Machine learning is a technique of data science that helps computers learn from existing data to forecast future behaviors, outcomes, and trends. Machine learning and AI touch your life, every minute of every day, be it the applications you use at work or how you choose products to buy (Amazon). Even who you marry or date (Match.com, Tinder etc.,)
All technology firms are now focused on AI – the rise of smart machines that augment human capability. McKinsey report estimated that activities taking up more than 20 percent of chief executive’s working time could be automated by an intelligent assistant, saying – “I really need that!” (Source: ft.com)
At Fiind, our vision is to simplify the work of Marketers and Sales personnel, so they could focus on closing the deal. We believe that more than 80% (subject to discussion) of the working time of Sales folks could be automated by intelligent assistants that act as the cerebral system of CRM.
What is Machine learning and how is it different from AI?
AI means getting computers to do all the things that it takes human intelligence to do like reasoning, understanding language and the visual world, navigating, and manipulating objects. Machine learning is a sub-field of AI that deals with the ability to learn. Learning is the one thing that underlies all the others.
What does a machine learning algorithm “look” like?
There isn’t just one algorithm. We have many algorithms and approaches today – based on statistical approaches (e.g. Bayesian learning, multinomial logistic models), evolutionary techniques (genetic algorithms), logical induction (boosting algorithms) and approaches that mimic the brain (neural networks). No one algorithm is good at everything. At Fiind, we use a combination of models, most commonly referred to as ensemble modeling.
How is Machine Learning used at Fiind?
First and foremost, we use machine learning for lead intelligence. In other words, identifying leads that are sales ready. The process touches five primary areas, here’s a snapshot where we use ML.
- Start with clean Customer data. We are looking at 3 lead sources. Fiind Library with 2M+ companies, Customer House list and External leads such as the leads from Media Planning. Fuzzy logic algorithm is used to match company names to websites. There are numerous models that work in tandem to identify missing values in firmographic / demographic data
- Discover signals. Now we need to identify which companies are sending the signals. Some examples: companies that are hiring for media planners, companies that just had a new CMO or VP of sales or marketing. This also involves creating new features that are more powerful in the model.
- Identifying the signals from the existing customer transactions to determine the look alike/ pattern recognition. We use adaptive boosting algorithm and association rules to arrive at the scoring. A typical workflow as described by Rubens :
- Use the market data (wins/losses/no-interests) to improve the models. A closed loop system with the customer apps will provide us real-time data from market tests. The error function will pick this up and adjust the model in near real-time.
- Use the user feedback from the tool (example, someone likes the data or corrects it) to improve the accuracy. The feedback icon is integrated into all our apps and makes it easy for the user to provide their input. This helps in changing the features/variables that are input to the model.
