What is a scoring weight?
A scoring weight is how much influence each factor has within the overall message score. The factors that go into the score — evidence strength, recency, surface fit, journey stage, message clarity — are not treated equally; the weights set the balance between them.
What do the weights control?
They determine what the engine prioritises when choosing between competing messages. Evidence strength and surface fit, for instance, carry more weight than message clarity — so a strongly supported, well-placed message beats a marginally clearer but weaker one. Adjusting the weights changes the engine’s taste in which messages win.
How is a scoring weight different from the message score?
The message score is the result — a single number for one candidate. The weights are the recipe that produces it — how much each ingredient counts. The same underlying signal can produce different scores under different weights, which is what makes the weights the lever for tuning behaviour.
How are scoring weights set?
They are configured in the pipeline rather than fixed in code, so the balance between evidence, context and clarity can be tuned to a store’s priorities without changing the signals themselves.
See scoring in the platform
The engine weighs evidence, context and clarity to choose between competing messages.