Ethical Personalisation: 4 Use Cases for Privacy-Enhancing Technologies

Isn’t it annoying when ads for products you’ve previously searched for follow you around the Internet? The worst part is you can’t do anything but ignore them (or, in the best of cases, opt-out).

And don’t you hate it when you grocery shop online and get product recommendations for food you can’t eat? If my diet is gluten-free (and retailers can easily infer that from previous purchases), why do I continuously get ‘personalised’ recommendations that don’t match my nutrition needs?

The algorithms that most retailers use for search and product recommendations don’t take the customer’s voice into account. They simply infer aggregated relationships from our shopping behaviour signals (ie. clicks, add to carts, purchases, search queries, lists, etc.) and ‘predict’ our next potential purchase without any input from us, their customers. The machine learning models behind these inferred relationships are rarely elevated nor explained, leaving customers in the dark.

This gap between a customer’s needs/expectations and a personalised shopping experience that randomly uses an individual’s past events creates a feeling of frustration and a major disconnect. Naturally, this gradually erodes brand perception and substantially damages the trustworthy relationship between customers and brands.

The truth is, we all love that feeling of surprise when good recommendations and relevant search results are displayed, but only when done correctly. We want to feel understood. We want to understand why products are being shown to us. We want to feel in control and not spied on or objectified. Now, how can we design shopping experiences that feel personal and relevant, yet expected and enjoyable?

From Surveillance to Trust, inspired by WIRED

Digital Trust & Ethical Personalisation

There must be no place for dark patterns in data acquisition — no more data shadowing and messy, misleading consent banners. It’s time for brands to build relationships around trust. Time to build an online experience together and nurture a culture of sharing between brands and shoppers. Time for retailers to be intentional and demonstrate they care for their customers by treating them with fairness and respect.

This is not only the ethical move to make; it’s the most enduring (and profitable) option for brands to stand out from the competition and build a deeper, lasting relationship with their customers.

Now, how can brands move from good intentions to actions? Here are guidelines to help brands design ecommerce personalisation experiences that evoke digital trust.

Data Privacy: Transparency & Control

• Data collection & consent: What do you know about your customers? What is being tracked? Ensure data collection is understood and explained in plain language within the experience, rather than in long indecipherable consent banners that are confusing for shoppers. And considering consent, keep in mind that even if data is self-declared (vs. inferred), it doesn’t mean it’s zero-party data. In most cases, data ‘willfully’ collected from customers are classified as first-party and require consent.
• Data processing: How is your data processed? What type of inferences are they taken?
• Data storage: Where is your data stored? How long will it be stored? Will it be shared with other parties?

Ethical AI

• Purpose-driven: How can your customers’ personal data be turned into value for each of them? What is the outcome of your AI models? Check it’s transparent and explainable. Analyse for any bias (i.e. gender, race, etc.) in the training models and take action to make them more inclusive.

Privacy by Design: Joyful & Customisable UX

• Design experiences that elevate all of the above in a way that customers understand it, and have agency not only over the data, but also over the experience (i.e. presentation layer of data). Have responsible design principles been applied?

4 Use Cases for eCommerce Search

One-to-one personalisation that respects one’s privacy is possible. No PII tracking is needed.

Here is a selection of relevant use cases that apply customers’ intent, affinities, and past purchase signals to personalise relevancy and experiences.

1. Contextualisation (in-session)

The lifetime of shopper intent is in a particular moment and varies within a single session. Accordingly, intent-based personalisation is computed in real-time. Momentary choices such as style, brand, colour, etc., are inferred from events during a session and serve as a signal to re-rank products in subsequent queries.

• Data: Events such as search queries, products viewed and added to cart, are read and computed in sequential order within a single session. No PII data (ie. userid, userdevice, IP…) is ever read, processed nor stored.
• Visit type: All visitors. No need to be logged in.
• Use case: Boost and visually identify products that match the visitor’s in-session intent affinities in search results.

In-session Contextualisation

2. Search History (locally stored)

Search history captures intent declared in past visits and serves as inspiration in future visits.

• Data: Search history data is stored locally in the browser. Not feasible for cross-device.
• Visit type: All visitors. No need to be logged in.
• Use case: Show trending products to returning visitors based on their past search activity.

Products recommendations based on search history

3. Explicit Affinities

Personalise your customer’s search & discovery experiences based on their explicitly declared preferences, rather than inferred ones. Secure active and unambiguous customer consent.

• Data: Customer preferences are continuous and require storage and management controls. This persistence of affinity data (associated with customer ID) can be delegated to the Customer Data Platform (CDP) of your choice, or, ideally, stored in decentralised personal data spaces (PODS). See BBC Together + Data POD example.
• Visit type: Logged-in customers.
• Experience: There are many ways to make this a joyful experience; it doesn’t need to be a boring form with many boxes for shoppers to tick. Design interactive and conversational experiences that are highly visual by using product swipes, quizzes, chatbots, etc. Use AI to suggest affinity recommendations to allow your customers to curate their preferences in a fun, transparent and effortless way.
• Use cases:
  - Affinity Search: Boost and visually identify products that match the customer’s affinities in search results. For example, Food Profile: Allow customers to personalise the shopping experience (search, navigation, recommendations, etc.) based on their nutritional preferences, such as allergies and diet.
  - Fav Filters: Let customers save frequently used filters for easy and quick access.
  - Saved Searches / Subscriptions: Let customers save searches and be alerted of new products after launch.
  - Customer Preference Analytics: Inspire customers with preference-related analytics and insights.

Search personalisation based on explicit affinities

4. Purchase History

Past purchases can be used in multiple ways. Mapping orders to relevant customers’ value dimensions, such as spend per category, savings, and impact on sustainability, is critical for engagement.

• Data: Purchase History data can be centrally stored and made available via distributed search.
• Visit type: Logged-in customers.
• Use cases:
  - Search in My Orders: Allow customers to search in their past orders and easily repeat a purchase.
  - Buy It Again: Boost products that customers have previously purchased to the top of search results and category pages (and visually identify them).
  - My Shopping Story & Analytics: Inspire customers with past purchase-related analytics and insights.

Purchase History Search

Wrap up

Shoppers deserve a new type of relationship with their favourite brands. One anchored around trust, transparency and control. In-store surveillance has to evolve — none of us enjoys being watched while we shop without a clear and understandable value exchange agreed upon beforehand.

Brands that apply digital ethics and privacy-first design principles to the way personalised experiences are created will be embraced and loved by their customers. The ability of brands to elevate the customer benefits and value proposition in use cases such as affinity and recommendation controls, and purchase history in search & analytics, will help them build more authentic and enduring connections.

Trust and purpose will definitely win. A new age of commerce has already started.

References

• Why Digital Trust Truly Matters Digital Trust: McKinsey
• Zero Party vs Declared Data: Sergio Maldonado (PrivacyCloud)
• My data, My privacy: Ethical Data Analytics: Adriaan Van Roosum (Simple Analytics)
• BBC Play Together + Data POD: BBC.co.uk
• Exchanging personal data to see what we need and want — a short explanation: Corina Stirbu
• Session Contextualization: Jose Jurado (Engineering @ Empathy.co)
• Customer Preferences: Food Profile: Alexander Martinez (Product Design @ Empathy.co)
• Favourite Filters: Cláudia Bennasar (Product Design @ Empathy.co)
• Buy it Again: Marta Ruiz (Product Design @ Empathy.co)