Sales forecasting and sales opportunity discovery using AI and deep learning
Selling is a process, especially in a B2B context, in which human interactions play a crucial role. AI and Deep Learning play an important supporting role in uncovering opportunities and thus support these interpersonal relationships. Sales teams often generate and have access to large amounts of data. However, given the amount of data, it is difficult for people to detect all relevant opportunities. And that is exactly where AI, as forecasting software, can add value.
This was also the view of one of our customers who engaged the Trendskout AI software platform to exploit sales opportunities. This concerns a production company based in Belgium, employs about 40 employees and has an annual turnover of 20 million euros.
Customers vs product ratio
This organization has a very broad customer base of active customers. The generated turnover is comparable to the classic 80-20 distribution: 80% of the turnover is generated by 20% of the customer portfolio. In addition to a number of large customers, there are a large number of smaller customers who only place an order sporadically. On the other hand, this company has a solid but rather limited product portfolio. Each type of product is available in an extensive diversity of shapes, sizes, colors and quantities.
A team of account managers is called upon to serve the existing customer base. Their task is to maintain close contacts with all customers and to upsell where possible. Each representative manages a portfolio with hundreds of active customers, so it is impossible to identify all possible sales opportunities themselves without assistance.
6 months after rollout, total relative sales increase was 8.5%
Discovering sales opportunities using AI
To uncover all this untapped potential, an AI software company was sought, with experience in sales forecasting, who could get to work with the available data without months of preliminary study or integration process. The choice fell on the Trendskout AI and Auto ML software platform.
This customer has been working for some time with a Microsoft ERP package in which all transactions with customers of the past years were tracked. Since the Trendskout AI software has ready-made plug-ins for common software systems and databases, the connection could be quickly established without custom programming. This was an important element in the choice of the Trendskout AI software.
As a result, the connection with the data went smoothly and it was therefore possible to quickly start building the AI model. The Trendskout analysis “Sales Opportunity Detection” was chosen for this application. With this analysis, the Trendskout AI searches for patterns in historical data to create future projections from there. To this end, a sales forecasting model is built based on all available order data from the past. From this model, the Trendskout AI software then generates a monthly forecast indicating which customers should focus on in order to guarantee optimal turnover.
Design of the AI within Trendskout
This analysis was set up according to the intuitive CONNECT–ANALYSIS–AUTOMATE flow which is characteristic of the Trendskout software. As training data, a historical data set from Microsoft Dynamics was chosen that contained the transaction & order data over a period of 3 previous years. For the real-time analyzes and forecasts, the connection was established with Microsoft Dynamics ERP, as mentioned above.
As AI functionality, the “Sales Opportunity Detection” analysis was chosen, which is a ready-to-use AI analysis in Trendskout. The Trendskout AI software generates a custom AI model based on the supplied data. During the configuration you can choose over which period the machine must predict the opportunities.
As an automatic step, Trenskout generates the data in tabular form, accessible to the sales team via an interface that can be easily consulted in a browser. This generates, for each sales rep, a list of customers ordered by priority where action must be taken in the following month, supplemented by the expected turnover per customer.
This complete flow was set up within the visualized interface of Trendskout, without the need for any coding and within a period of 5 weeks from start-up to delivery.
Determining the “next best action”
The suggestions generated by the artificial intelligence soon proved to offer clear added value. Each account manager now clearly knew which customers to focus on each month, and this soon paid off. After the application went into production for 3 months, a relative sales increase of 6% was recorded on a monthly basis.
After the added value of the solution was demonstrated, the customer wanted to further deepen it. The AI model generated a monthly overview of the biggest opportunities and with which customers they could be found. However, each seller still had to determine the correct follow-up action – mail, phone call, personal visit, offering a promotion …
To support this as well, the Trendskout AI was used to build a Next Best Action model that suggests concrete actions for a predefined goal: in this case, generate more margin. In concrete terms, a list of actions for each customer that will most likely realize a sale. This reduces the administrative role of the salespeople in the pre-sales phase to an absolute minimum and allows them to focus on developing deeper relationships with their customers.
Clear increase in turnover as a result
The addition of this analysis resulted in an additional turnover increase of 2.5%; 6 months after the roll-out of the initial AI flow, which brought the total relative turnover increase to 8.5%.
If you also want to get more out of your data and your sales process, we are happy to assist you.