I joined the Amazon Advertising Platform in New York City in August 2011 just when they were starting to build out the advertising technology stack. I joined as a Principal Product Manager to build out their analytics capabilities and after my first year there I got promoted to building and then subsequently running the global shopper and marketing insights team. I moved to Seattle In October 2012 to build out the team at the headquarters and returned to New York City in April 2014. My team built out the Amazon advertising ROI measurement platform, brand diagnostics and analytics and advertising effectiveness via Brand Lift Measurement, Offline Sales Lift Measurement and Marketing Mix Models (MMM). These data and analytics products provided business intelligence and reporting throughout the campaign lifecycle spanning Digital upfronts, pre-sales planning, optimization, measurement, end of campaign performance and advertising effectiveness. These products were cross-screens and cross-channels (i.e. desktop web, network, mobile, tablet). The team also developed a professional services model providing consultative data services to the largest advertisers and their agencies on the platform. I hired, developed and mentored teams of 25 people spanning product management, data science, data and software engineering.
Key product accomplishments include:
- Launched Cross Channel Attribution for all advertisers.
- Launched ROI Measurement – measuring Online and Offline Sales Lift in CE, CPG, Automotive & Fin Serv. (Offline partnerships with DataLogix, Gfk, Acxiom). Calculate sales lift using attribution windows and synthetic test & control groups.
- Launched Retail brand diagnostics and retail health analytics.
- Conducted Marketing Mix Models to inform advertiser’s investments across channels and formats (across advertisers and vendors).
- Launched campaign benchmarking tool.
- Launched Brand Lift Measurement (partnering with Nielsen).
- Launched the Amazon Card Engagement Framework (ACE) for Financial Services advertisers.
- Launched Customer Centric Targeting and Measurement.
- Launched Amazon’s brand analytics platform (Amazon Brand View Pro – ABVP).
- Launched predictive models to drive platform performance optimizing for clicks, conversions, purchases and app downloads.
Platform stats:
Technologies deployed: Amazon EMR, DynamoDB, S3, Redshift, Tableau, R, SQL, Hive
Measurement Models
Last-Touch Attribution. With this model we attribute a conversion to a user’s last click on an ad campaign, or to the user’s last impression of an ad if the user has never clicked on an ad from that campaign. We can attribute conversions for up to 30 days after an ad impression but default to a 14-day attribution window. We typically use last-touch attribution to optimize the targeting, placements, and creative of campaigns. Last-touch attribution does not by itself measure the incremental impact on conversions of an ad campaign, unless explicit test and control groups are created for the campaign, where the control group would see ads that are unrelated to the campaign.