cases Retail

How Predictive Analytics Helped a Furniture Retailer Save €529K

A Swiss furniture retailer spent €667,000 annually on direct mail campaigns. Using predictive analytics, this budget was reduced by 79%. Sales remained the same.

Service: Predictive analytics

Timeline: September 2026 to present

Country: Switzerland

Industry: Retail

Company: NDA

Summary: A retailer with 105,000 customers sent out catalogs with a low conversion rate of 1.2 (%). An AI model identified customers who were highly unlikely to make a purchase. These customers were excluded from the mailings. The result was massive cost savings with a doubled conversion rate.

Business area

The company is an established Swiss furniture retailer. It sent out approximately 445,000 letters annually. At a cost of €1.50 per letter, the budget reached €667,000. Furniture is rarely purchased, which explains the low conversion rate.

problem

The company sent letters to all customers, even though only a few actually made a purchase. Sending a letter to someone with no need is a waste of money. Simply shortening the list was too risky. The company needed a data-driven approach.

Solution

Step 1: Consolidate data into a single source

First, we gathered all available customer data in one place. This included purchase and communication history (which mailings each customer received and whether there was a response). We organized everything so that we could track all of each customer's activity over time.

Step 2: Create a dataset with behavioral metrics

For each customer, we calculated simple indicators that describe their purchasing behavior.

When did you last buy something?
How often do they shop?
How much do they spend on average?
Which product categories do they prefer?
Did they ever reply to the letter?

These indicators help the model understand the difference between active and inactive customers.

Step 3: A predictive model was trained

We trained the model to answer one question: Will this customer buy something in the next 3 months?

We opted for 3 months because people rarely buy furniture. Looking at a shorter period like one month, almost no one buys anything. There isn't enough data to learn from it. Three months provides the model with enough examples of actual purchases to identify patterns.

The original idea was to find 1,000 to 3,000 customers who would definitely buy and focus on them. But an honest data analysis revealed a different picture.

With a customer base of 105,000, the proportion of those who will buy in the short term is very small. Behavioral signals were limited and essentially restricted to purchase history. Therefore, it was statistically difficult to reliably determine who in the range of 1,000 to 3,000 people would buy.

Step 4: The strategy was changed from “Find buyers” to “Remove non-buyers.”

Initially, we tried to identify the 1,000 most likely buyers. However, the data showed that it is difficult to accurately predict who the buyers will be. will The purchase is difficult because behavioral signals are only available to a limited extent.

We have changed our strategy to excluding non-buyers. Although finding buyers is difficult, the model identified individuals with no purchase intent with an accuracy of 99%. By selectively removing these inactive segments, we were able to significantly reduce the size of the mailing list.

Step 5: Implementation of the segmented mailings

We removed inactive customers from the mailing list. Now, only customers who have shown potential interest receive a letter. The next campaign reduced the number of letters sent from 105,000 to just 20,000.

Results

79% – Postage costs decreased from €667,000 to €138,000 per year.
79% – The company now sends 92,000 letters per year instead of 445,000.
€529,000 – Annual savings after the use of Predictive AI.
x5 – The conversion rate has increased fivefold.

Book personalized demo

See what your data can do
for your business

Find out how to get started and what results to expect