architecture for measuring ad - developermarch · adtech data and measurement april 25, 2019 yahoo...
TRANSCRIPT
Architecture for Measuring Ad Performance at Scale Divya B
AdTech Data and Measurement
April 25, 2019
● Yahoo Bangalore (8 years)
● Yahoo Sunnyvale (2.5 years)
● Apply Cupertino (2.5 years)
● Uber Bangalore (10 months)
Journey
01 Project Overview 02 Architecture 03 Real time ingestion 04 Batch ingestion 05 Aggregation 06 Technologies 07 Wins
Agenda
01 Project Overview 02 Architecture 03 Real time ingestion 04 Batch ingestion 05 Aggregation 06 Technologies 07 Wins
Agenda
Intra-channel
Optimization
Creative Optimization
ROAS
Ad Partner Data
Performance Metrics
Conversions Data Science
●Spend
●Number of installs
●Number of clicks
●Number of trips
●Number of orders
●Number of sign up
●Number of first trips in N
days
●Number of trips for a region
●Cost per trip
●Cost per first trip
●Cost per sign up
●Projected first trip
●Lifetime value
●Cost per projected first trip
01 Project Overview 02 Architecture 03 Real time ingestion 04 Batch ingestion 05 Aggregation 06 Technologies 07 Wins
Agenda
Architecture
Attribution
Ad shown to an user
Ad shown to user
App installed
User installs the app
Signed up
User signs up
01 Project Overview 02 Architecture 03 Real-time ingestion 04 Batch ingestion 05 Aggregation 06 Technologies 07 Wins
Agenda
Real-time
Data
Ingestion
01 Project Overview 02 Architecture 03 Real time ingestion 04 Batch ingestion 05 Enrichment 06 Technologies 07 Wins
Agenda
Events
Ingestion
Batch
Spend
Ingestion
Batch
01 Project Overview 02 Architecture 03 Real time ingestion 04 Batch ingestion 05 Aggregation 06 Technologies 07 Wins
Agenda
Aggregation
Data Quality
● Unit tests
● Integration Tests
● Dashboards and alerts
01 Project Overview 02 Architecture 03 Real time ingestion 04 Batch ingestion 05 Aggregation 06 Technologies 07 Wins
Agenda
Technologies
● Apache Spark
● Apache Kafka
● Athena
● Piper
● Java/Python/Go
● Apache Cassandra
● Apache Hive
● Apache Hadoop
● Celery
01 Project Overview 02 Architecture 03 Real time ingestion 04 Batch ingestion 05 Aggregation 06 Technologies 07 Wins
Agenda
Wins
● Scale
● Plug and play
● Data accuracy and freshness
● Extensible
● Modular
Questions?
Proprietary and confidential © 2019 Uber Technologies, Inc. All rights reserved. No part of this document may be reproduced or utilized in any
form or by any means, electronic or mechanical, including photocopying, recording, or by any information storage or retrieval systems, without
permission in writing from Uber. This document is intended only for the use of the individual or entity to whom it is addressed and contains
information that is privileged, confidential or otherwise exempt from disclosure under applicable law. All recipients of this document are notified
that the information contained herein includes proprietary and confidential information of Uber, and recipient may not make use of, disseminate,
or in any way disclose this document or any of the enclosed information to any person other than employees of addressee to the extent
necessary for consultations with authorized personnel of Uber.