Probabilistic MAB
Image selection
90%
Exploitation rate
10%
Exploration rate
Spectare treats images and text as separate, interchangeable modules. When a visitor arrives, the engine independently selects the best text atom for their intent and the best image for their audience and buying stage, then assembles them together. Because image selection is decoupled from text selection, Spectare can discover that Headline A converts better when paired with Image C rather than Image A, without a marketer setting up a single test.
Image selection uses a multi-armed bandit algorithm. 90% of the time it exploits the statistically proven best-performing image for the visitor's audience and stage. 10% of the time it explores a random alternative to gather conversion data on untested combinations. As your image library grows, the engine continuously learns which visuals resonate with each segment (developers, managers, executives) and at each stage of the buying journey.
When brand integrity requires a specific image to accompany specific content (a screenshot of the analytics dashboard next to the analytics feature description, for example), you can lock them together with a shared Group ID. Locked pairs always travel together. Everything else is optimised freely.
This is one atom from the Spectare content library. Spectare assembles the right atoms for each visitor in real time, based on who they are and how they arrived.