Airline managers still rely on decades old methodologies, data and tools that will not be relevant in the new world of distribution which raises the question:
Who will win the next round?
1. Faical Allou – Business Development
faical.allou@Skyscanner.net
February 2017
How technology influences airline planning
2. Executive Summary
• The way people purchase tickets drives how managers run airlines
• In the GDS-only world, airlines built hubs and reduced connecting time to appear on the first page
• Since the mid 90s, in a “sorted by price” world, airlines have cut on the service level to offer the
lowest price possible
• The next iteration will be messaging and voice which are fundamentally different ways to
purchase:
o A very limited number of options can be presented
o The sorting will be based on an aggregation of Schedule, Price and Service
• Airline managers still rely on decades old methodologies, data and tools that will not be relevant in
the new world of distribution which raises the question:
How to win the next round?
3. The Best Place to Hide a Dead Body is on Page 2 of Google…
unknown author
… or any other search/comparison engine
“WYSIATI” What you see is all there is
Daniel Kahneman, Think Fast and Think Slow, 2011
Airlines need to plan their schedule, services and prices to stand out in the
next generation of distribution because people can’t buy what they can’t see
4. The GDS display has driven airline network planning for decades pushing
airlines to reduce connecting times at their hubs
GDS “Neutral Availability Display”
Up to 8 results Airport to Airport ranked by duration
- Moving to the next page 2 involved typing code for the
agent
- For the user, it involved asking the agent to type code
- Duration is easy to calculate and “seems” natural but
certainly not “neutral”
- Create hubs to offer as
many Airport to Airport
combinations as possible
- Influence agencies to bias
the results
- Code share to “hog” the
first page
- Reduce connecting time to
the (unsustainable?) “limit”
(e.g. 45min in CDG)
... and “adjust” the block
times
What should we
do to be on the
first page?
!
5. Search can be made from everywhere to everywhere, include “nearby”
airports, can be started from a map etc etc
Result page can display up 10 results above the fold (depending on screen size)
ranked by price by default
- Seeing hundreds more results involves only scrolling, or clicking on a link
- Many portals are now able to search by budget / from airport to
everywhere etc …
- Reduce prices ideally to be
1€/£/$ cheaper than
competitors
- Cut costs to sustain low fares
- Fly to alternate airports
- Code shares still work
The subsequent web-based distribution channels were price-driven and
provided the option to show hundreds of results on desktops
What should we
do to be on the
first page?
!
6. How many airline planners know what users filter for?
2,100,000
1,200,000
510,000 490,000
Nb of stops Departure time Airline Duration
Number of interaction with filters over a period that generated ~4m click-through
Whilst airlines have been focusing on cutting costs and reducing trip times,
they have lost track of what travellers want and optimized on the wrong KPIs
While non-stop vs stops is an actual user preference, duration is a lot less
important which means that 90 min connection is not a lot worse than 45
Since the early 2000s, many airlines have rolled back the tight connecting
times and unpeaked their hubs to avoid misconnections and better use
infrastructures (e.g AA “rolling hub” in DFW) – and now introduce bare bone
seats to be the cheapest
Unconverted
searches are
barely looked at
7. Users are willing to pay more to get more, even for something as trivial as a
better booking experience; they only need to be “nudged”… WYSIATI
5%
20%
Users not chosing the cheapest
old layout new layout
Old layout
New layout
What will happen when we show more of the relevant product attributes?
The new layout forces users to make a conscious decision
8. Flight attributes are currently displayed after the choice is made
• NDC will provide a more accurate representation of the amenities
available which will enable fair product comparisons
• Only a matter of time before they become filters
- Offer what users look for
- Price (or discount) based on the
features users are willing to pay for
… but what is it? How much?
Note: source for amenities is Routehappy API
What should we
do to be on the
first page?
?
… so the next iteration of comparison engine will take into account more
product attributes …
9. 3 options ranked by score = how “good” the option is (currently only duration and
price)
• Users don’t want to be shown hundreds of options to chose from (they just
need to know the tool compares hundreds, afraid to miss out)
• New web technology enables faster iterations and the ranking will reflect actual
user preferences so that optimizing the rank will truly improve travellers
satisfaction
- Would code shares still
work? Need to keep
only relevant options
- Improve “itinerary
score” => How is it
calculated? How does it
influence choice?
What should we
do to be on the
first page?
?
… but the world is now mobile with small screens, limited information and
cumbersome interactions; travellers increasingly rely on recommendations
10. Only 3 choices available, and
need to swipe between them
The new messaging platforms are where users are today; these show even
less options and travellers will rely even more on recommended options
How can airlines stand out in the messaging world?
3bn+ installs
11. How many choices are too many? (2?)
What option is going to be delivered?
What are differentiation factors?
Apple Siri
1Bn+ devices
Google Assistant
1Bn + devices
Cortana (Windows)
1Bn + devices
Facebook M
1Bn+ user
accounts
Amazon Alexa
275M+ amazon user
accounts (with cc info)
The next major wave is AI powered voice interfaces, where users can
probably only listen to one option; what airline will it be?
What should we do
to be on the first
(only) “page”?
?
12. What do we do with what we know?
“If you can't measure it, you can't improve it” - Peter Drucker
“Without data you’re just another person with an opinion” - W. Edwards Deming
13. Airlines need to adjust their product (network, pricing and services) to stand
out but they now need to understand the coming generation
GDS-only era Website/apps Messaging AI & Voice
How to
stand out
We are here
1960 2017
- Create hubs to offer as
many Airport to Airport
combinations as
possible
- Influence agencies to
bias the results
- Code share to “hog”
the first page
- Reduce connecting
time to the
(unsustainable?) “limit”
(e.g. 45min in CDG)
... and “adjust” the
block times
- Code shares still work
- Reduce prices ideally to be
1€/£/$ cheaper than
competitors
- Cut costs to sustain low
fares
- Fly to alternate airports
- Would code shares
still work? Need to
keep only relevant
options
- Improve “itinerary
score” => How is it
calculated? How does
it influence users?
- Offer what users look for
- Price (or discount) based
on the features users are
willing to pay for
… but what is it? How much?
What should
we do to be on
the first (only)
“page”?
?
Mid-1990s
14. People purchase differently and designing the right offer for them requires
access to the relevant information
GDS-only era
Marketing Information
Data Tapes (MIDT) for
PNRs
Website/apps
Structured user interactions
from search to ticket
Messaging
Loosely structured flight
requests in text
AI & Voice
Loosely structured flight
requests as soundData type
Information
Travel booked from
airport to airport
Search, location, filters
(preferences for flights)
User info & preferences (access
to all requests across verticals)
Intonation (mood?) …
We are here
We still plan based on this
1960 Mid-1990s 2017
What are we going to do with this?
For 20 years we have
barely used this rich data
15. Airlines delivering what users wants (and what they are willing or able to pay
for) will win, as long as people can find them “WYSIATI”
GDS-only era Website/apps Messaging AI & Voice
1960 2017
Large Hub and Spoke carriers
won
Low Cost
Carriers are
wining
Who will win this and how?
Mid-1990s