How AI is (almost) Transforming Marketing
Datorama recently published a white paper called "How AI is Transforming Marketing -- Making the Impossible Possible". Because AI is the shiny new thing, I was naturally curious.
As it turns out, the paper was pretty good. Here is a summary:
AI helps in three areas:
Collecting data from disparate data sources
Creating and maintaining the data model that unifies the data
Generating insights by trolling through the complete data set
In the first case--collecting data--AI can pull in feeds and files. Datorama provides a deep bench of connectors that make this process easy and constantly improving. This is handy.
AI also has the ability to create an maintain the data model. This is critical because the data model is how you get the "single source of truth". It is the framework the unifies the data and provides access. Using AI to create dynamic data models would truely be useful because data models have a tendancy to be rigid. If incoming data is not well formated, errors can happen and whole sections of data may be excluded. Presumably, AI provides a more elegant solution for this.
Insights are the final area that AI helps out. Being able to scan bazillions of data points to find relevant trends and outliers is useful. Unfortunately, there are three main problems here. First, I haven't seen an AI that is really good at this yet. AI tends to lack judgement and likes to return all kinds of stuff that is irrelevant. Second, AI analysis is normally based on only one or two dimensions. Until AI can combine 4 or 5 or more dimensions, it won't really be doing anything special. Third, the main selling point of AI insights is the speed at which they can be created. However, most marketing organization function best at a slower cadence. Daily insights are not really actionable for most marketing organizations.
Overall, the article was well done and thought provoking. There are some real benefits to AI that Datorama seems to be tapping into. I'm sure the weaknesses will eventually get ironed out. But we're not there yet.
You can download the whitepaper with my annotations on it here.