Podcast host Karyna Mihalevich, CPQ Functional Lead at CLARITY, welcomes Matthias Hirsch, Senior Product Manager in Customer Experience & Industries at SAP. Drawing from his vast experience since 2007 in the Discrete Manufacturing Industry and his crucial role at SAP since 2016, Matthias provides a distinct viewpoint on the role of AI in sales, particularly in enhancing and streamlining the sales of complex products. Today’s episode continues the discussion on the important topic of the transformative power of AI in configurable products, a subject that was extensively explored in a previous Quote-to-Cash podcast.
Our guest Matthias Hirsch, Senior Product Manager in Customer Experience & Industries at at SAP and the host of the podcast Karyna Mihalevich, CPQ Functional Lead at CLARITY.
Content
Key Functionalities of SAP’s Intelligent Product Recommendation
SAP’s Intelligent Product Recommendation, as Matthias Hirsch explains, is among the latest SAP product innovations that has already begun to transform the approach to sales in various industries. This innovative system is structured into two critical sections:
- the runtime
- the design time applications.
The runtime aspect enables the application of machine learning within CRM and sales processes, allowing users to interact directly with the system. Meanwhile, the design time component functions similarly to a product configurator, where users can model rules and design user interfaces tailored to their specific sales needs.
A key feature of this system is its ability to handle the unique complexities of industrial manufacturing and equipment sales. Products in these sectors, such as robots, forklifts, heat exchangers, pumps, and compressors, often have individual data models, structures, and domain models. The Intelligent Product Recommendation system is flexible enough to accommodate these variations, thereby ensuring that the recommendations are as accurate and relevant as possible.
Matthias uses an example to illustrate this point: a scenario where a shop floor manager in the automotive industry needs a new painting robot. Unlike traditional approaches where customers specify exact models and accessories, this system allows customers to express their needs – such as the number of car bodies to paint within a certain timeframe and the requirement to switch colors frequently – and then translates these needs into technical specifications for the robot. This marks a significant shift from standard product configurations, as the system starts with customer requirements and then suggests suitable products based on those needs.
Furthermore, Matthias highlights that the system’s integration into various business processes, such as CPQ, commerce, or CRM systems, is a part of its runtime functionalities. This integration is crucial as it allows the system to be a part of the overall sales process, from generating a recommendation to finalizing a product proposal.
Additionally, SAP provides pre-trained models that users can utilize immediately, offering initial recommendations based on general data. However, for more tailored results that align with a company’s specific historical data or product catalog, SAP has made it possible for users to train their own models without needing extensive programming knowledge. This feature democratizes the use of advanced machine learning sales tools, making it accessible to admin users or end-users who may not have a technical background.
Integration with Other Business Platforms
As Matthias Hirsch details, the SAP Intelligent Product Recommendation system provides comprehensive integration capabilities with both SAP and non-SAP solutions, establishing it as a versatile instrument for diverse business applications. This seamless integration is pivotal to its infrastructure, facilitating sales process automation and ensuring streamlined operations with improved efficiency throughout the sales cycle.
Matthias distinguishes two primary types of integration relevant to the system:
- systems of record
- systems of engagement.
Systems of record refer to platforms where essential sales and transaction data are stored. This data is vital for training the machine learning model underlying the product recommendation system. Examples of such systems include S/4HANA, ECC, CPQ systems, or any commerce systems where historical sales data like quotes, carts, and orders are stored. The SAP Intelligent Product Recommendation system integrates with these platforms to extract necessary data, laying the foundation for accurate and effective product recommendations.
On the other hand, systems of engagement are the platforms used by sales personnel in their daily operations. These could be CRM solutions like SAP Sales Cloud, CPQ solutions like SAP CPQ, or various commerce solutions. The integration with these systems is crucial, enabling sales teams to access and utilize the product recommendation results, thereby facilitating sales process automation and streamlining sales activities.
A key feature of the SAP AI integration is the API-first paradigm.
