Appreciate a good black box algorithm, but maybe understanding the inputs would be helpful to know if we're actually leveraging those inputs. Further, the ability to back-test their likelihood to close vs observed reality for each organization would help establish credibility. Very few sales leaders will leverage a (completely) black box solution.
"HubSpot uses the most current predictive machine learning algorithms known as black boxes to provide accurate predictions. With a black box, data scientists understand the input and outputs of the model, but how the input is transformed into the output is unknown. These models have been proven to outperform white box models, but it is not possible to break down how each individual input contributes to a contact’s score. Instead, the focus is on the overall predictive performance of the model." -- please see Determine likelihood to close with predictive lead scoring .
Note: Please search for recent posts as HubSpot evolves to be the #1 CRM platform of choice world-wide.
"To set values for these properties, HubSpot analyzes the following data:
Analyticsandconversioninformation (e.g., web page visits, time of last visit,email interactionsincluding clicks, opens, and replies, and form submission events).
TheLifecycle stageproperty. If a contact's lifecycle stage value isCustomer, theLikelihood to closevalue will be cleared, and theContact prioritywill be set toClosed Won.
Firmographic information provided by HubSpot Insights about the contact’s company.
Firmographic information about your business and HubSpot account.
Interactions logged in the HubSpot CRM (e.g., tracked email clicks, meetings booked)."
I will also share this feedback internally to see if we can edit the article to add more detailed information.