Creating the advantage through automated processes and artificial intelligence.
Sector: High Tech
Company size: Fortune 1000
Performance Area: Customer Service
Goal: Deflect human intervention from the company’s support engineers and deliver high customer satisfaction with relevant online technical support portfolio.
Challenge: Existing AI implementation to the company’s online technical support portfolio only achieved a 12 percent relevancy of producing the correct answer to customers’ searches. From the company’s 50,000 active users, the average number of searches per month were nine. The company’s average cost for human intervention and have a support engineer assist the customer is $800.00 per ticket.
Results: After implementing Sonasoft’s AI Solution, the company increased its search relevancy from 12 percent to 60 percent with the best answer as the number one search answer. When the results were expanded to include the best answer within the first five results, then Sonasoft’s AI Solution scored over a 90 percent relevancy rate. With more input to aid the system’s machine learning capabilities, Sonasoft’s AI relevancy scores are expected to increase.
Company size: Large bank with over 500,000 customers
Performance Area: Non-preforming Assets (NPA)
Goal: Reduce the number of loan defaults and focus on lending to prospects that have a strong probability of paying off the loan
Challenge: Although the bank conducted due diligence and screened all loan applicants, it still found that it had a significant percentage of loans that defaulted. To gain insights and try to prevent non-preforming assets, the bank implemented Sonasoft’s AI solution.
Results: Sonasoft ingested data of over 500,000 borrowers across 44 variables to determine the root cause. Sonasoft’s AI Solution then was able to provide a formula with 72 percent accuracy that determined if a loan would be paid or default.
The Sonasoft AI Solution also revealed some interesting insights to the variables that could determine whether a loan would default or not. These discoveries included the type of sub-grade loan, grade, years of employment, debt purpose, etc. that could determine whether the loan would be paid or defaulted.
Company size: Large company with significant infrastructure
Performance Area: Customer Support
Goal: Forecast the number of customer support tickets with particular attention to tickets that fall under the ‘password reset’ category.
Challenge: The company needs to forecast the amount of recourses to quickly resolve support tickets. Since the e-commerce site is a B2C, immediate customer satisfaction is necessary to remain relevant; after all, customers are just a click away from going to a competitor. However, the company cannot overstaff its support team either. The company also realizes that some events and times of the year will cause a disproportionate amount of support tickets. Therefore, the company needs to be prepared for these types of fluctuation with adequate and flexible resources.
Results: Sonasoft ingested data and applied deep learning along with seasonal decomposition. The best AI solution gave about 75 model of accuracy. This allowed the company insights on how best to staff and place resources in order to achieve high customer satisfaction and still be efficient in operations.
Company size: Global Manufacturer of Electrical Products
Performance Area: Forecasting sales within its CRM
Goal: Wants insights into the probability of winning opportunities in its CRM and optimize the likelihood of completing a sale
Challenge: The company had a lot of deals in its sales pipeline that were tracked within its CRM. Letting its sales staff predict the probability of closing these deals was not working. The company then had to find a way to accurate predict its sales forecast. Therefore, it turned to Sonasoft’s AI Solution to gain insights.
Results: Sonasoft ingested data and engineered its AI solution for new variables that could capture the events such as ‘how many previous opportunities were won for the account in consideration’, etc. The Sonasoft solution also looked for interaction between other variables including accounts and opportunity owner, products and opportunity owner, etc. Sonasoft’s AI Solution achieved 82 percent accuracy in identifying whether the opportunity was won or lost. It also gave insights such as which account is handled by which opportunity owner (sales rep) or which product is best sold by which opportunity owner, etc.
Sector: Large Retailer of B2C consumer goods
Company size: Global Company
Performance Area: Forecasting Sales and Demand
Goal: Accurate forecast sales and demand; AI will benefit not just revenue but also logistics of maintaining sufficient inventory
Challenge: The company needed to forecast its revenue as well as its inventory. The benefits from the revenue forecast are obvious, but accurate inventory forecast equally affects the company. Too low of an inventory will impact sales and customer loyalty. Too high of an inventory creates a space problem that prevents new items to sell. It also creates downward pressure to discount the items in order to move them at a reduced profit or even loss.
Results: Sonasoft ingested data on historical sales, weather information, public holidays, specific promotions, price cuts, etc. Sonasoft then ran multiple algorithms for statistical insights and deep learning. The best benchmarking solution was chosen with an 80 percent modeling accuracy. In addition, Sonasoft was able to reveal which product promotions worked best and which ones did not. Overall, Sonasoft was able to give insights to drive input variables to maximize sales.