From Automation to Gen AI: what's changed and where does it leave us?

Back in the before-times around 2020 I did a piece of consulting with a global FMCG to look at where they could be more effective or efficient by automating parts of their marketing processes.

What became a useful filter was this 2x2 which we used to evaluate parts of the process based on how frequently they was done and how difficult (or how much craft) they required. From that list we then prioritised based on the ratio of effort vs return of automating it; was the juice worth the squeeze?

It enables us to categorise work into four quadrants

1. Automation at scale: This was the area of greatest opportunity, processes that were highly repetitive and required little skill or craft that were easily automated. Typically these were activities we could use existing technologies to execute with increased speed and accuracy like validating file formats, resizing of adverts, writing alt tag copy, creating eCommerce images & descriptions. This was the heartland of platforms like Celtra, Vidsy and Smartly.io who were well established in this quadrant.
2. Automation: The next most promising area was automating activities that were low craft but low frequency and that had high value such as briefing. Please note planners and strategists this doesn't mean the process of writing a creative or strategic brief! It refers specifically to the process of a client requesting the production of assets. Automating production briefing? ensured consistency of input resulting in greater speed, accuracy and interoperability: tasks could be more easily broken down and shared across time and geography. i.e. they could be more easily produced by automation.
3. Intelligent Automation: Work in this quadrant tended not just to be more efficient to produce, but because the automation efficiency enabled a higher degree of variation, also became more effective. Intelligent automation enabled mutivariant testing of modular assets like emails, banner ads and web landing pages along with the elements that fed into those assets such as copy generation, translations, image variations and creating short animated video segments from still images. There were emerging players in this quadrant like Ad.Lib and Spirable (both of whom I then worked with) and Pencil AI.
4. Intelligent Augmentation: The area of least opportunity pre-Gen AI was activities that required the highest level of craft and were done the least frequently; originating creative ideas, strategic insights or developing original copy lines or images.?

I went back to the same global FMCG earlier this year to undertake a similar exercise for one of their product categories. The change I observed is described by the bellying out of the curve as automative capability went from a concave to a convex zone of addressable use cases.

Not only had platforms such as Celtra added significant generative capabilities to the production automation function, platforms such as Pencil.ai had incorporated the generative capabilities of LLMs to move both up and down the production chain enabling brief creation, original copy generation and visual creative concepts. In parallel and in part enabling these evolutions, generative platforms like ChatGPT, Claude and MidJourney present unprecedented opportunities for both incumbents who leverage them for first mover advantages but also disruptors who are unencumbered by legacy technology, processes and business models.

The result is two key shifts that change the fundamentals of marketing processes

1. Unlimited variants; as the barriers to producing highly individualised asset move quickly to zero we are now fast approaching a situation where every consumer touch point can be a personalised experience. This has significant implications for how we plan and execute these touchpoints, what it means for a brand guidelines as we approach an era of liquid or fluid brands that respond in real-time to personal preferences.
2. The role of human intelligence or craft; where does human creativity and strategy play a role? If software can do all of the heavy lifting, what role does human taste and discernment play in guiding the software?

We are at the very beginning of these two trends and their ensuing trajectories and interrelationships will result in emerging threats and opportunities across the marketing landscape.