Granular volume planning and micro-targeting
+ 1–2% revenue, -1–3% tactical spend
Granular volume planning and micro-targeting
Key use cases
An automotive OEM achieved a ~60% lower forecasting deviation via machine-learning-based daily sales forecasts
Case example 2 / 9
Many businesses struggle to identify performance deviations in terms of volumes and margins systematically. The solution needs to cover channels, regions, models, and customers to create accurate forecasts, leading to improvements in product pipeline attractiveness.
By programing a machine-learning algorithm with market, competitive, and sales data and applying a random forest model to these variables, the OEM created a mini data lake. It prepared the insights from this data lake for senior executives in visual form and made them continuously accessible in a physical war room for management gatherings. This made fast and informed decision-making possible by enabling joint 360-degree analysis on current sales issues. This solution led to better sales forecasts with 60 percent greater analytic accuracy than conventional forecasting methods.
~60%
Car dealers can increase sales using a predictive model based on sales drivers to identify postal code areas with the highest sales potential
Case example 1 / 9
Identifying sales potential at the postal code level and deriving insights from these findings can be a challenging task. Based on an advanced analytics approach, a large OEM developed a solution that enables dealerships to identify additional sales potential at the postal code level.
The predictive model analyzes large amounts of internal and external data that are then run through an exploratory analysis via SparkBeyond, which automates substantial parts of the data science process. The result is the discovery of a set of 50 top features that affect vehicle sales and the creation of a deployable predictive model to identify additional vehicle sales at a granular level.
lower forecast deviation
+1–2%
revenue
Sales war room for analytics-based decision making
Sales war room for analytics-based decision making
60% lower forecast deviation
Lead management and sourcing
Lead management and sourcing
+18% sales conversions
Next product to buy recommendations
Next product to buy recommendations
+8% points order intake
Incentive spend optimization
Incentive spend optimization
5–10% savings on incentives
Consistent cross-market pricing execution
Consistent cross-market pricing execution
+2% price
B2B dynamic pricing for online spare parts
B2B dynamic pricing for online spare parts
+2% revenue
Post purchase cross- and upselling
Post purchase cross- and upselling
+30% post-purchase sales conversion
Optimized stock vehicle configuration
Optimized stock vehicle configuration
+2–3% return on sales (automotive specific)
A premium automotive OEM netted significantly more sales conversions by implementing an advanced analytics scoring algorithm
Case example 3 / 9
Lead generation, management, and follow-up are some of the most important competences of a strong-performing sales force. However, many companies struggle to generate high-quality leads and find it hard to manage and track these leads successfully, leading to limited transparency and low conversion rates.
A premium automotive OEM transformed its old rule-based scoring system into an advanced analytics-driven approach for better end-to-end lead management. It created a lead database using 360-degree customer data and developed a state-of-the-art machine learning model for scoring and to boost sales conversions. The improved lead-scoring model enabled the sales department to identify high quality leads that produced a conversion uplift of up to 18 percent. The system also filtered out more than 32 percent low-quality leads, thus significantly increasing lead follow-up efficiency.
18%
more sales conversions
Up to
32%
fewer low-quality leads
Up to
A multinational advanced industries player increased order intake by 8 percentage points year-on-year in its distribution channel
Case example 4 / 9
The manufacturer had two goals: increase orders and build long-term selling capabilities. It developed an algorithm to provide “next-product to buy” recommendations. By analyzing all the customers across regions and product mix, the model could make meaningful recommendations that led to an 8 percentage points increase in order intake year-on-year.
For example, if a distributor ordered a large number of diverse switches and fuses, the analytics engine would identify panel boards that other, similar customers would typically also order. The company then trained sales representatives to explain the recommendations coming from the engine to customers.
+8% points
order intake
A global OEM captured ~5–10% savings on its incentive spend through an analytics-enabled incentive optimization approach
Case example 5 / 9
A large OEM developed a real-time approach and tool to simulate and visualize the impact of each type of marketing and sales incentive on sales volume and profitability. By combining internal and external data sources on VIN levels and by running several standard analytics, the OEM managed to derive clear country-level measures to optimize incentive spend which is typically 10 to 20 percent of revenues. Its next horizon involves further enriching the approach with predictive advanced analytics models, allowing fully data-led decision-making on incentive spend levels per geography and model under varying side conditions like volume, profits, or spend.
~5–10%
baseline savings achievable on incentive spending
A large global premium OEM achieved a ~2% price increase through a data-driven pricing execution program
Case example 6 / 9
Due to decentralized pricing structures, a global premium OEM operating in over 15 markets restructured its pricing approach to embrace the concept of parameters to set prices and calculate the maximum discount for a given deal. Based on historical and deal-specific parameters, the developed model calculates both a recommendation for the initial offer price and the maximum discount dealers can give. Representatives in the three involved markets were closely involved in the process of designing and refining the model, making sure it reflected the practical requirements of dealers and agents. After thorough tests, the model was rolled out in a single pilot market. The results were impressive. With the help of the tool, dealers achieved an effective price increase of 2 percent, even though the brand was already the most premium-priced one in the market. Subsequently, the new approach was rolled out to another 12 markets in several implementation waves.
+2%
price increase
Global rollout
of the developed pricing approach
A large OEM captured about EUR 700 of additional revenue per car by identifying order-specific and customer-tailored option upselling potentials
Case example 8 / 9
After a successful sale, sales representatives often think about additional, after-closure offers to capture incremental contribution margins for their business. To automate this process, a large OEM implemented an algorithm that suggests options with high purchase probability, and develops tool-supported, ready-to-send emails containing option proposals for sales representatives.
By analyzing historical data on consumer buying patterns, configuration options, and customer demographics, the OEM applied analytics-based behavioral psychology and big data approaches to identify order-specific, customer-tailored option upselling proposals. This led to a conversion rate that exceeds 30 percent and additional revenues of about EUR 700 per success case.
>50%
post-purchase conversion rate
~700 EUR
additional revenue per conversion
A global OEM optimized build-to-stock vehicle orders by using machine learning algorithms for optimal dealer and country sales recommendations
Case example 7 / 9
A global OEM wanted to know how dealers and country sales organizations could configure their stock vehicles with the right equipment levels to increase profitability while simultaneously decreasing time-on-stock. The result was a front-end for dealers that provides on-demand lists of recommended configurations for country sales organizations and fills up their stock with vehicles expected to have high profitability and short time-on-stock.
By analyzing data on build-to-order / built-to-stock vehicles, customer configurations, and vehicle list prices, the OEM developed two recommendation engines. One gives feedback to dealers about their configurations based on expected time-on-stock and contribution margin. The other proactively provides lists with proposed preferable configurations for dealers and the respective country organization to drive profits while decreasing time-on-stock.
2–3%
RoS improvement
Decreased
time-on-stock
An industrial B2B online spare parts distributor introduced a self-learning dynamic price engine that delivered run-rate estimates of 2% of revenues
Case example 9 / 9
To capture the full value in pricing and to understand the win-rate behavior and the willingness to pay of customers, a large industrial B2B spare parts distributor developed a self-learning pricing model, dynamically incorporating existing and new information while maintaining sufficient “guardrails.”
The company analyzed historical data on customer willingness to pay, customer demographics and competitor prices, and ran A/B tests to measure win-rates for similar customers at different price points. It eventually segmented all customers with similar willingness to pay. On top of their existing sophisticated engine, they combined the self-learning pricing model with personalized cross-selling recommendations across the customer journey, to maximize the economic benefits for the B2B company.
+2%
impact on run-rate revenue estimates
–1–3%
tactical spend