Lead scoring
Lead scoring is one of the prominent actions performed by AI in marketing automation to find potential leads based on the behavioural analysis of the previously converted leads. As businesses strive to become more data-driven, AI-powered lead scoring has become an essential part of today’s marketing. By automating the analysis and prioritizing leads, this AI-powered lead scoring system can help sales and marketing teams to focus on the most potential opportunities.
Lemnisk creates the most advanced fully functional lead-scoring system powered by AI that precisely analyses the history of data from multiple data points and segregates the potential leads scores that have the propensity of converting into sales. The machine learning algorithm tracks the behaviour of previous leads and analyses the data from the data points collected from the journey of previously converted leads. These data of previously converted leads can be analyzed from the journey and the actions performed on the websites to find the potential leads as per the score that was achieved based on the history of their journey. Based on the predicted score, each lead will be classified into hot, warm, and cold buckets which will be ingested into the CDP to support conducting different campaigns for the predicted leads.
This supports the sales team to focus on the leads with a high score rather than focusing on the leads with the least score. This predictive lead scoring saves a lot on your marketing budget and your efforts by focusing on the predicted data that has the potential to convert and make purchases.
Purpose
To prioritize lead follow-ups and spend resources on the more predictable deals.
To differentiate lead nurture approach, messaging, etc. based on the lead bucket
To understand what predicts the propensity of a lead to convert and utilize it in other analytics + modelling
The machine learning algorithm mainly focuses on making the AI perform better in the lead scoring system. The algorithm will collect enormous data from various data points of the platform in which the user has performed a set of activities and got converted into a potential lead. This data and the behaviour of the lead will be analyzed by the algorithm. By analyzing this large amount of data the algorithm creates the scores of each lead that has the potential to get converted. The algorithm will be constantly exposed to new data collected by the platform to learn new behaviours of the user and the behaviour of the user to predict the score of the lead and its potential to achieve sales.
Use Case
User A constantly visits the websites, shows interest in the product, and reads numerous posts about the products which show the possibility of making some purchases based on the behavioral analysis from previously converted leads. User B rarely visits the website and shows the least interest in the products. The algorithm will analyze both users' behaviour and gives a score accordingly. Based on the scores, users can be bucketed into hot, warm, or cold, and further campaigns can be created to nurture the leads and support making sales decisions like,
Predicting the activities that are most likely to lead to a sale.
Making offers that are most likely to result in a conversion.
Finding the best way to follow up with a lead.
Also, Lemnisk is capable of tracking the following behaviours,
Hours since the users drop off from different stages.
Different stages of page visits
Hours since the user dropped off from different products
Total product visits
Total page visits
Users entered their age
Users searched on Google
From the following data points:
URL
Page
Utm source
Users entered city
User entered province
Cities derived from IP address
Internet service provider
Operating system
Browser
Device
Day of week
Hour of the day.
Lead Nurturing:
The leads that fall under the warm and cold bucket will be nurtured by setting specific campaigns to try converting these leads into the hot leads that have the potential to get converted.
Once the model is validated, cold and warm leads will undergo nurturing and validation, prior to the assignment to a financial advisor to ensure that they are ready to purchase.
Hot leads will be assigned directly after generation for immediate action. They will still receive nurturing communications to help advisors to close the lead.
All unconverted leads will be returned to the nurturing pool for re-nurturing. Ex: Invitation to webinars, use for retargeting campaigns, etc.
All data of leads will be used on media targeting campaigns for us to have a higher probability of generating high-quality leads.
Highlighting capabilities of Lemnisk's Lead Scoring AI
Lemnisk makes the lead scoring more capable of making precise predictions for optimal conversions with the following features.
Online conversion data, available with Lemnisk can be enriched with offline conversion data (conversion that happens through the call centre, agents, etc) for better modelling and predictability.
Available variables/features will be filtered based on their predictability power and only relevant features will be fed to the model.
Outlier detection is done to remove anomalous leads while modelling. This further leads to better predictability of future leads.
Multiple machine learning models like Logistic regression, Random Forest, and Gradient Boosting algorithms will be evaluated with different sets of hyperparameters and the best-performing algorithm will be used in the final scoring of the leads.
The model will be evaluated on multiple metrics like ROC AUC (Area Under Receiver operating characteristic) score, Precision, Recall, and F1 score. Thresholds for different buckets can further be tuned based on these metrics.
Model interpretation will be done to understand how the model is behaving with different input features and check the impact of the variable in the output score.
Use case 1
This use case interprets the detailed analysis of lead scoring AI using one of the clients of Lemnisk. This use case interprets each and every behaviour of the leads in detail by tracking the performance and analyzing the historical records.
SHAP findings that interpret the behaviour of the users on product pages.
SHAP findings that interpret the HSDO( Hour since a user drop off) from ePlan pages
Demographics that show the leads' browser and ISP contributions
While Chrome (web and mobile) and Safari bring the highest traffic, there are no material differences in users converting better with one over the other.
Similarly, PLDT and Globe Telecom are the most used ISPs, but there is no material difference in conversion rates.
Use case 2
A client of Lemnisk used the platform to target the right audience to upscale their business, the lead-scoring AI supported the client in predicting their leads and categorizing them into Hot, warm, and cold leads to support engaging the optimal leads for better conversion. As a result, the prediction performed so far.
The chart shows the result of leads and conversions between the period of December to March. As per the result, the conversion target achieved is 0% and the KYC target achieved is 19%, from the cold bucket. From the warm bucket, the conversion target achieved is 1.8% and the KYC target achieved is 17.3%. And from the Hot bucket, the conversion target achieved is 1.5% and the KYC target achieved is 20.3%with these data, Lemnisk supported boosting sales by facilitating exact targets to make the conversion and also supports setting separate campaigns and nurturing the lead from warm and cold bucket to hot leads. The client faced a lag in KYC, so Lemnisk took a step ahead and triggered the scored leads to perform KYC to support the client in gaining customer data for further communication and engagements.
Last updated
