Common approaches to building lead scoring models
There are three different approaches to building lead scoring models:
Manual Lead Scoring
This is the analysis, agreement and adapt approach to lead scoring. Using some basic analysis techniques to identify variables that are likely indicators of your desired outcome happening. This analysis informs a conversation with key stakeholders (we suggest Marketing, Sales and Customer Support/Success) to validate and prioritize. Based on this conversation weighted values are assigned and a model is deployed. Ideally, this model is routinely evaluated and adjusted. Manual lead scoring is, in our opinion, the best way to get started with lead scoring as it allows for quicker implementations with lower effort and has the benefit of opening a dialogue between revenue focused teams.
Logistic Regression Lead Scoring
This method employs building a statistical model, often in excel, that determines the probability of an outcome. It works really well for Lead Scoring as we are often looking to determine a binary outcome - will a deal close or be lost. Taking this approach does not eliminate the need for cross-functional alignment but rather shifts the conversation to a more objective focus. To get the most ideal outcome from applying statistical modeling amore advance Multiple Logistic Regression Analysis which determines the likelihood of a binary outcome based on multiple independent variables. You might be happy to know that we are building Logistic Regression capabilities into Breadcrumbs allowing for an ideal hybrid approach - manual focus with statistical support.
Predictive Lead Scoring
You will often hear Predictive Lead Scoring referred to as Machine Learning or AI based lead scoring. Essentially, Predictive Lead Scoring is the application of statistical models like Multiple Logistic Regression to your data at scale in a continously optimizing fashion. Hypothetically this is nirvana from a lead scoring perspective. The reality is that without manual intervention it's almost impossible to inform the model with the nuances of strategic priorities and point in time factors. Ultimately, predictive lead scoring often leads to false positives as a result and more often leads to poor adoption because of it's black box nature.