Why Were Students Not Saving? Segment Aware Personalization at Dorothy Perkins
Overview
From a hidden discount to board level strategy, how segment signals that the data could already see became visible and useful experiences, governed by simplicity and a business case per deployment.
Personalization Specialist
Capita · Dorothy Perkins
Segment analytics, rule based logic, merchandising
Segmented 50% of UK sales
Making the data's segments visible in the experience
Dorothy Perkins' data kept describing audiences that its website refused to acknowledge. Students, one of the brand's most important demographics who were entitled to a 10% discount, had to hunt to make the system know who they were. Regional customers browsed a product list ranked for a national average that described nobody. The challenge was not finding data; it was getting that data to show up in the web, become visible and useful experiences, and doing it with a discipline the board could trust.
At Capita and Qubit, we designed, tested, and scaled segment personalization delivered through Qubit's platform, under a program philosophy of simplicity and business cases behind every deployed experience.
01: The Tension
Three gaps between what the data knew and what the experience showed
1: The offer nobody saw
The student discount excited a 10% usage, meaningful for the demographic, but availability was hiding at the end of the journey, far after the conversion choice. The offer did little work in the moment of browsing, when it could actually influence. The ecommerce team was glad to see it and correct it.
2: A national site for regional tastes
Product performance showed strong geographic patterns: certain products rank indexed in certain cities. The product display ignored all of it, serving one national average for everyone. The average described the typical user; the regional intent was going unused.
3: The legacy mindset
Personalization was a buzzword, meaning it was organizational fat. Personalization stalls when it is treated as an abstract capability. The program's code governance kept personalization focused on concrete, size tested treatments, so that personalization stayed legible to customers and credible to the board.
The data already knew who was shopping. The experience just kept acting as if it was day one, and every user was a discount unclaimed and a relevant product unranked.
02: The Craft
Building segment aware experiences
Four layers, each converting a known segment signal into a visible experience without adding lag to page loads. Try the two live prototypes below.
Student Pricing Toggle
A Student Prices toggle in the header. When activated, the student price versions instantly render. This persists across both the product listing page and the product details page, framing the whole journey with the discount.
Geo Behavioral Merchandising
IP lookup checks the location, serving a carousel of the most added to bag products from the user's region. It replaces the generic, one size fits all product display, ranking products prioritized by local popularity.
3: Momentum & Orientation
Supporting new visitors. Banners explaining urgency on product pages, and welcome popups greeting first time visitors.
4: Experience Governance
Simplicity and a business case per experience. Every treatment was built individually, with no personalization for its own sake, keeping results readable enough to be discussed at board level.
03: Rigorous Execution
The student toggle is a deliberate design choice: user controlled personalization. Rather than inferring who might be a student, which is a fragile and assumptions heavy approach, we let users identify themselves with one tap, then rewarded that identification instantly. Price is a sensitive surface: transparency beats inference on sensitive surfaces.
The geo-merchandising relied on fresh behavioral data, not static assumptions: aggregate clicks over the trailing 24 hours of local shopping activity, so trending in Chicago was a statement of fact about yesterday, not a merchandiser's guess about the region.
Each experience had to pass the governance check: a control holdout, a clear discount mechanic, a measurable outcome, a discipline that kept things simple enough to scale and credible enough for the board.
04: Measuring the Lift
of UK sales segmented at peak, the personalization experiences contributing to this share.
student price feedback delivered in session without a reload, unifying checkout and retail experience.
behavioral geo-merchandising: metro-level trends replacing the national average ranking.
strategic standing: personalization discussed at board level, integrated directly into the brand's growth strategy.
The key has been to keep things simple and build a business case for each and every experience we want to deploy.
Key learnings
Let users direct personalization
Price is a sensitive surface: price control is the best segmentation. Rather than guessing, a simple self selection toggle is direct, respectful, and visually rewarding. Personalization under control beats that personalization done to the user. Make them feel in charge.
Behavioral beats demographic
The national best seller list describes nobody. Disaggregating by metro level rankings on the trailing 24 hours of local activity keeps the merchandising relevant, reflective of regional trends, and highly converting.
Governance is a feature
The control holdout should not be the bureaucracy; it should be the quality control. It forced every idea to state its mechanism and its metric before launch, which is why the program built long term credibility instead of accumulating clutter. Simplicity scales; complexity collapses.