Lead scoring is a powerful tool that can help your organization in several ways. You can use lead scoring to prioritize leads — if you have a lot of leads, and not a lot of time. You can use lead scoring for forecasting. And finally, you can use it for prescriptive analytics, aka, you can use it to change how you approach different leads. In this quick, no-math introduction, we will show you how lead scoring works, how to leverage it, and how to get started using Google Analytics data.
I’m Brett from Anvil Analytics and Insights and we’re gonna go over at a high level what is lead scoring and how you can kind of use it at your business? Specifically, if you’re doing anything lead-based or like taking appointments like at a healthcare site. This is part of a series on lead scoring so we’re gonna go in-depth in some of the different ways you can track, we like to track from marketing standpoint so we’ll talk about like Google Analytics and how you can track your Google Analytics data and have it set up for lead scoring as well as going into the nitty gritty, the detail of actually modeling the data and how that works, if that is also something you’re interested in, but right now it’s this is really high level and really for marketing folks who are kind of interested in things like machine learning and analytics modeling and seeing how you can use it in your business and how you can actually leverage it. So I’m gonna talk a bit about what makes up lead scoring: lead scoring really has three components to it, the first one is really your past user attributes, the data that you’re gonna be pulling in from all of your history to be able to build your models, so this is things like, hey what sources they come from, do they come from Google, do they come from Facebook, did they call you, did they text you, how are they getting into the system? So say they’re making appointments at your healthcare network. The second thing is what we call the response or what happens, the outcome, and here the response would be did they actually show up for the appointment or not? And then the last thing is you actually build your analytics model that can give a score, a percentage, say 0% chance they’re going to show up or a 100% chance they’re gonna show up, and this is usually some different model, some are called like logistic regression or logic, logit that’s a really nice thing to basically encapsulate that and give a score and something we can manipulate. And then the outcome is whenever a new person comes in and makes an appointment we can actually give them a score and say, hey this person is likely to show up or not likely to show up and here’s what that could kind of look like. Basically you could have sort of an interface and could build it out in something as simple as Data Studio or whatever worked with the existing tools that you have, and say, hey here’s my upcoming appointments for the day and then here is the likelihood of each one of these patient IDs of missing their appointment, and this is really helpful too because it’s all about maximising people’s times and also not maybe even controlling staff. So it’s about controlling cost as well, so you could say, hey you know we currently have you know like a hundred patients that come in a day, we have to call each and every one of them, so we have to have a staff who’s just dedicated to calling these people, well if you can narrow it down and say, hey I don’t have to call these five people, these six people and these other people I know they’re very likelihood just showing up, they only have a 10% chance of missing or a 20% chance of missing, I’m gonna focus on these people that are 60 70 80 plus percent chance of missing, or maybe say if this person is a 99% chance of not showing up, we may not even get to that person right away and say maybe if we still have time we’ll call that person but let’s assume they’re not even gonna show up, and that’s another thing to where you can come back and say, hey I’m not gonna worry about that person so if they’re at the end of the day say, we’re gonna just go ahead and finish what we’re doing and close everything down assuming they’re not going to show up because the likelihood they’re showing up is pretty low and if they do, yeah we can you know put things back up, but we can also save time with the likelihood they’re not gonna show up, that’s great, that really comes into these thresholds. So the really important thing here is how we get there and how we get there is data about the existing users, so collecting that data from all the different sources and then combining it together, so if you have you say your Facebook data and you have your Google Analytics data and then you have your actual appointment system, say you’re using something like InQuicker or an internal system, that you were able to see a lot of things about the appointments and be able to help string all that activity together to a single person and then say, then you can really start putting that data in and then we can weigh that through the model, the logistic regression or whatever it is, and what it does it will come in and see all these different what we call features, these different attributes, and say, hey you know in this particular instance Google search is really important or maybe you know phone calls are really important and it combined these things together and says what’s the percentage the person is likely to show up? So it might be if you visited the homepage, on a Tuesday and you came from Google search there’s a very high chance you’re going to show up, but maybe if you had a phone call and it’s in the morning and you didn’t visit the homepage, maybe there’s very low chance that you’re gonna show up, all these different factors that weigh in and maybe things you wouldn’t even think about, the model can really use and say when these things are all together this is the percentage chance that that person’s gonna show up. So it’s really helpful in that way, and the other cool thing too is maybe phone calls were a really good indicator say 10 years ago that someone’s gonna show up, and the model then might have said hey if they have come in a phone call versus you know Facebook they’re gonna show up and Facebook’s not gonna show up, but over time your model’s gonna learn as things change and maybe phone calls are becoming less and less reliable, that model’s gonna adapt to that and maybe what maybe 10 years ago it might have said one score, today it’s gonna say a different score, so it’s always continually learning and that’s a great thing about it. So I hope you can subscribe to our channel and learn more about it and we’ll also put in a link to more videos about the subject as we grow and learn all about lead scoring and how it can help your business.