We are living in the age of the algorithm. In a previous article I briefly touched on three types of algorithms to help with three types of strategy: brand, relationship and business.
In this article I will deal with the first of these: the use of algorithmic intelligence to help create and harvest demand for brands.
As algorithms stretch further, humans are needed more, not less. Brand algorithms, at their most holistic, can be the ultimate cultural mavens — informed by the whole internet, by what people are reading, watching, sharing, reviewing and searching. But as we will see, whilst they know everything about everything, they can also be the ultimate pub bores without human intervention.
Platform algorithms
Let’s start with the role of platform-specific algorithms. These algorithms (such as Google Performance Max and Meta Advantage+) are becoming indispensable tools in media planning and buying. But they are limited in three ways.
First, if their big promise is to constantly optimise, they are (still) most effective at optimising towards mid and bottom-of-funnel metrics, rather than upper-funnel ones. Google Performance Max, for instance, is single-mindedly focused on conversion.
Second, they optimise within their platform but not beyond it. A platform like Google, offering everything from Gmail to YouTube to Maps, is impressively diverse — but it still does not help us determine how much we should invest in Google platforms versus other options.
Third, they are limited by the inputs they are given. If the assets are poor, they will be optimising towards the best of a bad bunch. And if, in the years to come, generative AI increasingly rolls its sleeves up to create assets of its own, it will tend towards mimicry of the best-performing category assets, diminishing meaningful difference.
Algorithms as inputs
To best use algorithms requires a shift in perspective. As outlined above, one way is as an autonomous final link in the chain: pieces of magic to which you give initial direction and assets, and then leave them to optimise as best they can. This is absolutely correct. But to get the most out of algorithms we need to also think of the various algorithms as start points — sources of insight around human intent — that we can combine with other sources of insight and brand understanding, to create ownable and native experiences.
Signals of human intent have deepened and broadened. Therefore, we need to think holistically and beyond the intelligence of any single source of truth. Sometimes user intent is self-evident; however, often it is less so. For example, the use of social search to find user-generated content that impartially verifies recommendations already unearthed by Google. In emerging platforms we need to understand whether (for instance) ChatGPT is used more for creative exploration, versus Perplexity for more technical queries, and when accuracy is vital.
Algorithms help us understand how each platform is used and the content within it that people love — this insight will help us craft ever better inputs for the platform to optimise.
Algorithms beyond the platforms
As well as understanding platform algorithms, we can also create brand algorithms that are platform-agnostic.
These stand astride the internet, finding patterns in what people are watching, sharing, liking and creating in culture. They encompass both structured and unstructured data. The ultimate know-it-alls, they can help us find untapped category entry points in culture.
For example, at Carat we recently used clustering AI to understand various drinking moments for a portfolio of spirits brands. In the pre-going-out drinking moment, we uncovered nuances about how people were premiumising their experience, swaying to 90s cheese rather than hipster cool, and deciding to call it ‘prinks’.
This approach can also uncover whole new occasions to associate with your brand. For example, we used it for a coffee brand and found a rising cultural link between coffee drinking and active social activities — whether that is a coffee after a walk for an older audience, or coffee drinking around skate parks for younger audiences.
Algorithms help us uncover insights such as this, but only humans can determine whether they are ownable, and how they can be expressed in a way that is true to the brand and native to the platform. Again, it is the partnership between algorithmic intelligence and human intelligence that can create better inputs for platform algorithms and beyond.
Communing with the algorithms
Algorithms are increasingly indispensable tools to help us build brands and harvest demand. Platform-specific algorithms, whilst essential, are not the answer by themselves. They need to be supplemented by a more holistic algorithmic approach.
However, all of this only highlights the ongoing need for human intelligence, albeit in an evolving form.
Moving forward, there will be a need to bring together media and creative expertise. Media expertise, which can commune with the algorithms to understand how channels are used, the brand opportunities, and what brilliant native content looks like. And creative expertise, from agency creatives to platform natives, who can use this to craft meaningfully different brand experiences.
Featured image: Pig (2021), Neon Films