intern project - tech
TRANSCRIPT
Intern Project - Final run-through7/28/2016
Team Technology AKA ‘Notorious Data’
Desean
Front End Developer
Zach
FinanceGus
Product Manager Niket
Software EngineerShamanth
Business Analyst
Monetizing Data byIncreasing Conversion through Travel Recommendations
Customers Love Recommendations
40%40 million / 100 million
75%62 million / 83 million
?13 million unique users / month
Sources: Spotify,Statistica, Kissmetrics, Priceline
http://www.slideshare.net/upload
Overview
ResourcesProject Goal Specification
Analyze search and booking data to identify user patterns
and make monetizable recommendations
Air Hotel Rental Car Package Cruise
Retail SOPQ OPQ
PricelineGroup priceline.com
Data: June 21 ,2016 - July 21, 2016Booking records 1.2 Million, Searches records 263 Million
Tony PadovanoChief of Staff for CTO
Dan O’ConnorPrincipal Insights Analyst
Zachary HorneSolutions Architect
Tools
Mentors
Agenda
1. Timeline2. Insights3. Ideas4. Execution/Recommendation
Timeline
2016
Today
Week 1
Milestone 1Planning
Milestone 2SQL Query
Milestone 3Data analysis Milestone
4Findings from dataMilestone
5SlackBotMilestone 6Recommendations
June 20 - June 24Task 1
June 27 - July 1Task 2
July 4 - July 8 Task 3
July 11 - July 15Task 4
July 18 - July 22Task 5
July 25 - July 29Task 6
Week 2 Week 3 Week 4 Week 5 Week 6
1. Understand data 2. Analyze data 3. Visualize data
Insights
© 2016 priceline.com
Data Analysis Methodology
To better understand customer search behavior, we mapped searches to bookings...
How did we map a search to a booking?
1. Site Server ID (Cookie) matched2. Travel Dates +/- 2 Days of check-in and check-out dates3. Area ID/City ID Matched
Data Analysis Trends
80% of customers search for hotels multiple times before booking.
Customers search more than once, perhaps looking for a better price or considering travel alternatives.
Number of Hotel Searches
Perc
ent o
f Tot
al H
otel
Sea
rche
s
25% of customers return to search for hotels on multiple days.
Customers could be considering alternate plans, or looking for better deals during this time.
Perc
ent o
f Tot
al H
otel
Sea
rche
sNumber of Search Days
Data Analysis Trends
35%+ of customers search multiple cities before booking.
These customers might be more flexible in their travel plans.
Perc
ent o
f Tot
al H
otel
Sea
rche
s
Number of Cities
Data Analysis Trends
© 2016 priceline.com
Insights Summary
• Many customers search for hotels:– Multiple times– In multiple cities– From multiple properties
• Data allows us to…– See which destinations have the most flexible travelers
– See what other destinations customers have considered
– And the same for specific hotels
Ideas
© 2016 priceline.com
Tableau Demo
https://nw-tabprq-201.corp.pcln.com/#/site/finance/workbooks/1576/views
Makes and Confirms Bookings
Traditional Travel Agency Online Travel Agency
OTA?
Assists in Searching
Shares Advice and Knowledge
Booking Engine
Search Engine
Recommendation Engine
Execution/Recommendation
© 2016 priceline.com
Slackbot Live Demo
© 2016 priceline.com
Potential Customer Facing Applications of Bots
Destination Recommendation
Hotel Recommendation
Technical Details
Recommendation Engine API
Hotel Listings Search Request
www.priceline.com/stay/#/search/hotels/
Impact
★ Multisource Value Creation
★ Valuable Destination and Property Insights for MDMs
★ Unique Site Feature with Potential to Drive Direct Traffic
★ Value Add can Increase Repeat Propensity
5.5% ? Average New Customer 12 Month Repeat Propensity
★ Increase Conversion by Offering Relevant and Compelling Recommendations
★ Increase Customer Engagement via Interactive features
1.4% ? Average Hotel Retail Conversion
Next StepA/B TestingUsing Existing Engine
Q&A