Contact Recommendations can be used to retrieve a ranked list of people at a target company who are most relevant for a given sales motion (use case), such as prospecting, deal acceleration, or renewal and growth. The recommendations are derived from past user interactions and account activity, and are ranked by a machine learning model. These recommendations can be used to build targeted call lists, prioritize outreach, or guide sellers toward the right stakeholders within an account.
To find Contact Recommendations select the appropriate useCaseType (for example, PROSPECTING or DEAL_ACCELERATION) and the
target company (filter[ziCompanyId]). These inputs tell the model which sales motion you are running and which
account you want to focus on, so that it can surface the most relevant personas based on historical past user interactions or engagements.
Behind the scenes, the model leverages data such as the user's past contact views, exports, and copies for the PROSPECTING motion, or contacts from a user's CRM related to closed won deals for the DEAL_ACCELERATION motion. The model uses this data to infer which types of people are most likely to drive success for the selected motion. It then finds similar contacts at the target company and scores them using a combination of similarity and propensity signals.
The endpoint returns up to 100 recommended contacts, ordered from most to least relevant. To control the number of recommendations returned,
use the page[size] parameter with any integer value between 1 and 100. Each recommendation contains additional metadata (meta)
that describes the reference person used to form the recommendation. Each recommendation includes the general similarity score (score),
a re-ranking score (reRankingScore) which uses several propensity signals (such as contact similarity, contact quality, title boosting, etc.)
to refine relevancy, and explainability metadata (meta) that describes why this person was recommended (for example, the reference person
and source of the interaction).