AI Talk: Closing the AI-ready data gap in modern marketing  

Please register or log-in to watch this on-demand episode

 

Views on news 

Marketing organizations are racing to adopt AI while simultaneously trying to contain it with about 76.6% of marketers now having AI policies in place. Yet, more than half of marketers report feeling overwhelmed by the pace of AI change. Organizations are governing AI adoption without first planning for it, building guardrails around a road that has not been mapped.

 

Beyond the AI dilemma, marketing organisations also struggle with a decision architecture problem, regarding what data and how should be leveraged – a recipe for AI deployment failure. AI governance without a clear marketing strategy turns platforms into control mechanisms instead of growth engines.  

 

Fragmented systems versus composable architectures 

Fragmented data can impact four different areas negatively – compliance through fragmented consent data; customer experience due to inconsistencies in multi-channel communication; operational risk as a result of decision latency; and delays in AI adoption and automation. Fragmentation can also break down trust. Another aspect of the problem is technology – marketing stacks have grown organically without any end-to-end design.

 

These tools often operate in siloes too.  Also, the data that has been collected previously needs to be cleaned to make it suitable for gaining insights. As data standardisation and consolidation can be expensive, it’s important to get the internal buy-in through demonstrating the value that data projects can bring to the table. Explainability is key to data – if you can explain how you collected and analysed your data, you can trust it. Fortunately, one of AI’s most compelling use cases is cleaning and assessing data, as well as finding holes and patterns in it.  Therefore, marketers don’t need to wait till they can feed pristine data into their AI models but, rather, improve the model through iterations.  

 

Composability – the ability to change or replace parts of a system independently without having to re-platform workflows – makes the decoupling of data decision logic and execution and learning possible so each can evolve at their own pace. In terms of processes, composability offers flexibility in how new data sources, context and feedback are introduced in a timely manner. Composability enables AI models, data sources and decision-making policies evolve separately, as well as the adoption of new protocols and inference or feedback mechanisms. Another frequent issue is that monolithic platforms can’t keep up with the speed of change in AI technology.  

 

To ensure smooth collaboration between different enterprise software, there must be a unified data source that all participants can access. Companies have already started using AI-ready CDPs that support both structured and unstructured data processing and see improved conversion rates as a result. AI-ready CDPs not only prepare data for activation but they also provide context for decision making through matching historical with real-time data. For reliable results, observability and explainability must also be built into these platforms. Future-proof CDPs must also support agentic activation through dashboards and co-pilots. CDPs shouldn’t be designed to hoard data but connect and orchestrate intelligence across different platforms. It’s also key to success that a decision owner is assigned to every AI use case within the marketing function as well.   

 

The panel’s advice 

• AI governance is about giving teams the confidence to use AI.  
• Teams often spend more time on reconciling marketing data than acting on it.  
• Composable systems are like Lego, you can build with them whatever you want.  
• AI can break the iron triangle and create better things faster and cheaper.  
• Make the most of available technology while it comes at a subsidised cost thanks to the AI boom.  
• Find a high-value use case that is small enough to tackle and impactful enough to demonstrate AI’s value – and work backwards. 
• Think of which marketing workflow could create the greatest value with AI. 
• Narrow scope. Don’t try to modernise everything at once. 
• Read Dario Amodei’s essay on the Adolescence of technology.