and this article) does not correct this fundamental disjoint in the network planning process.
With Bob’s approach, the machine can solve the puzzle about where and when flights should operate, and how much Vestal Atlantic should charge, at the same time. In addition to the standard airline metrics (see article) there are four simple key performance indicators that Vestal Atlantic’s team will keep their eyes on:
1. Customer problem solving score (CPSS) - not a net promotor score or (NPS)
2. Aircraft utilisation per day & per week
3. Number of estimated passengers per flight
4. Average revenue and profit per flight and per aircraft
The operations planning process, which prepares flights for dispatch, rosters crew and other accommodation planning and so is broadly unchanged. In the future though, once his workforce base is more established, Bob may introduce some form of variable crew rostering to put more crew onto flights that are busy to ensure passengers always get a good service.
Step 7: Bob develops automation protocols
The automation protocols are designed to make life easy for everyone. The staff are happy because automation takes away much of the lame repetition and heavy lifting regarding offer generation and distribution, flight planning and dispatch.
The passengers and the buyers are happy because they make it easy to navigate the process of travel and when value-added services are presented these are more relevant and valuable to specific users, making the purchase more surprise-and-delight rather than pay-to-avoid-pain.
The artificial intelligence that Bob installs behind the automation will do five things:
1. Drive traffic to Vestal Atlantic’s website & app
2. Distribute customer-centric & contextual offers
3. Dynamically price travel products based on the value and loyalty characteristics specific to individual buyers
4. Automate all booking-triggered activity, such as proof of originally contract and proof of changes*
5. Continuous improvement, powered by machine learning.
* Contractual elements are likely to be handled through a blockchain process (see this article and this article)
Step 8: Bob launches his airline, learns what he got wrong & improves as he learns
Vestal Atlantic is now ready to go. The data and analytical processes involved are quite different to what a normal airline used in 2020, but more and more airlines are likely to adopt these methods in the next ten years. As Vestal Atlantic evolves AI will not just be limited to the commercial product, it will define the organisation itself too (see article).
In years gone by airline entrepreneurs might ask people at parties whether they have thought about buying tickets yet. Bob does not need to – he already knows…
Would you like to use artificial intelligence and machine learning to help you develop your business? Get in touch.
ricardo DOT pillon AT millavia DOT com, author
oliver AT ransonpricing DOT com, editor
Are you enjoying Airline Revenue Economics? Why not check out some thought-provoking articles from the archive:
For even more mind-stretching analysis, buy a Club Class subscription and enjoy:
Iceland’s Banks – the Nordic island’s airline may have the world’s best-designed network
 | A guest post by
|