I recently discussed this topic and more with Rio Longacre, MD at Slalom, and Jon Williams, global head of agency business development at AWS.
Gen AI: Making sense of the mania
During our exchange, Longacre delineates a paradigm shift within the Gen AI landscape. Advancing past mere experimentation, enterprises are now crafting strategic pathways, pinpointing domains where it can authentically effect transformative change.
“In recent months, there has been a significant transition, which is exceedingly favorable. Rather than simply 'experimenting with various approaches', the focus is now on establishing a Gen AI strategy. They are actively seeking out sectors where Generative AI could yield substantial benefits and drive progress. They are eager to allocate resources to these areas, be it eCommerce, operations, or creative endeavors; they aim to conceive and trial concepts that hold promise. Should these initiatives succeed, it's fantastic. They will then look to monetize them. If not, that's perfectly acceptable as well; they can adjust and explore alternative avenues.”
The future of AI as a fellow marketer
Williams discusses the areas where Gen AI is excelling as a marketing aide, in a sense. "Amazon Q represents a novel form of generative AI-driven assistant designed for customizing work to suit your business needs, facilitating discussions, troubleshooting, or producing content. Leveraging the insights and knowledge embedded within your organization's data repositories, including codebases and corporate systems."
“You could use Q, for example, to:
* Learn a brand style guide, then
* Use that information to turn a press release into a blog post that adheres to those standards, then
* Analyse how a brand has shown up on social media, then
* Create new posts around those releases that will make sense to followers, then
* Analyse the results of those posts, and finally
* Summarise them for review for teams
"It's akin to this self-perpetuating loop of gradual productivity enhancement, driven by harnessing the generative AI functionalities inherent in a bot, seamlessly integrated into the systems and data proprietary to your enterprise. We're merely at the nascent phase of this evolution, which is tremendously promising."
3 Ways marketers are leveraging Gen AI for greater efficiencies and cost-savings, according to Longacre:
1. eCommerce company: This company has written descriptions for 10,000 product SKUs using Gen AI in a couple of weeks, saving them months of time and about a million dollars.
2. Paid media campaigns: As it relates to paid media tools, such as those designed for Amazon Marketing Cloud, there’s a background image generator specifically tailored for crafting lifestyle images. Findings indicate a remarkable 20 to 25% increase in conversion rates for products showcased with lifestyle images compared to those with a plain white background. Swiftly deploying these features, testing their efficacy, obtaining results and subsequently, optimising based on these insights is a gamechanger.
3. Banks and finance: The bank’s creative briefs are now being generated by artificial intelligence, reducing the time spent on back-and-forth communication with agencies by approximately one week.
Despite employing segmented tactics, companies frequently encounter resource constraints. Streamlining the development of creative briefs, visual content, and product descriptions enables swifter customization of on-site interactions. This move towards personalization doesn't necessitate relinquishing control entirely to Gen AI. Rather, it resembles an authentic one-on-one dialogue or conversational AI interaction.
Keeping the human in the loop with AI
Longacre emphasizes that each scenario he outlined involves human intervention. Given the early stage of AI development, it's unsurprising that most brands are beginning with 'human in the loop' applications. In these instances, AI generates outputs that are then reviewed and potentially refined by a person. 'Human in the loop' scenarios facilitate efficiency gains while mitigating risks stemming from errors or unforeseen outcomes.
"Perhaps the text is crafted by Gen AI, but a human provides oversight," Longacre explains. "While the image may be generated, it's not immediately deployed.
"We're starting to witness this trend, but generally, human supervision is present. Even in the case of chatbots. Chatbots have been in existence for quite some time, many relying on machine learning. It's crucial to discern when escalation is necessary. Knowing when to transition from the chatbot to a live agent for specific scenarios remains highly important."
Setting your brand up for success with Gen AI
In the journey of crafting a Generative AI strategy, Williams points out five key elements that play a pivotal role in ensuring success. They are:
1. Tech stack: Your tech stack is vital. You should have the ability to explore models, test use cases and choose the right ones.
2. A solid, mature first-party data foundation: Generative AI relies on the data to function properly, which means you must have robust data ingestion storage and management capabilities to make sure that the first-party data is accurate and as close to real time as possible to provide accuracy in the model outputs.
3. Human oversight: You still need that human in the loop intervention to make sure that what you thought was going to happen is happening and there are no anomalies.
4. AI-specific analysis skills: Leveraging AI requires the ability to interpret and accurately apply AI model outputs. You must ensure that your teams have the expertise to understand how the tools work and how best to put the data to work.
5. Process redesign: Consider existing processes or workflows that need to be resigned to take advantage of General AI.
Start small with AI for big results
When advising brands and organizations on implementing AI, I suggest beginning with modest steps. Initiate with a small-scale application that is easily quantifiable and doesn't demand significant upheaval. A prime area where our clients have achieved considerable success is in fine-tuning subject lines or enhancing the content of emails and paid advertisements. With the option of human oversight, this presents an excellent opportunity to explore diverse segmentation strategies and message variations through experimentation.
