Google’s SEA and South Asia Frontier VP Sapna Chadha speaks to WARC Asia Editor Rica Facundo about the impact of artificial intelligence on creative effectiveness, decision making and best practice.

WARC: AI can enable marketers to produce creative efficiencies at scale but efficiency is not the same as effectiveness. What does it mean to use AI to drive effectiveness and what's that key difference between effectiveness and efficiency?

Sapna Chadha: Many of our clients ask us how they can reach more customers. In the traditional world, to reach more customers, you have to pay more. But we’re moving away from that model with more customers using AI solutions to hit their targets more effectively, without it costing more.

This is happening because they’re being more strategic with their goals and specific about who they want to reach and seeing it pay off, and they are now leveraging AI in the form of automation, optimisation and improvements in measurement. This provides a sandbox for experimentation, which in today’s world of AI is really important. It gives people that flexibility to do the trials that they’ve always wanted to do.

In the past, marketers were just held ransom to the goals they had to hit. Now they can hit the goal and save money while doing it.

Cost savings is often described as a benefit of AI-driven solutions. Are there any other benefits?

People don’t often talk about time as a factor for both efficiency and effectiveness. People are more effective when the cycles of delivery are shorter. Gone are the days of taking 18 weeks to run a campaign. Now with a shorter window of time, marketers can run while ensuring that they are hitting their targets at the same time.

The other benefit is being able to optimise towards effectiveness in real time. Marketers follow a calendar of events of key occasions and seasonal moments. Keeping up with this calendar can be difficult and marketers run the risk of falling behind. But what we’re seeing is that with AI, marketers can even get ahead of key moments because the data science enables campaigns to be optimised in real time.

In the past, there had to be these war rooms during a campaign where people were monitoring it every hour to ensure that the campaigns were as effective as they could be.

The Harvard Business Review says there are a quadrillion permutations of how somebody can serve an ad, at a specific moment, to a customer. How do you optimise with a quadrillion permutations? In the old world, people used to sit back and try and figure this out. Now with the way that the models are working, it allows you to do that more effectively.

Creative decision making has been a matter of subjectivity but AI-driven recommendations mean that algorithms increasingly dictate best practice. What does this “informed” approach mean for creative effectiveness if storytelling is reduced to what the algorithm recommends?

Let’s use the example of having to labour over the various permutations of a headline. You still have to put in the time but now, you can do it in a more structured and refined way. E-commerce platforms are doing this exceptionally well. They have numerous merchant feeds on their platform. Is someone sitting there to write every title that exists?

Now, you can conduct real-time testing. Before, you could A/B test and see results over time. But now, that feedback is happening in the moment, so that you immediately understand what’s working better. We have tools that do it but our customers are building their own bespoke solutions to be able to help them do more real-time testing. So creatives are being tweaked and changed and optimised at a different speed than it has existed.

Despite the hype around AI, there’s still a sense that creatives are hesitant to use it because of how it might impact their job. How do we address this tension?

AI won’t be able to have the empathy to create human-led insights or truth. But it can help us write better briefs faster and more effectively. I used to tell my marketing teams that a brief should take you a long time to write. But now it’s the opposite.

AI should help us to work more frequently and deliver better work in a shorter amount of time, thereby making it more effective. Most of the time, you’re chasing a trend but then, the trend is over. If a trend only lasted for 48 hours, you might not be able to react to it but now, with AI, I would argue that you can.

And so the creative is going to be tasked to just work at it at a different pace and to have the tools to be able to do that.

The modalities change but who’s responsible isn’t. I think the tools are going to allow us to do what we've always wanted to do, such as act in the moment, trendjack in the way we wanted to or provide extremely personalised communications.

How do you see AI impacting the briefing or ad testing process?

I used to say that agencies weren’t able to crack the brief because they’re constrained by time and resources. AI will help us overcome these challenges and do marketing the way we’ve always wanted.

Testing shouldn’t be that we tested for 30 days and saw that was better. Rather, it should be that we tested for 48 hours, we know this is the insight and therefore we know it’s working and we’re optimising for that.

It allows you to optimise in the moment for where things are more effective and without having a media planner trying to figure it out on their own. It allows you to do it in a more scaled way, which will give people more confidence because there's a science behind it.

If we’re constantly testing and experimenting, what does this mean for how the industry views what’s best practice? 

There's the time frame for when best practice changes. Before, best practice could last for a few years. Now, best practice is in real time.

Also, best practice for each brand is going to be very different. That's also the power of AI. You can take best practice from other categories but you can also create and refine best practices in your own organisation.

There's actually an overload of best practices in the industry right now. How do you even figure out what is best practice? This is where you could argue that tools such as AI allow marketers to sift through the millions of best practices that will help us get to more refined and synthesised insights.

A concern of AI-driven solutions is that it can be a black box that’s making decisions for brands without understanding unique attributes or specified nuances of a brand or audience. What advice can you give marketers on how to mitigate this risk, and ensure that there’s a balance between trusting the algorithm and a nuanced and unbiased understanding of human behaviour and motivations?

We always advise our clients to be as clear and specific as possible with what you’re trying to achieve. And the more that you can feed the models, the better.

Are people being hands-on enough with the tools? Actually, what happens when you work on a product like Performance Max is that you're feeding it constraints and what to optimise for.

This specificity also translates into the output. We’re listening to our customers on what they are expecting from us in return, such as providing details on asset performance, or audience personas and insights. 

I liken it to when creating an analysis in Google Sheets. When the answer comes out, you don’t question why the answer is the right answer because you know it’s right. As AI matures and people begin to experience its ability to deliver, they will also begin to build confidence and expertise in leveraging it to their advantage.