Personalshop
Customer Scoring & Sales Forecasts
Data-based models support targeted customer communication and inventory planning – from catalog selection to the early detection of out-of-stock situations.
The Company
Tyrolean mail-order retailer with multi-channel distribution
Personalshop is a Tyrolean mail-order retailer that sells a range of brand-name, licensed, and private-label products at discounted rates to a closed customer base. Customer communication is conducted via an extensive mail catalog of over 100 pages, newspaper inserts, advertisements, emails, and newsletters, supplemented by eight brick-and-mortar stores. The company employs approximately 400 people and generates annual revenue of around 250 million euros.
The Challenge
Key elements of the assignment:
IMARK supports Personalshop in shifting two central management questions of the mail-order business from experiential knowledge to a data-based foundation: Who should receive the next catalog – and which items will be in demand, when, and in what quantity?
-
Development of a prospect scoring system that predicts the expected yield per customer – order total minus returns – before every catalog mailing, thereby guiding the selection of recipients
-
Establishment of a planning model to forecast demand per item and to detect imminent out-of-stock situations at an early stage
-
Transformation of both models into continuously usable decision-making tools for marketing, purchasing, and catalog planning
The cornerstones of our approach:
Both models are built on the transaction and campaign data already available within the company. IMARK combines these data sources into predictive models, the results of which flow directly into operational planning.
Analysis of customer-specific success of past mailings and linking it to the respective catalog content (promotions, cover product, placement)
Inclusion of previous ordering behavior per customer – purchase date, purchase frequency, and ordered product categories
Calculation of response probability and expected value per customer as a basis for selecting catalog recipients
Development of a sales forecasting model that takes into account catalog presence, placement, pricing, seasonality, and product attractiveness
Combination of the forecast with current inventory levels and expected returns to provide early-warning signals for product availability
Methods
Methods and techniques used (among others):
-
Scoring models based on individual response probabilities
-
Expected value calculation per customer
-
Customer analysis by recency, frequency, and product categories
-
Sales forecasting models with seasonality and placement effects
Result & Impact
Benefits for Personalshop:
The models provide Personalshop with a reliable, repeatable basis for decision-making for every catalog mailing and for ongoing assortment planning.
Servus Handels- und Verlags-GmbH
Headquarters
Polling, TyrolBranche
Retail & Consumer GoodsProject Period
2024 & 2025-ongoing
Testimonial
at Personalshop
Data Expertise Meets Collaborative Partnership
“The collaboration in the field of Predictive Data Modeling impresses us with its high analytical competence and deep understanding of complex interrelationships. Especially with large data volumes, this leads to well-founded insights and concrete solutions with real added value for our company. I particularly appreciate the collaborative, appreciative, and consistently solution-oriented cooperation on an equal footing.”