Last week, at our Shaping the Future of Consumer Marketing event, we emerged from stealth to unveil our Intelligent Customer Data Platform. I am humbled by the enthusiastic participation of thought leaders from some of the world’s most loved brands, including Alaska Airlines, Belvedere Vodka, Domaine Chandon, GAP, Kendra Scott, lululemon, Nordstrom, Moët Hennessy USA, Starbucks, TGI Fridays, and Pacific Market International.
It was not very long ago that a group of visionary investors took a bet on what was not yet an answer ‒ it was a commitment to trying to solve a problem that might not be solvable. Now, our launch comes after 21-months of ground-breaking work by an amazing group of engineers, data scientists, product managers, marketers, and operations personnel. What makes this milestone so special is that launching the company was predicated on an ambitious set of criteria: the product had to work, at enterprise-scale, with phenomenal customer success stories to prove it. I want to take this opportunity to share a bit more about our journey.
Why we started Amperity
While building enterprise scale marketing software at Appature and IMS Health, Derek and I were struck by how many companies were using only a fraction of their customer data. Business leaders, data specialists, and marketers all knew that they needed to leverage more of their customer data in order to compete, and they went to extreme measures in an effort to do so.
We watched as brands spent tens of millions of dollars and thousands of people-hours on schema mapping, manual data cleaning, normalization, and ETL coding to transform data into usable formats. When the databases were ultimately built and populated, they were rigid and prone to break if data was missing or incorrect, or when new data sources were added. Simply put, the outcomes utterly failed to deliver and grew worse every day.
The most formidable hurdle, however, was identity resolution at scale. Consumer brands have neither the software nor the expertise to resolve customer identities across sources. Most technologies require common keys in order to link records and form customer profiles, but the majority of systems have not been built to integrate with one another. This means no common keys, no matches, no unified customer profiles, and a huge amount of valuable customer data wasted.
This first hand look at the struggle to make data actionable galvanized us to find an answer. We vetted countless vendors that claimed to solve this problem, but all were using the same manual, slow, and incomplete methods we had seen fail again and again. We talked to roughly a hundred brands and they all shared the same data struggles. After realizing that no purchased or built-in-house platform offered the capabilities that brands needed to unify and activate all their customer data, we set out to build a solution ourselves.
Our unique approach
Our quest led us to the University of Washington and Dr. Dan Suciu, one of the world’s leading experts on data management, with particular emphasis on working with uncertain data. Dr. Suciu’s research focuses on probabilistic matching, parallel data processing, and data security. He is a fellow of the Association for Computing Machinery, holds twelve US patents, and is the recipient of numerous honors and awards in his field. With his help, we explored a revolutionary new approach: probabilistic data matching across sources using machine learning and scalable computing power.
We quickly saw the tremendous value in this new way of working with data. First, raw data could be ingested and stored without cleaning, normalizing, or transforming it. This is because the intelligence of machine learning, coupled with scalable computing power, enabled us to analyze each unique piece of data, categorize it, and match it with incredible scale, speed, and precision.
Second, probabilistic matching does not require common keys. Instead, data is matched based on the likelihood that the information describes the same individual. This process involves analyzing terabytes of data, training machine learning models, and adding layers of customer data-specific intelligence. Using these methods, brands can build multiple customer databases that operate in parallel, each optimized for specific use cases.
The probabilistic approach might sound complex, and in execution it is, but the results are simple: it unlocks all the customer data that brands couldn’t otherwise use.
As we sought feedback with our network of marketers and data specialists, it became clear that if we could build a machine-learning powered customer data platform that used a probabilistic approach, we could solve the customer data problem. In January 2016, after building a small core team and raising our initial round of funding, Derek and I co-founded Amperity.
Amperity’s Intelligent Customer Data Platform
First and foremost, the Amperity platform is the foundational layer that enables brands to use complete customer data for any data-driven experience they can imagine. We offer a seamless pipeline that continuously ingests raw customer data from any source, leveraging machine learning and scalable computing power to resolve identities and match records. The resulting customer profiles are rich with attributes about individuals’ in-store and eCommerce purchases, search and browse behaviors, email and social interactions, and more. If the data exists, we will ingest and unify it.
We built Amperity to empower marketers and analysts, not only through data unification, but also by giving them the direct ability to activate that data themselves. Our visual and SQL interfaces are both intuitive and powerful, so data can be explored, segmented, and activated by the people best positioned to understand and use it intelligently.
Amperity operates at trillion-entry scale and in real-time, in order to serve the comprehensive customer data needs of today’s global consumer brands. Without scale and speed, brands cannot leverage all their data in the ways that are most valuable to consumers: personalized and timely interactions across site, mobile, social, email, customer service, and in-person experiences.