MadKudu is an AI-powered marketing intelligence platform that helps marketers build models to better score, prioritize, and understand leads and accounts.
The platform also uses machine learning to help you identify leads that are a good fit and make predictions about how much they'll spend.
We spoke with Francis Brero, Cofounder and CRO at MadKudu, to learn more about this marketing AI solution.
MadKudu is a marketing intelligence platform that supports complex businesses by removing operational challenges, enabling teams to work faster and more intelligently.
We leverage a few levels of machine learning throughout our product.
At the core, our models are built to help predict outcomes for customers. However, we also use models to generate features (aka predictors) and overcome feature sparsity typical in B2B, especially on the third-party enrichment side. There are different types of algorithms used, from random forests to non-linear regressions; however, a strong specificity of our platform lies in how we utilize homegrown optimization functions to better fit the highly contextual needs of B2B sales.
Logistic predictions such as lead and account scoring and prioritization, classification, such as segmentation to fuel personalization, data science (PCA, kNN…). Also, audience analysis to better understand your ideal customer profile (ICP) and what makes a lead a good fit/likely to convert, and value predictions for the size of wallet estimation and predicted spend.
Compared to non-ML solutions, MadKudu offers predictive models based on historical conversion data rather than manual point assignments. Predictive models help fight pre-conceived notions and identify non-obvious trends. For example, companies based in more expensive zip codes have more budget and thus more buying power. On the behavioral side, we factor in time decay to better represent the fluctuations of intent over time. Each decay is different based on the type of event. For example, opening an email has a small impact overall and a short effect. Registering for a webinar will have a longer-lasting impact on intent. These are critical to take into account to ensure we understand where the prospect is in their journey at any given time.
Compared to other AI-powered solutions, our algorithms optimize for intelligibility by end users to foster high adoption levels rather than optimizing for pure predictions and failing the “black box” test. It is important not to leverage features that have high predictive power but low explainability. For example, “company uses .ly domain” might be predictive of a high budget, but this information alone won’t enable sales to be more effective in their outreach. However, it might be tied to the fact .ly companies (that are still operating) are scale-ups at this point.
Not really, but you are likely not going to need the AI part of our models if you are looking at less than 2,000 new leads per month. However, you will want the data science part to facilitate first and third party data manipulation, real-time orchestration, and decision making. Also, the exploratory capabilities help with strategic assessment.
We work with amazing companies, primarily in the B2B SaaS space, like InVision, Clearbit, and Intercom.
Education. People are still confused about when they should use AI, what it should do, when NOT to use it, and why you might not want self-learning.
Design standards are a huge challenge that we are actively working on to make sure we get end-users to adopt AI rather than feel like they are being ruled by it.
AI will bring automation to a new level and allow for much more granular personalization. However, this will require stronger and better-defined marketing strategies that can leverage AI capabilities. AI in itself doesn’t solve a new problem. It helps scale the solution of making the journey relevant to the user by adapting. AI will bring the highly personalized experience of interacting with a great sales rep at a high-end fashion store to the masses at scale (understanding your needs, tailoring the offering to who you are and what you might not know you are for).
Always start with the problem you are trying to solve for your customers. AI should help scale the solution beyond rule-based automation due to the volume of decision points or data required to decide. Don’t do AI for the sake of it. While AI companies can help with best practices, it is critical to not fall prey to the desire to auto-magically make all your problems simply go away. AI is a tool, not a solution.