A key feature of the SAP AI integration is the API-first paradigm. This approach ensures that users get a runtime user interface (UI) that dynamically renders based on the underlying data model, similar to what is done in LO-VC or AVC. This pre-designed UI significantly reduces the need for custom UI design, streamlining the implementation process. However, Matthias points out that the flexibility of the system extends to allowing businesses to use their own UIs, integrating them through the APIs provided by SAP.
This dual approach to integration not only allows for the use of backend configuration systems, known as systems of records like S/4HANA or CPQ, but also facilitates integration with systems of consumption. As part of the SAP suite, the Intelligent Product Recommendation system seamlessly fits into the broader SAP ecosystem, enhancing its utility and applicability across various business contexts.
Ideal Users for Product Recommendation System
When considering the ideal users for SAP’s Intelligent Product Recommendation system, Matthias offers insightful perspectives on its suitability across different business scenarios. He acknowledges that the system is particularly effective for complex product sales that require a nuanced understanding of customer needs and technical specifications.
Matthias points out that machine learning, the driving force behind this system, is inherently data-driven. For effective implementation, businesses would ideally draw upon substantial historical sales data, spanning the last 5 to 10 years, to train the initial model. This requirement can pose a challenge for some customers, particularly small and medium enterprises (SMEs), who might not have extensive sales data at their disposal. However, SAP addresses this challenge by offering methodologies to augment naturally grown data with synthetic data, enabling these businesses to leverage the system effectively.
Delving deeper into the question of which customers might not be the best fit for this product, Matthias flips the perspective to emphasize those who would benefit the most. He explains that the system is ideally suited for complex product sales where technical intricacies and customer-specific requirements are paramount. This contrasts with sales environments focused on emotional or consumer-like buying experiences, where the product recommendation might be more straightforward and less technical.
Matthias further elaborates that the Intelligent Product Recommendation system excels in environments where products are not typically sold from stock but are instead customized or produced in a lot size of one process. This customization aspect is crucial for businesses dealing with complicated products requiring detailed configuration, such as industrial equipment or specialized machinery.
Real-World Applications
Despite its recent introduction to the market, SAP’s Intelligent Product Recommendation system is already making waves in real-world applications. Matthias shares that he had presented at the DSHE summit in Germany, a prominent gathering for the German-speaking SAP user community. There, he discussed a compelling use case involving a customer they work with.
This particular customer specializes in the creation and construction of tap changers, a critical component used in large electricity transformers. The challenge they faced was intricate: they needed to accurately understand the specifications of a transformer to design an appropriate tap changer.
“This scenario is a quintessential example of where our Intelligent Product Recommendation system plays a pivotal role. Utilizing this innovative tool, we could provide tailored suggestions on the most suitable tap changer for a given transformer, considering specific customer requirements. This use case goes beyond merely showcasing our product’s technical capabilities; it’s about illustrating the effective implementation of AI-driven sales strategies in complex sales processes, transforming the way products are matched to customer needs.”
Additionally, Matthias notes that SAP is engaged in various proof-of-work projects with other customers. These collaborations are vital as they assist the clients in harnessing the power of AI for their sales strategies and enable SAP to build robust references. These references are vital for establishing a more robust presence in the industry, demonstrating the practical advantages and tangible success of SAP Intelligent Product Recommendation in harnessing AI in sales. They underscore the efficacy of AI-driven sales strategies and stand as evidence of the system’s capacity to unlock new opportunities in machine learning, especially within the sales sector.
Adapting to Dynamic Product Catalogs
In business technologies, one of the most significant challenges is adapting to dynamic product catalogs. The flexibility to adjust to changing product specifications and offerings is crucial. Matthias explains the concept known as the “cold start problem” in machine learning, which is particularly relevant when dealing with dynamic product catalogs. He describes the approach of SAP’s system as a preemptive step before launching the product configurator. The key, according to Matthias, lies in focusing on the ‘hard factors’ of a product – those elements that are essential and less susceptible to change – over the ‘soft factors’, which are more prone to frequent updates.
