Use Case - Lead Scoring Model

Let me show you how you can use Count dot ai to build a really great lead scoring model. So in the Canvas, I've added a few different datasets that can give us a sense of how a lead is doing and its intent. I've got access to social media interactions, various different data sources from our website, including the website page visits, sign ups, demo requests, and content downloads. I've got access to product sign ups and any email engagement that's been going on through our marketing, and then customer contracts, deals database, and then obviously the full list of leads and the companies that that they belong to. So we've got a really big rounded dataset here of all the different direction points that our leads are having, both current leads and historical leads. And I've written a few different methodology points that I want to use to inform the model. So I want to use all these different data sources to give us an accurate lead scoring model using historical data. So I want the AI to look at our historical leads and where they actually converted, and then use that to create a model which we can use to project on live leads going forward and give them each a hotness score. I've also asked it to flag any data quality issues that we might need to look at to make the model even better and ask it to invalidate any particular notes we have on buying persona and also take into account that there's a bit of a change in the last six months if we changed our pricing model. So now we've got all this data and this methodology. I'm gonna add an agent to the canvas and give it access to all this data and the methodology as part of its context and ask it to start building the model. Hey, Cam AI. Could you use these data tables and this Nota methodology to build as a lead scoring model, please? Off it goes. Okay. So the agent's finished. It's done a huge amount of work for us here as it's dug in. So you can see that it's looked into what makes a successful lead in the past. It's validated our ideal buying persona. It's then built a behavioral pattern and feature importance based on previous deals, looked at firmographic patterns, etcetera, and come up with ultimately a hotness lead distribution, which we can then use to score live leads. It's also come back with a list of data quality issues for a v two model. So it looks like there are thirty two demo no shows incomplete and various other things that we might wanna go fix. For example, only two does had expansion MRI, which maybe doesn't quite fit what we expected. So all this is summarized at the top. And, thankfully, all that is done, I can go interrogate. So I can look at any cell. I can show its input to see exactly what SQL it's written and even edit this if I needed to. So now it's finished. I can then bring in my team and share this with my team so they can come in and interrogate this model with me and look at the logic, start using this and turn it into a report. We could even ask the agent to build this in this as a dashboard so we could start using this regularly, And then start to work on potentially a a v two model looking to bring in other data sources to or new ones to to give us better data quality as well. As you can see, already a fantastic start and a way we can start using, our data to give us a better lead score.