How to build a lead scoring AI using XOKind no-code platform
Are you looking to transform your historical data into an AI model capable of predicting the outcome of new leads? In this article we will explain how you can build your AI model with no code using the XOKind AI platform.
Step 1: Select your model in the App Zoo
Go to ai.xokind.com to access the platform.
Select the “Lead Scoring” model to start building your AI.
Step 2: Connect your historical data
Use your tabular historical data to create and train your custom lead scoring AI. Via the XOKind AI platform, you can simply connect to your Snowflake database or upload your data in CSV format.
In this example, we will use the UCI public dataset.
How to Prepare your data?
Gather all your historical data on the action you want to predict. Make sure to add a “Conversion” column to train the algorithm and annotate if the conversion happened with a YES or a NO.
Once the conversion column is saved. You can upload your csv file to the platform
Step 3: Build & Train your model
This is usually a complex and long step while building your own lead scoring AI. In this stage you will need to select the column to use as “input” in the model and the column you want to predict (the output).
Input columns: historical data points on conversions that happened or failed. Select as many columns at first, in order to let the algorithm identify what really affects your conversion rate.
Output column: the outcome you would like to predict for your next leads.
Ignored columns: Non-relevant data points that do not impact the conversion such as the user ID
Once the inputs and outputs are set, you can now click on the SAVE AND TRAIN button to build your model.
You have the option to select your model presets to tradeoff between speed and accuracy.
Note: If your training data set is large or you selected “Best Quality” it might take up to 1 hour to create your model. You will receive a confirmation email once your model is ready for the next step.
Step 4: Review Results & Deploy
At this stage, your AI is ready! You can review your model accuracy, and improve it by fetching new data to train your model.
You will also learn about the SHAP values to better understand what data columns impact the decisions the model is making.
If you are satisfied with your model, simply click on Deploy to move to the next step!
Step 5: API Endpoint + Playground
Now it’s time to use your custom lead scoring AI model and make some predictions!
Using the playground:
You can enter manually the input data of a new lead to predict in real time the output by clicking on Predict
Using the API endpoint
You can also integrate your AI model within your product or tool using the API endpoint.
Simply select between React (JS) or cURL to generate the code, copy and paste where you want to integrate your Lead Scoring AI.
How to use a lead scoring AI?
Your AI project should be defined for a specific use case, so you can build and train a model specifically to predict an outcome that will trigger action or identify opportunities.
Lead scoring AI, most popular use cases:
- Trigger discounts only for visitors not likely to convert for online retailers
- Power an intelligent paywall when your visitor is ready to convert for SaaS companies
- Identify and prioritize leads most likely to convert for sales teams
- Create a custom fraud detector for your business
Turn your data into predictive insights in record time
Need help defining your AI project?
Book a 15-min consultation with an AI expert to identify the best use cases for your business.BOOK CONSULTATION