Equals 11’s Post

Transcript

How the model and how instant works. So when we talk about instant, it's important to say that first of all instead needs data collection. So we need to define the historical data or historical segment of data that is going to be used for creating the forecasts or the predictions then. And we will have a a next stage that it's going to be at the data preprocessing step. In this preprocess step, what is going to happen when you are going to use a model for predictions? You need first to organize your data and probably you will need to do some cleaning to your data because you can have a lot of blanks you can have like. Typos, misspellings, So anything can impact your results. So in that way it's really important to have like clear data for this stage, for the preprocessing data step. And then what is going to happen, We are going to create the best part. We are going to create the model and creating the model, it's not a big deal, it's important of course, but yeah. It's it's it's a easy thing to do here in self course because yeah, Selfors and insane, of course they have native capabilities that are going to allow us to do this easier. Once the model is built, we are going to validate. Most of the common scenarios would be to provide like the best samples or the best scenarios and the worst the worst ones. So in that way we can train the model and say, oh look, this is our best, this is like a good one and this is the not that good and this is the worst. Once we have that type of validation we can iterate multiple times with that validation and if you think at some point that the the the all the data that you are getting. All this important insights that you are getting from Einstein are accurate for you. And yeah it makes sense. You can deploy, deploy would be the last step and then also you can distribute the data and then what is going to happen, you need to iterate, it's a loop. So those predictions are going to be improved throughout the iterations because when we talk about instant prediction builder. We also need to understand that data is alive and constantly. Data is going to change. And of course, if data changes, we also need to probably, yeah, fix or adjust something in our model in order to all the time guarantee that we are going to get the best insights and the best predictions. That being said. So, so there's obviously this could be used really on any kind of data that's within Salesforce first right, Is that a fair statement? So one of the use cases I've had with a couple of different clients is about sales data or client data. So they're trying to look at patterns in say a buying cycle or renewing of contracts if it's a a subscription based type of solution or trying to figure out what customers might buy. Or might have a higher probability of buying something else in the future. So as I understand what I think you just said. Is that as the business evolves, so let's say we get a prediction score of you know 75 for or higher for something. If we can improve that number because you're working with real time data, it's going to change the variables going forward because we're going to increase probability in certain circumstances. And as the business evolves and let's say we improve our customer renewal rate, we're going, it's going to find the next most likely renewal. And and move those people up in priorities that fair they or did that make sense? Yeah, exactly. Basically with this tool we will have access to either standard objects and custom objects. So any customization that you have already in yourself or or it's going to be available there for instance and you are going to select that object and based on those record you can start analyzing those type of trends that are going to be the recommendations, in this case the predictions. And based on those predictions, yeah, that would be, yeah. And our next episode, because we have also in Selfhost, next best action that is going to take the predictions and it's going to recommend strategies and actions to the users. Yeah. But you know, this is like a really impressive big field that it's continually growing. And first of all, I guess that this is a really key point to. To start taking advantage of your data, you need to probably start trying to make those next steps to move forward with your data, because this is also related to. Yeah. The maturity level that you have related to your data because there are a lot of companies that are extremely conservative. So in that way they are not that open to explore these possibilities. They will rather rely on reports, dashboards that tools that of course are effective, but probably you can work with this more efficiently to have the same results.