Finding Value in AI

Ever since the launch of ChatGPT, technology vendors have been adding as many Generative AI features as possible to their product portfolios. Meanwhile, marketers have been experimenting with the technology in search of possible use cases that can benefit the customer. The one question that few people have been asking until now, is whether the technology can deliver value for money to the business. That is beginning to change following a series of high-profile analyst reports criticising the slow adoption and poor return of AI projects.

Beyond the Hype

We're starting to see an increased disconnect between the media hype behind AI and the practical reality, which can be explained by two main problems inherent to AI in its current state. The first issue is that AI is a supporting technology rather than a fully featured end product. It is not a capability that directly improves internal efficiency or the customer experience. AI is only beneficial if consumers or businesses use it to speed up existing business processes or to unlock new customer insights. Even then, the AI model needs to be cheaper to run than the existing business process it is replacing.

The second issue with the AI bubble is that AI is not actually new. Businesses have been investing in different types of AI for years already. Traditionally, AI has been most commonly utilised for data analysis. Predictive lead scoring is one such example, a capability marketers have used for nearly twenty years. Data cleansing is another area in which we have been using AI for a very long time. More recently, we have seen new uses for AI in the data realm, particularly to support deeper journey mapping as well as ABM workflows.

The Cost of GenAI

Generative AI is notable because it extends the capabilities of AI into the domain of text and image generation. That does have substantial benefits within marketing, but only if content accuracy and brand integrity can be guaranteed. AI has proved useful for generating high volumes of low-value content. Chatbots have benefited the most from the technology, and have the clearest path to demonstrating ROI. However, many marketers are struggling to justify the cost of investment even here.

The challenge for tech firms is in monetisation, because most marketers are comfortable using some derivative of ChatGPT for any Generative AI requirement. As with most immature technologies, venture capitalists and big tech firms have been willing to absorb the costs of training AI models during the growth phases of the technology. The sheer amount of data and computing power required to operate an LLM means that AI is extremely expensive to operate, and that has been reflected in the costs of AI products that do try to turn a profit. With many companies struggling to bring AI into production, the ROI timeframe for AI technology is now under intense scrutiny. In recent weeks, we have begun to see business analysts questioning the financials behind the Generative AI boom.

Demonstrating ROI

Such concerns will only intensify as we enter the 2025 planning cycle. Corporate tech stacks are already bloated, and early research from Forrester indicates that marketing budgets are unlikely to rise significantly next year. They were quoting budget increases of around 5% next year, which is broadly in line with inflation. Technology vendors would like much of that extra money to be spent on AI, but budgets won't be able to afford such an investment.

We are starting to see Generative AI being used for more advanced use cases that may justify additional investment in time. In the interim, marketers are best focusing their AI efforts on data management, analytics and reporting. These are the areas where AI has long demonstrated strong ROI. Everyone wants to make data-driven decisions, but in practice, this is more difficult than it sounds. AI can help both in organising data and in spotting the trends that marketers are liable to miss. The only catch is that you need high-quality data to make the maximum use of the technology. Good data has long been a competitive advantage, but in the age of AI, having the best data is an even bigger advantage than ever.