Case study - How 4KST Machine Learning helped a bank increase its sales

How Machine Learning helped a bank increase sales

Every company that works with sales targets and forecasts needs tools capable of identifying the best business opportunities that are measured not only by the size of the deal, but also by the chances of closing the transaction.

Machine Learning, one of the branches of Artificial Intelligence, is a powerful technology for making these kinds of predictions, although few companies use it for this purpose.

The following case is a good example of how a bank was able to significantly increase its sales after adopting 4KST Machine Learning to predict which business opportunities were most likely to sell and thus manage their actions and know how far they were from their targets.

The financial institution's problem was knowing which deals to focus on. Its sales pipeline within the CRM (a tool that displays and records a potential customer's entire buying journey), although it indicated which sales stage a particular purchase or deal was in, didn't show which ones had the greatest chance of being closed.

To solve this obstacle, 4KST has developed two sales propensity models, one monthly and one quarterly. Each model indicates a score (i.e. a mark from 0 to 1,000) that measures the likelihood of a deal closing in the month (monthly model) or quarter (quarterly model).

"Based on these two scores, salespeople and sales managers will be able to look at the deals that are most likely to be concluded and determine their action strategies. If they have a monthly or quarterly target to hit, they'll know exactly how likely it is that they'll achieve it and can focus on the opportunities that have the best chance of hitting those targets. This way, sales managers don't just use the salesperson's "feeling" to know where they should focus and they have a powerful ML tool in their hands. " explains 4KST founder Riccardo Lanzuolo.

See real results:

Of all the negotiation proposals, the bank closed an average of around 20% per month. The predictive model developed by 4KST made it possible to discover that half of these sales were concentrated in deals that received scores above 800 points. What's more, deals with a score above 950 had a 98% chance of closing.

As for the proposals that received a score between 650 and 700, this probability was between 65% and 70%.

By looking at the previous month, the manager knew which score led to the closure and so he knew how far or close he was to his target.

This way, guided by a machine learning model that learned from the behavior of his sales and leads, the manager knew how to guide his actions with his team.

click on the image to enlarge

click on the image to enlarge

Faced with such precise and precious figures, salespeople were able to make rational decisions based on statistics, discarding negotiations with a low probability of success and optimizing efforts, time and bringing better results for the company.

As we've seen so far, Artificial Intelligence can be applied to statistical models to make increasingly accurate predictions in practically any company in any sector.

The case above refers to sales forecasts, but it is possible to mold the model for other purposes such as calculating a customer's credit score, the liquidity score of any vehicle or the school default rate, among others.

Our predictive models stand out from most of those available on the market because they are customized. This means that our proprietary algorithm is adaptable to any market niche. 

Are you interested in our solutions for your company? Talk to a sales consultant and find out more about our products!