Behaviour vs Similarity Signals
Two signal types, one smarter storefront
Overview
Liva 7 uses two primary types of signals to power its personalisation engine: behaviour-based and similarity-based. Each plays a distinct role in how recommendations are generated, and understanding the difference helps you choose the right strategy for your goals.
Behaviour-Based Signals
These signals are generated by what a shopper does on your site in real time.
Examples include:
• Viewing a product
• Adding to cart
• Completing a purchase
• Returning to a product multiple times
• Clicking on a recommendation
What they’re good for:
• Personalising recommendations for each individual shopper
• Powering strategies like Behaviour-Based, Hybrid, and Cart Recommendations
• Adapting in real time to changing shopper intent
Behaviour signals are ideal for tailoring recommendations to each shopper’s journey - especially when intent is clear.
Similarity Signals
These signals are based on how products relate to each other, regardless of shopper behaviour
.Examples include:
• Shared tags, categories, or collections
• Similar titles, descriptions, or attributes
• Visual or semantic similarity
• Manual curation (e.g. bundles or 'Complete the Look' sets)
What they’re good for
:• Recommending alternatives or complements
• Powering strategies like Similar Products, Content-Based, and Trending
• Supporting discovery, especially for new or anonymous visitors
Similarity signals are great for helping shoppers explore your catalog - even if they haven’t interacted much yet.
How they work together
Most Liva 7 strategies use a blend of both signal types. For example:
• Hybrid Strategy combines behaviour and similarity to balance personalisation and discovery
• Trending Strategy uses similarity signals plus storewide popularity data
• Fallbacks often rely on similarity when behaviour data is missing
Hybrid is the default strategy in Liva 7 because it adapts to both new and returning shoppers.
Choosing the right signal type
Use behaviour-based signals when:
• You want to personalise for logged-in or returning shoppers
• You’re optimising for conversion or AOV
• You have strong signal volume (traffic, events, purchases)
Use similarity signals when:
• You’re targeting new or anonymous visitors
• You want to promote visually or categorically related items
• You’re launching a new product with no behavioural history
Bringing it all together
Understanding the difference between behaviour and similarity signals gives you a powerful lens into how Liva 7 thinks. By knowing which signals drive which strategies, you can fine-tune your recommendations to meet shoppers where they are - whether they’re browsing for the first time or returning with intent.
