On this second occasion, we have had the wonderful pleasure of interviewing Ricard Bonastre, Co-founder and CEO of Lead Rating. We explain how in your company, Lead Ratings, they get to predict how hot is the lead, and how likely is this to convert.
P. Alpuente– What is Lead Rating?
A. Bonastre – Lead Rating is a company's marketing predictive uses a model of Software as a service (SaaS) able to develop models for algorithmic to improve sales and marketing processes.
We develop predictive models that improve each of the phases of the funnel sales; from the acquisition until the conversion or loyalty. For example, we have models that are geared to Lead Scoring. When you enter a lead, the software is able to tell which lead is most likely to convert.
P. Alpuente – Then, with an excess of leads you are able to see what leads you attack before because it may give you a greater chance of success.
A. Bonastre – Yes, this would be a classic model of Lead Scoring. The analysis of the historical information of a client to identify a few patterns that the predictive model used to predict in leads following what are the most likely to convert.
P Alpuente – I understand that the larger the database that will facilitate the client a greater possibility of generating a model that has the most success.
A. Bonastre – It is a balance between two things. It is obvious that the more volume of information there is, the more precise you can be the model, but sometimes there are also the amplitude of the variable. The more variables you have is also good for the model.
P. Alpuente – In the case of a supermarket, can you get to generate a potential cart with all the products that you have to buy a customer?
A. Bonastre – Yes, it would be a model of recommendation. We have made a draft of an e-commerce a company that sells perfume. They recommended always to the Top of Sales that would be a model of common sense, and we have added a model of recommendation using an algorithm that in time about five products over fifty.
P. Alpuente – What variables are taken into account in this case?
A. Bonastre- Variables that have to do with the user profile, variables of behavior and the historical with the type of purchases you have made. We do two analyses: one for users and another of the products, and we make a model of matching.
In the end, the model machine learning they have the great advantage that it adjusts itself. That is to say, we launched a first model and then the model has a way of auto-train because what you're feeding results.
P. Alpuente – I'd like to explain how an algorithm is capable of automatically learning more from you.
A. Bonastre – It is a matter of identification of patterns. For example, let's imagine that we analyze a lead. This lead is very influenced by the channel of acquisition, and that allows us to identify the real interest of the person. What they do is to identify patterns that are sufficiently stable to apply them on other shows. They are sturdy and the trend of the prediction is always good.
P. Alpuente – As I understand Ricard, you have a lot of clients in the sector of training. How a model of lead scoring is very similar to the behavior between different schools to the time of the analysis leads?
A. Bonastre – We work with models, Ad Hoc, and algorithms customized. It is true that there may be behaviors more or less similar, but in the end, every company has its own logic.
We invite you to watch the interview with Ricard Bonastre in video format to understand in more depth how to use predictive models to achieve success.
