AI Data Democratization
— Sharing knowledge, not data
AI Data Democratization for Marketing — Sharing knowledge, not data
January 2019, Brent J. Dreyer, Managing Partner, DataEM.com
What is AI Data Democratization for Marketing? Let’s start by first looking at just “data democratization”. Data democratization is the function of making digital information available to end users who are not data specialists.
In a marketer’s world, the CDP (Customer Data Platform) has taken a step in the direction of data democratization by creating a unified view of a customer, connecting data from across an organization’s silos of information, finding behavioral patterns that trigger communications out of connected marketing channels.
By the end of 2018, the CDP Institute reported that 78 vendors were offering CDP platforms, having attracted a total of $1.7 billion in funding. Many of these CDPs incorporate some level of machine learning, analyzing customer behavior, making predictions of outcome, suggesting and facilitating the appropriate next action.
Think of Amazon, an early adopter in the marketing space that utilizes customer behavior to enhance sales. As you look at a product on Amazon, its AI engine suggests alternatives, manages the appropriate sponsored ads and displays logical product add-ons. After leaving the website, your world becomes strategically inundated with email reminders and banner ad suggestions appearing everywhere you are connected to the Internet. A layer deeper are the products and services suggested by other vendors, triggered by the sale of your behavioral data by Amazon.
In previous example, Amazon runs their marketing world from their private CDP, generating over $258.22 billion in US retail sales in 2018, capturing almost 50% of all online retail spend. Data is the keystone to their successful AI engine, allowing them to perpetually build smarter and fluid machine learning algorithms.
How can the average marketer compete with giants like Amazon? Short of buying their data, there are other business models in place. For example, organizations can share their data with cooperatives, gaining insight from the collective analysis and renting lists of customers from non-competing entities, matching behavioral and demographic characteristics. But, to participate, you must trustfully release your customer data into the pool and pay to get other data back. Given the nature of the cooperative model, the data is not responsive in real-time as it may take weeks to compile, analyze and release.
Other avenues to competing with the Amazons of the world include releasing your data to the analytical giants like Epsilon or Acxiom, paying them to run your data through algorithms, supported by data collected from other clients, building predictive models of what your client looks like.
In the above scenarios, there are several common threads. You give up control of your data, you do not own your analytical models, the results are not based on real-time data and you pay for these services.
This is where AI Data Democratization for Marketing enters the picture. Imagine a marketer’s world where data can be processed with the collective intelligence gleaned from billions of consumer actions and transactions, in real-time, offering immediate suggestions, all done without releasing your data. That is the power of AI Data Democratization for Marketers.
How would this work? With AI Democratization, the knowledge is shared and not the data. In this model, a sponsoring institution or organization develops the core AI algorithms from a sample dataset and validates the predictions. The core algorithms, along with the AI interface and engine, are freely distributed to participating organizations. Those organizations install the platform on their own systems or cloud, running the AI models on their own data.
The marketers adjust the parameters (“turn the dials” so to speak) to validate the best settings against controls from their own data set. The intelligence in the system allows this to be done by marketers without the assistance of data scientists. The settings and scores from each participant’s “tuned” algorithms are fed back to the sponsoring institution or organization where the core model is adjusted, and the renewed intelligence is distributed to the programs’ participants. The cycle continues and the system gets smarter, without the distribution of data.
The mechanics of AI Data Democratization for Marketers may seem like a “moon shot”, but it will happen as momentum for AI democratization is building today in other industries. Take healthcare, for example.
The transportation or sharing of patient data is heavily regulated, posing limitations on the build of learning models that incorporate the necessary, deep datasets. This is changing today with AI Democratization, as the foundations are being built to allow hospitals to share knowledge without releasing patient data. Without a doubt, this movement will carry over to other vertical industries and marketing will be one of them.
To learn more about AI Data Democratization for Marketers, or to inquire about sponsorship or participation, contact me at BrentJDreyer@DataEM.com, or call 954.906.2590.