Customers using Bombora’s Company Surge intent signals are enjoying even more precise results thanks to a major advancement, the company announced . Human evaluation of the company’s new BERT-based B2B Topic Classifier shows a 26% improvement in topic prediction — that is, a more precise understanding of the categories business buyers are researching — along with increase in domain coverage (the companies those business buyers are from) against topics. Bombora now captures more of the research signals that b2b sales and marketing teams use to identify, prioritize, target and message prospects resulting in triple digit performance improvement.
Marketing Technology News: ShipBob Joins Shopify Plus as Certified App Partner for Order Fulfillment & Ecommerce Shipping
Bombora earned those improvements by basing its Natural Language Processing on Bidirectional Encoder Representations from Transformers, or BERT. BERT, a machine-learning technique that learns by attending to the context of each occurrence of a word — distinguishing “the stock of apples has dropped,” for example, from “Apple’s stock has dropped” — revolutionized Google Search toward the end of 2019, a year after Google open-sourced it. Among the businesses now using BERT are Microsoft, LinkedIn, Facebook, Wayfair and other
Bombora’s data science and engineering team spent a year designing, building, and deploying the new system. Implementing BERT demonstrates Bombora’s data science and machine-learning leadership , and the company will continue to add major feature enhancements on top of it. Helping to drive these efforts forward will be Amber McKenzie, Ph.D. formerly of PwC and DialogTech, who has joined Bombora as VP Data Science, bringing deep NLP expertise to leverage BERT in other NLP use cases.The company has already enhanced its web-page categorization in order to better focus on the digital content that most strongly suggests intent.
Marketing Technology News: Kimble Applications Welcomes New Members and New Roles to Executive Team
“Bombora’s engineers and data scientists never stop looking for ways to serve our customers better,” said Bombora Data Science VP Nicholaus Halecky, Ph.D. “The BERT-based B2B Topic Classifier demonstrates a substantial performance increase in Company Surge® intent signal quality, and we know this will improve our customers’ business results.”
