Advertisers and marketing firms are bracing for the impact of Google’s planned discontinuation of third-party cookies on its Chrome browser by 2024.
Coupled with privacy regulations which differ from country to country, the pressure is on for enterprises to boost their respective businesses while tiptoeing on rules governing the use of people’s personal data.
This is where AI opens a window of opportunity, offering organisations marketing strategies anchored on privacy-first standards.
During a panel session, as part of Meta’s recently held “Data and AI Performance Summit” in Singapore, several industry experts offered their take on the changing landscape of online advertising, and how technology can become a catalyst for both privacy compliance and business growth.
According to Sunil Naryani, Chief Product Officer, Media, at Dentsu APAC, performance and marketing have become doubly challenging due to two reasons, with the first being media fragmentation:
“The media planners need to consider walled gardens, the open web, the endgame, and retail media, and find ways to improve cross-channel planning,” Naryani explained.
Secondly, the need to protect consumer privacy entails several restrictions.
“Protecting consumer privacy means that absolute granular data for every consumer is no longer available. As a result, insights have become more challenging, and the phase-out of third-party cookies adds another layer of difficulty. Building first-party data is now essential, but that’s not easy. How do you then do holistic measurement after you’ve done a media plan? It feels like the marketing ecosystem is evolving, but we need more robust solutions,” he added.
Crumbling away from third-party cookies
Presently, four in every five brands across APAC, equivalent to 79%, are still heavily reliant on third-party cookies, and around 56% of leaders in the region expect the discontinuation of third-party cookies will hurt their businesses.
Despite these figures, a lot of businesses have already moved into owning first-party data, in anticipation of a cookieless world, observed Praveen Kumar, Managing Director, Customer & Marketing Intelligence Lead for Southeast Asia at Accenture Song.
“That means everyone understands the importance of first-party data. Companies that have first-party data are privileged, but not everyone has it. For example, telcos and banking are rich in first-party data, whereas some consumer goods companies in Southeast Asia are still embarking on building up that first-party data,” he said.
Given this first-party data quandary, Kumar identified three challenges faced by their customers:
- How to integrate rich, first-party data with the rest of the ecosystem.
- How to increase relevance with identifiable audiences.
- How to have a holistic personalisation journey and strategy.
“What we have seen with third-party ecosystem integration, for example, if you’ve downloaded an expectancy calculator app, that means you are probably expecting to be a mother soon. While you probably don’t have that data, third parties like telcos do. So integrate that third-party data with your own first-party information, and that will help enrich your audience and efficiency,” he suggested.
One organisation lucky enough to have an abundance of first-party data coming from both its e-commerce platform and in-store sales is the APAC business of French multinational retailer Sephora.
“How do we give our members a great membership experience? What are the perks they get? How do we encourage them to give us consent to share their data in exchange for something great? It’s about the whole package, and that’s basically our strategy and our success,” shared Boris Bataille, e-commerce Director, APAC, Sephora.
As advertisers and marketing professionals may soon be losing access to insights gained from third-party cookies, how can AI plug some of the holes left by such disruption?
According to Sephora’s Boris Bataille, AI should be utilised based on data that is most relevant to the business.
“One of our principles is to maintain the action-to-data ratio. We use various matrices to qualify our customers based on their purchasing history. What do they purchase? Are they more stores-friendly? Are they more online-friendly? Are they on the rise in their frequency? It’s really complex,” he explained.
Bataille emphasises the importance of understanding which data is most valuable in terms of volume and quality. He suggests that in an AI/ML context, it’s crucial to identify the most reliable source of data. Bataille also notes that determining the most valuable data is essential for making informed decisions and moving forward.
For Accenture Song’s Praveen Kumar, AI could be very helpful in enabling continuous data measurement.
“Typically, when you do MMM (media mix modelling), or any other analysis, it’s done once in three or six months. You get a nice PowerPoint presentation, and you don’t know what to do with that because your 3.3 sale is coming up. A continuous measurement framework and AI can assist to measure performance continuously, and help make informed decisions. For example, a 30% discount may not always be the best, as a good bundle could work better with only a 10% discount,” Kumar said.
In response to the changing needs of the 10 million businesses that use personalised ads on Meta’s platform, the social media giant has built a range of AI-powered tools. Among these is the Meta Advantage Suite, composed of automated solutions powered by new ML models, which can leverage first-party data to help automate a campaign or workflow from end to end.
“We believe that personalised ads are incredibly valuable to people and businesses. They allow people to discover products, and also make sure that people are finding relevant services for themselves. Businesses need to make sure that they can reach the audiences that they know will be interested in what they have to offer,” said Annette Male, APAC Agency Director, Meta.
Meta, she continued, wants to ensure that everything they are investing and developing meets those expectations around transparency and choice, as well as consent on data usage.
With the need to protect consumer privacy, gathering insights at a granular level is more of a challenge, Dentsu APAC’s Sunil Naryani acknowledged. To this end, he suggested a couple of strategies to get organisations moving towards an AI-powered roadmap.
“Sometimes, the data you put in an AI/ML model is already biased. Consequently, you are going to get a biased output as well, which is not the best-case scenario. Thus, it’s crucial to have a clear understanding of the data you’re working with,” he said.
The second strategy, Naryani pointed out, is collaborating with domain experts who can identify the appropriate problem that needs to be addressed. Certain KPIs may appear suitable for the current use case, but they might not be meaningful for the business. Therefore, it is essential to ensure that the combination is there.
Last but not the least, organisations should familiarise themselves with the accuracy of the AI/ML models as they go along in their data journey.
“Clearly there are a lot of ad products now that, at the backend, deploy a lot of AI and ML, so embrace it. It’s here to stay. AI and ML can make life easy, and they can really drive outcomes,” the executive concluded.