airline analytics for the 21st century
TRANSCRIPT
Amazon “the world’s most customer centric company” is exploring delivering products to dispatch centre
before they are bought to reduce delivery time … while airline still plan based on last year data
In deciding what to ship, Amazon said it may consider previous orders, product searches, wish lists,
shopping-cart contents, returns and even how long an Internet user’s cursor hovers over an item.
Airlines plan based on estimated traffic between airports flown last year
Challenging current airline planning with 21st century data capabilities
• 21st century data capabilities
Planning data available to airlines is collected at the booking and check-in; the overall shopping and travel
experience is virtually ignored
Inspiration ShoppingBooking / Ticketing
Check-in BoardingTo the airport Flight
DisembarkmentBaggage
ClaimFrom the airport
At the destination To the airport Check-in Boarding Flight
DisembarkmentBaggage
ClaimFrom the airport
Feedback: complaints,
reviewCompensations
Changes Cancelations
Traditional data Traditional data
Initial data extension suggestion for the purpose of this presentation
Shopping data is big data, but it is fairly “structured” and could be immediately available to most airlines
Booking
Ticket
Shopping
MIDT
BSP
Search
BoardingDCS
Challenging current airline planning with 21st
century data capabilities
• The search data, studying behaviour before the PNR is created
Most searched destinations are not always the most travelled since conversion depends on service and price
• The volume of searches is closer to the actual demand than the traffic itself
• Traffic is a subject to availability, price and convenience and as a result the destination where people fly are not always where they wanted to travel
• To matching the offer with the demand, travel supplier need to understand the unconstrained demand and not focus on traffic from last year
• When rankings match perfectly the correlation factor reaches 1; in large cities such as NYC, LON and SIN the correlation is above average (NYC:0.99, LON:0.99, SIN: 0.995)
Source: Travel Insight
Destination travelled*
NYC
BCN
DXB
BKK
AMS
ORL
AGP
ROM
DUB
TCI
Destination searched
NYC
BCN
BKK
AGP
DXB
AMS
ORL
ROM
PAR
DUB
From LON
Destination travelled*
LON
MIA
ORL
LAX
CHI
SFO
YTO
PAR
FLL
CUN
Destination searched
LON
MIA
LAX
ORL
SFO
CHI
YTO
PAR
CUN
LAS
Destination searched
BKK
DPS
HKG
TPE
TYO
KUL
SEL
LON
HKT
MNL
Destination travelled*
BKK
DPS
HKG
TPE
SEL
KUL
TYO
HKT
LON
MNL
From NYC From SIN
Note: based on the exit
In the top 10 travelled but lower in rank
In the top 10 travelled and higher in rank
Not in the top 10 travelled
Not in the top 10 searched but in the top 10 travelled
12th searched
11th searched
There are significant differences in how much the actual demand is satisfied among large metropolitan areas and an overall degradation over time
Source: Travel Insight
Correlation searches/exits
top 50 top 50-100 Linear (top 50) Linear (top 50-100)
Challenging current airline planning with 21st
century data capabilities
• The geolocation, going beyond the PNR
Travelers increasingly chose not to fly from their “home” airport for convenience and price while airline can only see their departure airport and thus misrepresent actual demand
• Leaked traffic is traffic from a city that doesn’t originate nor end in the IATA associated city code (e.g. travellers from Paris travelling from BRU)
• Every city has a home airport (or group of home airports as defined by IATA city codes) where airline assume traffic originates
• Airlines plan their network and revenue management on the basis of this definition and monitor performance “airport-to-airport”. No planning system in the industry is tailored to give insight into the “true” competition
• Users increasingly chose alternate airports to fly which questioned current processes
• Competition between origin/destination airports is widely misunderstood
Source: Travel Insight
0%
5%
10%
15%
20%
25%
30%
35%
40%
% “leaked” traffic from the top 100 cities in Skyscanner user base
Total
As way of example: Traditional O&D based market data allocates incorrectly 31% of the demand from Las
Vegas area to other airports such as San Francisco and Los Angeles
SFO
MNL
BKK
LON
LAX
SEL Las Vegas, NV+North Las Vegas, NV+ Henderson, NV
• Traditional Market data allocate the demand at the first departure airport ignoring the true origin of the travellers
• While not perfect IP based geolocation gives a better view of the true origin of the demand
• IP based geolocation gives a better understanding of where users live and work than mobile networks
Preferred departure airport for our users in Las Vegas, NV, North Las Vegas, NV and Henderson, NV when travelling international
Source: Travel Insight
Challenging current airline planning with 21st
century data capabilities
• Up-to-date decision making
Capacity planning is typically done 10 months in advance with very limited capabilities to adapt to the
current demand of travellers
Travel Searches for Travel in July 2016
• From BER, TCI was the 38th most requested destination for travel in Jul 2015 with a conversion rate of 10%
• Airlines in October and November planned their network for the summer, published and started selling
• For July 2016, TCI was the 24th most requested destination from Berlin (+14 ranks!) but because the demand for TCI came late (started in APR 2016) the capacity was not sufficient and the resulting conversion rate dropped to less than 8%; compared to the 18% conversion last year there is likely 25% of the demand that could not be satisfied
• Next year, on the basis of the traffic, (which was constrained by 25%) the capacity will likely be limited once again