How Bombora leverages AI to answer the who, when, why and what of targeting
In the rapidly evolving digital landscape, accurately resolving web traffic to its associated business domains has become a cornerstone of B2B intelligence. Bombora has been at the forefront of this transformation, pioneering innovative solutions that redefine how organizations interpret and leverage web activity. Today Bombora’s products and solutions empower businesses to answer fundamental questions of who, why, and when specific businesses should be targeted – and, importantly, with what messages, content, products, and services. The answers to those questions provide critical insights for B2B sales and marketing organizations in their work to identify the right accounts to include in their outreach.
To answer these questions, Bombora’s identity resolution determines who visited a website at a company level, as well as the visitor’s level of seniority and functional area. Then, our content extraction and classification models determine what materials were consumed by visitors. Next, our anomaly detection capabilities identify the significance of those events to determine if they were real research that indicate intent or just noise. All of these data points power a range of Bombora solutions including Company Surge®, Identity and Enrichment, and Digital Audiences.
Bombora’s approach to AI
These impressive capabilities are possible because Bombora’s AI technology spans a spectrum of methodologies. In our large-scale distributed environment we leverage a variety of machine learning (ML) techniques to handle tasks such as visitor-to-company-domain creation. We also harness the power of Natural Language Processing (NLP) to identify which materials were read on the page and classify them into our taxonomy. Large language models (LLMs) are used to accomplish taxonomy-related tasks.
Our advanced technologies empower us to extract actionable insights from unstructured data and drive innovation in the content recommendation systems.
Bombora’s domain resolution employs three primary methods:
1) We achieve exclusive, deterministic resolution through IP address and first-party cookies, hashed business emails from registration forms, and Single Sign On providers (thanks to Bombora’s Co-op). This is a unique and highly foundational part of the system that provides us with high confidence in IP-to-domain assignments. As such, this serves as a “seed” to our probabilistic identity resolution models.
2) Over years of experience, Bombora has developed best in class AI technology using behavioral modeling powered by Bombora’s Data Co-op and deterministic traffic signals. As a result, Bombora generates proprietary IP-to-company data, used in combination with external vendors, to feed a composition-based model to maximize both reach and accuracy of domain resolution. These AI models are a combination of a variety of ML solutions, including supervised and unsupervised classical machine learning as well as deep learning models. Multiple, intertwined techniques are used to provide the most robust AI solution.
3) For website identity resolution performed via third-party cookies, Bombora partners with reputable, consented vendors (like LiveRamp) to create a ‘graph-of-graphs’ approach to generate a high-quality composite signal with minimal false positives.
Bombora is committed to providing the highest quality domain resolution, and we’ve had to adapt our work to external challenges to keep that promise. For example, starting in 2020, we responded to customer needs by building an AI model that recognizes when employees work from home. Identifying web traffic as originating from a company employee rather than an unrelated individual is a complex challenge, but today our AI solution has achieved high reliability in doing just that, delivering large-scale visitor data resolution while maintaining exceptional quality.
More recently, Bombora responded to the expected deprecation of cookies. We developed a deep learning AI model that connects multiple anonymous profiles, which enabled us to generate a densely connected graph of events linking unknown visits to the known ones. This allowed us to experiment with probabilistic event linkages and probabilistic identity resolution predicting best domain on events.
As we researched to develop our probabilistic domain resolution model, we discovered that VPN solutions have transformed over the years. We had to pivot and find ways to identify IP addresses that belong to Zero Trust Network Technologies type of VPNs, which allow multiple corporations to share the same IP addresses.
Bombora continues to improve its AI models, leverage emerging technologies, and deliver future-proofed identity resolution for our customers. As the digital and external world changes and evolves, we will continue our research and model refinements to make sure that events are resolved to the correct domain.
Unlocking data-driven decisions
Our innovations have not only improved the accuracy of domain resolution but have also unlocked new possibilities for data-driven decision-making. Companies can now identify emerging market trends faster, detect competitor movements more accurately, and personalize their outreach with greater precision. By embracing continual innovations within our solutions, we are committed to setting the industry standard for web traffic-to-domain resolution—helping businesses navigate the digital world with confidence and clarity