Forrester sets five criteria for what truly counts as an AI agent in marketing
According to Forrester, the concept is becoming blurred as vendors sell more and more tools as "agents"; the firm expects a marketing agent to carry out work on its own, plan to achieve a goal and be able to explain its decisions.
What's happening
In a post on its own blog, Forrester says it is time to draw a line between what is and isn't an AI agent for the marketing function. The firm defines AI agents as applications that carry out tasks, make decisions, and interact with data and other systems on their own on behalf of a business or individual. A marketing agent, in turn, must meet five criteria: it must be marketing-focused, autonomous, responsible and goal-oriented, and it must carry out multiple tasks in an adaptive workflow.
Whose desk it lands on
The executive in charge of marketing technology. Which of the tools we use today under the "agent" label actually carry out work on their own, and which just wait for commands?
Our view
The CX Brief's interpretation, not a claim made by the source.
Forrester's criteria are most useful at the buying stage: a tool that can't carry out work on its own, plan to achieve a goal or explain its decisions doesn't count as a marketing agent, whatever it is called. The firm's warning backs this up: inconsistent definitions amplify delivery risks and deepen concerns that AI will fall short of expectations. Reading vendor pitches against these criteria lets buyers see disappointment coming before the contract is signed.