For instance, when certain fundamental aspects of a product like its core functionality or key components remain constant, despite changes in other attributes like color or minor features, the system can adapt more smoothly. In scenarios where entirely new components are introduced, Matthias outlines two strategies. Firstly, the system can gradually learn and adapt through a feedback loop, where it’s retrained with new data, such as monthly sales quotes. This method ensures that the model stays current with market trends and customer preferences.
The advantage of using synthetic data is its immediacy; it allows the machine learning model to quickly integrate new parameters without waiting for extensive real-world data accumulation.
Alternatively, for a more rapid adaptation, Matthias suggests the creation of synthetic data sets by technical product managers. These data sets are designed to reflect potential changes or new features in products. The advantage of using synthetic data is its immediacy; it allows the machine learning model to quickly integrate new parameters without waiting for extensive real-world data accumulation.
This capability is not just confined to SAP’s technical team; it’s accessible to end users as well. Every technical product manager, equipped with knowledge about their product’s operational range and specifications, can develop these synthetic data sets. This proactive approach enables a swift response to market changes, ensuring that the SAP Intelligent Product Recommendation system remains a robust and adaptable tool for sales in a dynamic market environment.
Future Innovations in Product Recommendation
As SAP’s Intelligent Product Recommendation continues to make its mark in the realm of AI-driven sales, future innovations are already on the horizon, poised to further enhance its capabilities. Matthias emphasizes the potential integration of generative AI into the SAP Intelligent Product Recommendation system. He notes that while large language models like Bard or GPT-4 have their strengths, they also have limitations, particularly in tasks like classification crucial for product configuration. However, these models can play a pivotal role when used in conjunction with specialized neural networks.
An innovative example Matthias presents is the use of a chat function or email for product configuration. In this scenario, a customer’s email detailing their needs could be directly fed into the Intelligent Product Recommendation system. The system would then employ a large language model to extract the necessary information and map these needs against a specialized machine learning model. This process not only simplifies the task of configuring a product but also ensures that the recommendations are highly tailored to the customer’s specific requirements.
The integration of generative AI models extends beyond just parsing customer needs.
The integration of generative AI models extends beyond just parsing customer needs. Matthias envisages a system where, once a machine-generated recommendation is made, it can be further refined by feeding it back into a generative AI model. This model would then consider the customer’s context and the initial recommendation to create a custom-tailored value proposition, significantly enhancing the personalized aspect of the sales process.
This approach to product recommendation represents a fusion of the broad intelligence of large language models and the precision of specialized machine learning algorithms. It exemplifies the synergy between different AI disciplines, aiming to provide a seamless experience where the end user need not worry about the underlying complexities.
Looking forward, the vision for SAP’s Intelligent Product Recommendation is one where the sales process is not just about transacting but also about building relationships and understanding customer needs on a deeper level. By automating repetitive tasks and enhancing the quality of customer interactions, sales representatives can focus on higher-value tasks, such as fostering customer relationships and personalizing AI-driven sales strategies.
The Bottom Line
The SAP Intelligent Product Recommendation system is a standout example of SAP product innovations, illustrating the significant role of AI and machine learning in business, especially in the sales sector. With its ability to automate complex product sales, seamlessly integrate with various business platforms, and adapt to dynamic product catalogs, it’s a critical driver in shaping the future of sales technology. This system not only streamlines complex sales processes but also opens new avenues for personalized customer interactions and more efficient AI-driven sales strategies.
Karyna Mihalevich effectively highlights the transformative potential of these innovations, observing a notable transition from labor-intensive, manual methods to streamlined, AI-driven, custom-fit proposals. As the embodiment of SAP’s commitment to advancing sales technology, the future of the SAP Intelligent Product Recommendation system is geared towards revolutionizing and personalizing the sales experience. This leap forward in AI and machine learning marks the dawn of a new era in product configuration and recommendation, expanding the horizons of what’s possible in the domain of sales.