Artificial intelligence has several applications within marketing analytics. Everything from data cleaning to model configuration to fit optimization benefits from strategically implement programs; however, it is imperative that these AI solutions are transparent and narrowly tasked. AI solutions come in many forms, but all can be defined by their area of responsibility: broad or narrow.
Broad AI is not focused on a specific task and is designed for a general purpose. While this may be the next step for some solutions, like search, it is difficult to implement in measurement because it often lives within a black box. Broad AI is generally self-taught. After the programmers set up initial parameters, the system learns from new data sets and forms its own connections between similar observations. There may be a future where broad AI is useful in marketing analytics, but currently cannot be trusted to perform repeatable tasks.
Narrow AI is machine learning focused on a single task. While it can’t handle every task, it is very good at doing one thing. This makes it perfect for marketing analytics when the process is well-defined. Well-defined processes allow AI to focus on specific tasks to optimize. Leavened’s own ConfigAI feature uses artificial intelligence to bring speed to insight and ease of use to their marketing mix model (MMM) measurement. ConfigAI is focused specifically on the task of optimizing model configurations. This helps eliminate bias and expedite analysis by recommending optimal decay and saturation transformations.
Regardless of how a team leverages AI for analysis, it is important that any solution is transparent. AI and machine learning can become vague buzz terms when not strategically implemented for a specific purpose. The clearer the purpose for an AI solution increases the chance of success.