Product - AirDP Flashcards
AirDP supports which external feeds and how
- competition
- pricing model
- SOON target prices
By SFTP
Competition files for SQ DP
O&Ds with list of competitors and OIDs
Manual work on receiving SQ competition file
5md
- Make delta (new file every 2 months)
- Add parameters (stay duration, etc.)
- Handle issues (e.g. non-existing OIDs)
Why do we have OIDs in SQ competition file
To display reference prices in DSH
LHG and Continuous pricing: touchpoints
Only Digital, NOT call centre
Competition: solution to limit perf issues
Pre-computed utilities to avoid loading all competition data live
Engines for DP live and omnichannel
2 different engines (not omni re-use result of live)
Why reshooting
Not for Ludwig GDS bookings (“MIDT bookings”) as we can use shopping logs but we need to remap
CCM model: formulae
MNL: parametric utility based on price P and other attributes Ai:
U=aP+bi.Ai
Proba = e(Ui)/Sum(e(Uj))
Preparation of variables for CCM MNL
Normalisation:
e.g. by distance for price + elapsed flying time
Limitation of single optimisation vs multi-optimisation
Cannot take into account EY alternatives: each EY option is assessed against competition only
and market share of EY is difficult to assess
(still some functional and technical challenges)
How is airline preference estimated
Based on market share
Example of manual model tweaking thanks to weight change
Give more weight to short stay durations for close departure dates as more representative of business behaviour
CCM steps
Weekly estimation
Vetting of new model (vs previous one before calibration)
Calibration
CCM: objective of calibration
Finetune CCM so that:
- DP price close to reference price
- Purchase proba of competitors close to historical market share
Limitation of what omni-channel can do to reach live DP (by design not user controls)
Omni-channel is averaged (RT/OW, business/leisure…)
Problem of DP w/o competition
Can recommend absurd prices (like +100%) because ignores thinks like change of dates
Way of testing DP silent a bit more live
Put avl adj to map DP results (was considered during SQ silent but not done)
Split of business responsibility between the products RMS & DP
RMS+RAAV responsible for avg price, DP plays on the margin
SQ: DP reporting: which features
/DP at end of ticket designator (instead of /SAA)
FCPI H
Impact of Continuous Pricing for AC.com
In shopping, need to get Offer IDs
In pricing, need to call with selected OfferID
AirDP pricing models
1) PB based (if already a significant NDC adoption)
2) Fixed fee + net booking fee
AirDP average price per pex for airlines for external science
2.5 cts
AirDP: try before you buy: pricing model
Upfront fee + fee if a/l doesn’t continue + monthly fee for extension
For BA: 100k + 300k + x
AirDP: try before you buy: duration
7 months: 4 implem/calib + 3 trial. Extension max + 3 months
Air DP: Ludwig vs PROS on functionality
Ludwig allow mark-up and full itinerary optimisation
AirDP: max # of cls that can be closed for EY at time of cutover
4 (as opposed to 2 w/ 1S)
How does DP manage missing competition data (e.g. airline not distributed)
Offline process only based on distributed competitors but corrected by market share (estimated based on MIDT, etc.)
Online process can be input with infare data
Option behind continuous pricing DP and competition DP
Continuous pricing + customer segmentation
