B2B Data

Data Science and Demandbase – Yeah, We Do That

August 1, 2022

3 mins read

Data Science and Demandbase

Data Science and Demandbase – Yeah, We Do That

We recently highlighted the numerous modeling data attributes available from Demandbase. This blog will address how and where this data is used in the model building process.

There are many paradigms that can be used to represent the model building process in different levels of detail – a long popular example is CRISP-DM. For this exercise, we’ll use the fairly simple example shown below.

Demandbase can help with the model building process in two key ways. First, the more traditional way that we’ve supported Data Scientists(DSs) is in the Data Gathering and Data Pre-Processing Steps as shown below. Second, Demandbase recently launched D2 Labs – a group of data scientists, data analysts, and solutions engineers that can assist in the entire model building process, developing customized models and predictions for our customers. We’ll cover those more recent capabilities later in this post.

Data Gathering/Sourcing and Data Pre-Processing:

1. Data Gathering/Sourcing

  • Data Scientists (DSs) are puzzle solvers, i.e., they ask questions like, “what set of data attributes/variables/features will help me predict a certain outcome?” – an outcome example being, predicting which leads will convert into pipeline opportunities. This is an important question because if we can answer this, we can better focus our sales and marketing efforts and dollars on those leads with the higher probabilities to convert.

When predicting the probability to convert, DS’s may have some ideas on which set of data attributes will be most helpful based on domain and industry knowledge. But, there are attributes/variables/features that are predictive that they may not be aware of. That’s why DSs want “all of the data!” – they don’t know what they don’t know. Today, DSs can input tens or even hundreds of variables into their algorithms. That allows them to determine which of all of those potential features help better predict lead-to-pipeline conversions.

2. Data Pre-Processing

Demandbase can play a couple of roles here.

  • Data Quality: Most companies’ data is quite dirty – missing data, incorrect data, old data, non-standardized data, data/record duplicates – it can go on and on. DSs often spend 70-80% of their time gathering and preparing data. Demandbases’ Data Integrity offering can clean and standardize account data and contact fields, deduplicate records, fill in missing fields, and append many data attributes including firmographics and technologies used. This data can even be augmented with real-time data on buying intent and churn signals.
  • Many clients will have data they need in specific formats. Some data may even need to be aggregated and/or rolled-up for analysis – we do that!

The Entire Model Building Process

Demandbase’s recent launch of D2 Labs offers a significant advantage for customers that don’t possess the prerequisite knowledge to build their own models – as well as for customers that want to leverage Demandbase’s vast data repositories and take advantage of an Analytics team that works day-to-day with this data in conjunction with 1st-party data.

Our D2 Labs scientists wield custom data sets (1st and 3rd-party) to unearth new opportunities using proprietary propensity models. Customers can now engage with Demandbase’s data experts and data scientists directly to turn their incomplete, inactionable data into revenue-generating insights.

D2 Labs has worked with enterprise customers such as Microsoft Azure, Adobe, DocuSign, and more.

To learn more, contact your CSM or account manager. If new to Demandbase you can reach us here!

Related articles