apigee insights: data & context-driven actions
TRANSCRIPT
2© 2013 Apigee Corp. All Rights Reserved
Agenda
• What’s new in the world of Digital Transformation – change and context
• New analytics capabilities needed to address this changing world
• Agility in Analytical Process
• Insights Architecture
• Insights Services
4© 2013 Apigee Corp. All Rights Reserved
Self-Optimizing Digital Business
Optimize your digital business
by making
every customer interaction smarter
across every digital channel,
in real time,
using all
relevant contextual business insights
Your Digital Assets
API API
API
API
Self-
Optimizing
Digital
Business
5© 2013 Apigee Corp. All Rights Reserved
Pace of change is increasing
• After an initial startup, we saw > 1 new API every 10 days
• After 150 days, we observed slowdown in less than 10% of organizations
• Median number of APIs per organization: 14
© 2013 Apigee Corp. All Rights Reserved
New rapidly changing contextual signals occur
7
User Developer
Time Location
8© 2013 Apigee Corp. All Rights Reserved
Need for a new Big Data Platform
All Backends
All Channels
External
and
Internal Context
Apigee Edge
Apigee Insights
Correlations
Conversions
Segmentation
Anomalies
Time Series
User Engagement
Developer Adoption
API Information
© 2013 Apigee Corp. All Rights Reserved
Contextual Business Insights
10
Business Growth
User Engagement
Developer Adoption
API Information
© 2013 Apigee Corp. All Rights Reserved
Sporting Events and Beer Consumption in AU
Adjust on-hand stock of beer (by brand and packaging) for each type of sporting event
11
2008 Rugby World Cup
2008 Summer Olympics
12© 2013 Apigee Corp. All Rights Reserved
User Shopping Behavior
/storelocator/product/search/buy
/findinstore
< 3 days
< 1 day
Send personalized offers to at-risk customers based on online, mobile, and offline interactions
© 2013 Apigee Corp. All Rights Reserved
Speed up delivery of business insights
14
Collect
Analyze
Visualize
Act CAVA
© 2013 Apigee Corp. All Rights Reserved
Collect and Act faster
15
Apigee Edge
Apigee Insights
App Analytics
Internal Context
External Context
All Digital Channels
Action
Data
API Analytics
Developer Analytics
15
16© 2013 Apigee Corp. All Rights Reserved
Analyze faster: Abstractions
Flexibility
Spe
ed o
f D
eliv
ery
Map/Reduce
Entities, Events & Analytics Libraries
File System
Turnkey Solution
17© 2013 Apigee Corp. All Rights Reserved
Analyze faster: Widely used technologies
Easy access to skills (Javascript) accelerate development time
© 2013 Apigee Corp. All Rights Reserved
Insights: Architectural Components
Pre-Built Connectors
Visualization Engine
Core Execution Engine and Analytics Libraries
Entities and Events Collection Engine
Big Data Stack
Configuration Engine JavaScript Engine
Apigee Edge
API
20
21© 2013 Apigee Corp. All Rights Reserved
Demo: Retail Personalization
Apigee Insights
All Digital Channels
Internal Context
External Context
ActionData
© 2013 Apigee Corp. All Rights Reserved
Engagement Workflow
24
Kickoff Data Analysis Readout Insights
© 2013 Apigee Corp. All Rights Reserved
Analysis Process
Take core findings through multiple dimensions to deliver hidden insights
Data
Raw customer data
Sessions MapReduce
Service, Errors, Time, APIs, Status, etc.
Insights
25
© 2013 Apigee Corp. All Rights Reserved
Apigee Digital Business Platform
Web
Point of Sale
Partner
EdgeApps, APIs, Analytics
Mobile
ESB, SOA,App Servers,Databases
Apps
Backend Services
InsightsBig Data
ExchangeAPI Networks
© 2013 Apigee Corp. All Rights Reserved
Data Science Engagement Workflow
• Work with the customer to understand– What data they have available and its schema– Their technical and business goals (successes, failures, metrics, etc.)– Determine if there are other relevant external data sets
• Obtain a set of historical data for POC– Load the data into a platform instance on AWS using the bulk file importer
• Data files are staged in S3• Currently support csv, web log formats but can add easily add others• Files are stored in HDFS
– Data can be loaded into HDFS either• As time series (time stamp with an event payload)• Or as entities (a unique identifier plus relevant characteristics)
2929
© 2013 Apigee Corp. All Rights Reserved
Data Science Engagement Workflow
• View data samples in a Series Manager UI• Run MapReduce jobs in javascript (UI or via app specification)• Common jobs:
– Aggregation – group the data in time and by dimensions– Sessionization – group the data per end user interaction– Bringing in context (joins with other data sets)
• Results can be written as– Time series– Entities– As a final result in JSON
• Final results can be downloaded via a REST API:– /jobs/{jobId}/{runId}/results
3030