Skip to content
Back To Knowledge Hub

By Alan Chatfield


AI Operations: Implementing AI for Marketing

Generative AI is fast becoming a multi-purpose tool. Already, there have been signs that the technology will disrupt society in new and unexpected ways. It promises a new era of unprecedented productivity, while exposing businesses to new and unexpected risks. 

As widespread AI becomes part of daily business operations, it will impact the day-to-day responsibilities of many marketers, leading to a substantial change in their roles. Job descriptions will be redefined, focusing more on internal collaboration and strategy development rather than day-to-day campaign delivery.

Roles and Responsibilities 

After all, AI cannot mentor junior team members, navigate the changing business environment or defend the role of marketing to the board. Ultimately, AI lacks both accountability and the decision-making power to do those things. Yet, decisions still need to be made and strategies developed. AI can then support in validating and executing those strategies and the associated campaign plans once they are sufficiently detailed.  

AI changes how marketers approach their responsibilities, but it doesn’t change the tasks that need to be delivered. Instead, it provides the opportunity to automate routine tasks. Each task assigned to AI will require the associated model to be configured and then monitored. Every department will need an AI expert responsible for training and managing the AI models used within the team. For marketing, that will be the marketing operations team.   

AI Operations   

Managing AI is about more than just prompt engineering, although this will be an important skill in the short and medium term. It’s also about maintaining the data flows and technology integrations necessary for the AI model to deliver the expected output. The proliferation of code free integration tools is already making this significantly easier for the modern enterprise. 

Mapping data flows and designing integration use cases requires a level of technical expertise that only marketing operations can provide. AI can offer advice and suggested field mappings, but it cannot validate all the necessary use cases and downstream impacts on other systems. The need to comply with data protection and information security legislation adds an extra set of considerations, requiring human input from a specialist.   

It will rarely be possible for the typical marketer to configure a new AI model just from a single prompt unless that model is a standalone AI that doesn’t use any data or applications available across the wider business. Compliance standards such as ISO 27001 will demand a design and governance process in order to prevent data leaks and ensure the model is non-discriminatory. We may also see copyright and brand compliance concerns leading to additional legal reviews for AI generated content in many situations.  

Return on Investment 

No business will adopt AI just for the sake of it. As with any new technology investment, business cases will need to be written and budget secured before any new AI models can be deployed. That won’t change as the technology matures, but gaining approval for new AI implementations will become easier over time. Once marketers understand the costs and benefits of AI, it will make scoping new use cases for the technology easier.   

The technology will need to provide a return on investment. Much of the benefit from AI comes in the form of efficiency gains, which is generally the most challenging type of improvement to measure. The work hours saved through AI usage will need to be quantified in financial terms, as this will offset the technology or server costs of using AI – which is currently very expensive for your typical Generative AI model. 

Productivity Gains 

AI has long been touted as the source for any future efficiency gains, but older machine learning technologies were often expensive and didn’t always deliver sufficient output to justify the high price tag. We’ve seen predictive solutions removed from tech stacks because they weren’t cost effective. It’s unclear whether this same problem will affect generative AI as well. 

We are still in the very early days of the generative AI product cycle, and the most exciting use cases for the technology are yet to make it into production. That shouldn’t stop businesses from exploring use cases for AI. The technology does pose risks, but it will deliver substantial productivity and efficiency benefits for everyone. Marketers just need to ensure they have the operational and technical expertise needed to make the most of this new and exciting technology.