AI in sales Precise market prediction through artificial intelligence

Related Vendors

FAULHABER Drive Systems

Taiwan Machine Tool & Accessory Builders' Association (TMBA)

Modern supply chains and uncertain markets often make reliable demand forecasts a lengthy task for companies. The start-up Paretos specializes in AI-based market forecasts and demonstrates with Wika how production needs can be predicted.

The topic of artificial intelligence is currently coming across us everywhere - including at many industrial locations in Germany. Paretos, a start-up from Heidelberg, has used AI to create precise market forecasts. With their forecast and decision platform, they offer companies the opportunity to optimize their market forecasts based on AI. Since August 2023, Paretos has been working with the measurement technology company Wika. The aim of this cooperation, according to Fabian Rang, CTO of Paretos, was to better understand future sales volume. For this, more precise and automated processes are being developed, making the planning of resources, materials, storage, transport, and personnel more efficient and transparent.

Why an AI-based evaluation instead of a manual forecast?

Paretos' forecasting tool serves firstly to relieve employees, as it eliminates the need for complex data analysis and the calculation of forecasts. Secondly, the AI is more easily able to recognize patterns within the large amounts of data. These patterns are in turn significant for setting up a forecast. "For very complex decisions along the supply chain, there are more possible combinations than stars in the universe; a person can no longer keep track of them," says Rang. Through the more precise forecasts, the supply chain can be planned better as a whole. This leads to further advantages, such as:

• the saving of costs for transport and logistics,

• optimizing working capital and

• reducing overstocks, through increased accuracy in demand forecasting.

The use of artificial intelligence is also becoming increasingly important with regard to market competition. Currently, the use of AI is still a competitive advantage, but this will change over the next few years. According to Rang, the use of AI will soon become a hygiene factor. This means that data evaluation without the use of AI is significantly less accurate. "Anyone who does not use AI to improve forecasts will lose market share in the long run or disappear from the market entirely," says Rang.

What should be considered when using AI forecasting tools?

In order to obtain a valid forecast, Paretos uses a decision intelligence platform. This can derive suitable optimization measures and future scenarios using data and key figures. For example, stock levels or product and promotional prices are included, as well as external data such as forecasts and recommendations, or information on consumer sentiment, holidays, school vacations, and industry-specific action days. In order to draw reliable forecasts from this data, various predictive analytics methods are also used.

How was it done?

To generate a precise forecast for Wika, Paretos' forecasting engine was used. With this, machine learning models of various classes were trained on the Wika dataset. Statistical, machine learning, and deep learning models were used. In the end, a variation of a machine learning model achieved the best results for Wika's needs. In the next step, forecasts, so-called predictions, were generated at a low level of granularity, meaning with less depth of detail and content. These include for example the product family and the brand. Subsequently, the decisive comparison took place between the AI-predicted data and the globally aggregated data. In these, the predictions were on average significantly more precise than the benchmark values at Wika. In concrete numbers, the prediction benchmarks for comparable customers could be exceeded by about 30 to over 80 percent, according to Rang. However, more precise estimates can only be made after several months.

Currently, the project between Wika and Paretos is still in the roll-out phase of the Sales Forecast. As soon as this is completed, Paretos plans to expand the collaboration along the supply chain. For this, the first step will include capacity planning in manufacturing. This includes, for example, forecasting the number of units, personnel planning, and utilization of production lines. In the second step, purchasing will be analyzed in terms of demand planning and warehouse optimization. According to Rang, this approach can fully exploit the business potential in these areas.

How was the new software received by Wika's staff?

Even at the beginning of the pilot project in August 2023, there was great interest in the results and the potential that the Sales organization's solution offers. Therefore, the plan is to gradually roll out the forecast to the global Sales organization as part of the roll-out. In general, Paretos perceives a healthy skepticism from almost every customer at the beginning. Particularly among those who have to work with the forecasts afterwards and make decisions based on them. As Rang explains, this mistrust can only be turned into trust through transparency.

Which other industries can consider using AI-based forecasting?

The company Wika primarily deals with pressure and temperature measurement technology. However, a software like that of Paretos could be considered for other industries as well. The start-up from Heidelberg has already worked with well-known companies from various industries. This includes a cooperation with the fashion company Armedangels. Within this collaboration, the demand forecast could be optimized and the surplus reduced. But also German forwarding service providers like Hermes or GLS are already cooperating with the AI company. In conclusion, Rang says: "What works so well for optimizing logistics can in principle also be applied in production."