A prediction model in a database system may be configured to predict, for a given object instance, a respective probability of acceptance for each of a plurality of recommendations. A determination may be made as to whether the prediction model is associated with sufficient training data to produce predictions at a designated accuracy rate. When it is determined that the prediction model is not associated with sufficient training data, for each of a first set of object instances a respective first message may be sent that includes a respective first one of the recommendations determined based on a static ranking rule applying one or more criteria to one or more object fields associated with the respective object instance. The prediction model may be updated to include additional training data based on a plurality of responses corresponding to a respective first one of the recommendations.