Beyond the hype and the paranoia — what generative AI means for marketing

Over the last year, AI has gained a formidable reputation as a disruptive force due to the rapid adoption of chatbots and image creation services powered by generative AI — systems capable of learning from pre-existing content to create something new. It’s tempting to think of the emergence of these tools as sudden developments and imagine gloomy scenarios where they replace human talent, creativity, and innovation. 

 

The truth is far more prosaic. Companies from social media giants like Meta to global communication conglomerates like Publicis Groupe APAC have been using AI for decades as an enhancer of skills. At Meta, $30 billion to $33 billion of capital expenditure has been allocated to infrastructure in 2023, including AI and apps. 

 

Speaking at ‘Magic, Media and Machines,’ a summit on integrating media and AI organised by Publicis APAC and Meta in Singapore, Laurent Thevenet, head of creative technology at Publicis APAC said, “AI is a growth accelerator. It’s not just hype where everyone forgets about it the moment something new comes along. This will create new business models and opportunities.” Carlos Palacios, strategic connections planning lead at Meta added, “We have used AI to fundamentally improve the way that our advertising system works. The benefits are becoming apparent, and we have seen a 20% increase in conversions YOY and decline in cost per acquisition.” 

 

Some of these business models and solutions impacting creativity, media, and targeting have been co-created by Publicis Groupe APAC in partnership with Meta, and point towards greater collaboration, not conflict, between humans and AI. 

 

AI’s transformative impact on customer loyalty 

 

Instant messaging has evolved from being a purely personal communication medium. Today, customers often use messaging services to make enquiries or resolve problems with a brand. Citing a study from BCG, Anna Soliman, agency development lead for APAC at Meta said, “One in three APAC consumers message a business once a week.” 

 

The next step is AI-powered messaging. With millions of active users in Southeast Asia across Messenger and WhatsApp, it makes sense to use these platforms as a venue for customer-centric data-driven promotions, tailored to a brand’s needs. 

 

Originally developed by Digitas in the Philippines, Ultrascan (U-Scan) is an AI-powered bot accessed through Facebook Messenger and WhatsApp that can read consumer-uploaded receipts. Once requisite permissions have been granted, U-Scan begins a deeper engagement with consumers — from purchase reminders to instant prizes during promotions. 

 

Speaking about the advantages of using AI for Ultrascan, Soliman said, “The AI scans receipts with over 98% accuracy. Our AI also learns fast and can process thousands of receipts for any store promotion very quickly.” 

 

The lead time for creating a promo can be reduced to weeks instead of months, with instant gratification for the customer.  

 

Making communication more measurable and effective 

 

Correlating media spends to business outcomes is one of the most persistent challenges faced by the marketing industry. With AI solutions such as Robyn — an open-source marketing mix modelling code from Meta that builds thousands of models simultaneously — better predictions on spending and efficiency are now within reach. 

 

Stemming from requests by marketers who wish to quantify media data and correlate its business impact, Publicis Groupe APAC and Meta collaborated to create the MMM Model Centre, a user-friendly modelling tool that allows marketers to run different scenarios. Bhavna Shewakramani, solutions partner at Publicis APAC said, “We leveraged the skills of the Publicis technical and analytical teams and, using Robyn as a foundation, enhanced it to cater to specific client needs while maintaining its simplicity. People at varied analytical skills levels can use the tool.” 

 

How to deploy AI effectively 

 

With myriad use cases, navigating the path from AI awareness to using it in an optimal manner is something marketers and agencies are still grappling with. Publicis Groupe APAC and Meta’s experts had a few suggestions on the steps that ought to be taken to get the most out of generative AI-powered marketing. 

 

Test and scrutinise results: Experimenting is an essential first step on the AI journey. Meta recently launched Advantage+, a suite of ad automation solutions. Supriya Singh, product marketing manager at Meta said, “Advantage+ shopping campaigns help advertisers get smarter and faster on which campaigns are converting. It eliminates the manual steps of ad creation and automates up to 150 creative combinations at once. This helps advertisers more quickly learn what ads are working, while making the most of their advertising budget.”  

 

However, she recommended that brands test campaigns and learn from results before buying into a value proposition, since AI is amid a hype cycle. Examining the impact of a specific incremental change to a campaign is vital. Singh explained, “There’s a Harvard Business Review study that says if you run over 15 experiments a year on incrementality tests, you are likely to see about a 30% increase in ad performance.” 

 

Human oversight is necessary: With AI quickly acquiring buzzword status, there has been an enormous leap in the quantum of just average work, particularly through the course of this year. Jacqueline Pavich, Publicis global client lead, Publicis APAC said, “It can’t be just left to the machine. There has to be a human element. AI can expedite and accelerate how we reach consumers, but generative AI ensures that we can create content that is entertaining and hyper-personalised. My hope is that the combination will result in infinite possibilities to engage our end consumers.” 

 

Oversight is also crucial when it comes to ethical frameworks surrounding AI. Andrew Male, client partner, Publicis Sapient said, “Implicitly trusting AI to deliver what is right is definitely not the correct approach. Ensure that you have humans in the process at the right stage, so that any biases are addressed.” 

 

Moving AI in lockstep with governance frameworks: The newness of generative AI means organisations are in uncharted territory when it comes to the legal and governance framework within an organisation. Male said, “Governance is what runs businesses. Our ability to execute quality ideas needs to happen around the right framework, and guides on what we can and cannot do.” This is particularly true in cases where there are no legal precedents on how to move forward safely while ensuring that a brand and its partners are protected. 

 

The constantly evolving nature of generative AI adds a layer of complexity. Male said, “Generative AI capabilities of today are not going to be the same six months or two years later. That iterative change in technology is something we must think about, ingest, and bring into the business. Those guides and guardrails effectively create a safe space for everybody to be able to play, test, iterate, and learn.” 

 

As a starting point, Publicis Groupe has built its own secure version of ChatGPT, excluded from sharing any of the data with the LLM’s wider userbase. 

 

The path ahead for generative AI 

 

The experts from both Publicis Groupe APAC and Meta are confident that we have merely scratched the surface of what generative AI can offer, and the only way forward is to test, try, and iterate. Being adept at generative AI can be a competitive advantage both for brands and agencies, as well as individuals who wish to stay ahead of the curve. Or as Pavich put it, paraphrasing a quote she had come across at Cannes Lions earlier this year, “‘AI will not necessarily take our jobs, but people who understand it will.’ There is urgency — not just in planning for tomorrow, but for the next 10 years.”