AI (Machine Learning) at DDG
Recently, I was at a conference for marketing. And a marketing exec simply challenged me to answer this question—How are you different?
He stated that all vendors were making the same claims—"use modeling to improve outcomes.” At first, I was surprised at his implication that Data Decisions Group(DDG) was just another pretty face.
Then, as we talked I begin to grasp his perspective. So here is a summary of how DDG is certainly different from our colleagues in the marketing analytics space that applies artificial intelligence.
First, DDG has created proprietary analytical software that is Applied Artificial Intelligence in the form of machine learning to allow us to classify and predict—“Reach” is the service name. We are not simply applying SAS, SPSS, R or another vendor’s solution. The Reach platform is unique by definition.
And the system processes file input, file enhancement, model build, scoring and reports in minutes—FAST--as our offline version normally takes weeks. Think 6 minutes versus 6 weeks.
Second, DDG is offering this machine learning platform for Customer Acquisition singularly. Someday, we might extend into the other phases of a customer lifecycle but today we address acquisition only.
Third, the focus is B2C. Our datasets cover the United States for the over 18 year- old marketplace. In our custom work, we handle other datasets but not in the Reach platform today.
Third, we only analyze direct mail campaigns. We understand the world has gone digital, but our clients realize that the American consumer continues to engage with direct mail in a profitable manner. Our industry Best Practices data standardization, data hygiene and data integration methods give our clients an advantage before we start modeling. It is a cliché but true, a model based on GIGO does not produce good decisions.
Fourth, DDG’s dataset in heavily skewed to offline data feeds—public records, purchase transactions that allow us to offer almost perfect identity resolution. Thus, our data enhancement is very complete and accurate which reduces error in modeling--not only better scores but more stable scores. Digital survey data is used to augment the offline data as it allows us to increase the velocity of our consumer observation process.
Fifth, the DDG Reach platform produces both descriptive and predictive outcomes with visualization of the discriminating values. This is no Blackbox. As data scientists for our marketing clients we a. believe in transparency b. realize the need for both the C suite and the advertising agency creative staffs to understand the segmentation and thus the messaging guidelines depicted by our classifications and scores.
|Customer Acquisition||Upsell, Cross sell|
|B2C Population over 18||CRM universe, B2B|
|Channel: Direct Mail||Channel: digital|
|Individual data from transactions (offline)||Anonymous Values for digital behaviors (online)|
|Transparent descriptive/ predictive visualization||Blackbox, "trust me"|