What needs to be in place before machine learning for marketing begins?

We begin by reviewing historical datasets, prediction objectives and outcome definitions. We check what you already have before recommending new assets, systems or supporting work. Useful modelling requires sufficient relevant data and a decision that can act on predictions; simpler baselines must be considered. Missing access or ownership is identified before delivery commitments are finalised.