As we look ahead to the much-hyped AI revolution, businesses should learn from the last two major technological revolutions: the internet and social/mobile media. History tends to repeat itself, and as a tech “veteran” old enough to have worked through both, I’ve seen the risks of companies going all-in on the hype before the tech is ready. I’ve also seen the dangers of being too cynical and slow-moving.
The winning path forward lies between those poles – in a strategy that seeks to build augmented intelligence. Rather than aiming for the Jetson-esque view of flying cars and other complete paradigm shifts, augmented intelligence takes a measured approach. In this approach, companies apply AI to specific areas where it can have impact and value now, incrementally enhancing our everyday decision-making and letting us work faster and smarter. This approach builds an agile foundation to move quickly as those applications evolve toward a flying-car future.
Five things it takes to become an augmented-intelligence marketing organization
1. Prioritize data quality: The accuracy and comprehensiveness of your company’s data will make or break your success with AI. Every AI use case starts with training data and prompts into AI models. If that data is not accurate or is incomplete, it can lead to mischievous decisions – to the tune of millions, if not hundreds of millions, of dollars.
Building processes to ensure data hygiene, data without bias and comprehensiveness of your data will set the foundation for successful applications of augmented intelligence.
2. Commit to intelligence-led decisions: Everyone understands that AI is supposed to pull transformative insights out of Big Data. But what a lot of people overlook is that you need to start with a set of distinct questions you want to answer or decisions you need to make. As anyone who’s played around with ChatGPT has learned, the more specific the “prompt,” the more useful the output.
For marketers, an easy example is deciding which audience(s) you should be talking to in order to drive the highest likelihood of positive business outcomes. Once you have your targets, use AI to consider which channels will be most effective in reaching them, and how different creative (copy, visuals, etc.) will or will not resonate with different targets.
Perhaps more importantly, organizations need to be committed to trusting in this AI-derived intelligence, and being prepared for answers that might challenge preconceived notions or uncover bias. I’ve already seen augmented intelligence reveal some powerful and previously unseen insights for marketers. For example, Data Axle helped one of the world’s largest toy retailers use AI to see that its marketing spend was aimed at the wrong audience – that, in fact, grandparents are its prime target audience. Data Axle also worked with one of the nation’s largest coffee retailers, using AI to show that its high-volume buyers are more likely to be in the construction industry, in addition to the tech employees and remote-work entrepreneurs it previously focused on.
3. Use AI to enable experiences: Augmented intelligence also includes generative AI to actually deliver those experiences. For example, marketers are already testing generative AI for creative development – both to develop copy and visuals for hyper-targeted, hyper-personalized marketing messages.
Using AI alleviates one of the main problems with our current focus on personalized marketing: The closer we get to 1:1 messages, the more impossible it becomes to generate those messages at scale in time to be relevant. Generative AI can resolve this scale problem. But, again, augmented intelligence companies will spend the time now to build skill in using AI tools – in skillfully prompting AI on the front end and carefully curating the outputs on the back end – to make the most of the human-AI synergy.
4. Move toward predictive attribution/measurement: Being confident about attribution has become increasingly difficult with the fragmentation of channels. Instead of the 10 channels marketers were working with 30 years ago, they now have hundreds, and many have their own IDs. The volume and complexity of data is too complex for humans, but it’s the perfect application of augmented intelligence. Companies are already using AI/ML to analyze multi-channel fractional attribution and optimize campaign performance in real time, based on response rates, conversion rates, engagement metrics, etc.
But true augmented intelligence companies are starting to use AI-powered predictive ROI insights to put their campaigns on the right track from the very start. With this predictive planning, before companies spend a single dollar, they’ll be able to use historical data and performance benchmarks (by audience, channel, message, etc.) to model predictive ROI. They’ll have a confident sense of how a campaign will perform and use that to better plan and allocate budget.
5. Building an integrated ecosystem of AI: Typical Data Axle enterprise clients have 15+ ecosystem partners (mar tech vendors/tools) that play a part in their interactions with customers. The bloat of modern mar tech stacks is already a problem, but it’s about to get bigger.
Traditionally, one of the main challenges has been breaking down data silos between these ecosystem partners through data exchanges. But as more of these separate systems incorporate their own AI, the challenge is getting these AI systems to talk to each other.
We saw the need for AI-to-AI exchange early on, and we’re helping our customers enable this communication. Instead of sending records and attributes, we’re sending the AI models, like predictive audiences, with weighted and scored attributes to go along with the customer records. In essence, we’re allowing one AI to teach the other AI what it’s learned, what matters most, etc. So, rather than starting from scratch, we’re creating an integrated AI ecosystem that builds on itself.
This brings up a very tricky question of IP ownership: When an AI develops insights or models based on your data, who owns those models? To be an augmented intelligence company, organizations need to own that AI-generated IP – and that means they need to start now in putting the people, data governance processes and standards in place to manage this growing gray area.
History shows us we’re at the inflection point for AI transformation
Back in 1994, I was working to transform companies’ businesses from analog to digital at the dawn of the internet. I remember recognizing a critical inflection point where multiple factors came together to finally make this digital transformation imperative for every company. The first factor was the advancing capability and maturity of the technology itself. The growing business impact of these capabilities became increasingly hard to ignore.
But the real catalyst was consumer adoption. Once people started using the internet in their everyday lives, they expected businesses to do the same – and “e-business” shifted from a risky way to stand out from the competition to a necessity.
I could see the sharp divide between companies that had already started that transformation, or at least experimenting with certain applications of the internet, and those caught flat-footed when the shift happened.
We saw a similar tipping point with social and mobile media in the mid-2000s: All of a sudden, everyone had a smartphone in their hand, everyone was on Facebook and Twitter, reading blogs and downloading apps. The conversation shifted from using social and mobile media as a new way to reach customers to an essential way that consumers expected to interact with businesses.
And again, we can trace an unforgiving line between those companies that had started to establish a presence and a following in the social and mobile media worlds, and those that were suddenly trying to find a way into that vital conversation.
We’re rapidly approaching this tipping point with AI. The big leaps in generative AI completely changed the equation in the last eight months. We’re all hearing plenty about the incredible business opportunities around improving productivity and enabling personalization. But it’s consumer adoption that I’m watching.
AI is quickly weaving itself throughout our everyday lives – helping us text, helping us drive and helping the hundreds of millions of people signing up for ChatGPT with both fun and serious tasks. These are signals that we’re once again at the critical moment for imperative business transformation: Those organizations that can put the foundations in place to become an augmented intelligence company will thrive in the next decade, while those that can’t … won’t.
