Why Capturing Customers at the Purchase Intent Stage Can Sometimes Be Too Little, Too Late
In an age of audience and platform fragmentation, brand growth can be a tricky thing. V sat down with data experts Bryson Gordon, EVP Data Strategy at Viacom and Marc Ginsberg, VP & General Manager at AMEX Advance to discuss the power of predictive intent, as well as how it can impact brand affinity and grow your brand at the right time
V by Viacom: Let’s start with the obvious question: what is predictive intent and why is it so important for marketers to capture their audience at this stage?
Bryson Gordon: Predictive intent allows you to leverage massive sets of buying pattern data so you can predict purchase intent before an individual actually buys anything. Basically, we can predict what someone is going to purchase before they even start thinking about it. We call this pre-intent, and we know this part of the marketing funnel is really important to brand building — specifically within some key advertising categories. In some cases, by the time purchase intent is formed, the ability to influence brand affinity is low — you’re basically late to the party.
Marc Ginsberg: Very true. With traditional targeting, most advertisers fall back on demographics and survey data to target ads to TV viewers. With predictive intent, we can add real insights that help increase the relevancy of their campaigns. We see the types of transaction our card members are making across a wide variety of industries — everyday things like buying a cup of coffee to large purchases like a washing machine or a new car.
What we’re really looking at are those transactions in aggregate to identify patterns between lifestyle moments that matter and buying habits and preferences. This helps us to identify purchase behavior that foreshadows something bigger.
V: Why did Viacom and AMEX join forces?
MG: Basically, we saw a unique opportunity to help change the TV advertising industry and speed up the adoption of advanced audience segments using data analytics. Data has always been a huge part of how AMEX delivers important customized marketing programs to our merchants and card members. We’ve been able to build models utilizing billions of real-time spending data points across our network, which allows us to very accurately predict a specific consumer group’s buying behaviors and habits — often before they even know what they need or want.
Importantly, we then are able to measure the effectiveness of the model’s ability to predict intent against other sources using actual purchase data.
To be clear, American Express is sharing anonymized forecasts and models with Viacom — not personal data — to ensure that card member privacy is protected.
BG: From our perspective, it’s all about how we can help marketers. With digital, marketers have enjoyed a level of control over defining segments and delivering advertising messages to them. For years, this wasn’t the case in television, but this has radically changed with advancements in TV viewing measurement and data fusions. At Viacom, we’re now applying advanced targeting to the scale and premium context of television — partnering with AMEX allows us to add a real level of sophistication to this.
V: Context IS everything. What types of insights are you able to glean from all this data?
MG: Since we know the best predictor of future purchasing behavior is what consumers have bought in the past, and the time sequence of those purchases, we can study a group of Card Members who have followed that particular purchase behavior and then see what’s unique about their spending leading up to that purchase.
Let’s use an example. Let’s say you start to spend more money on lawn care and household goods. You’re also increasing spending on sporting goods. While these aren’t necessarily the reasons you may end up getting a dog, they’re powerful indicators that suggest your likelihood of getting a dog is pretty high. We know this because we use independent variable analysis (the idea that commercial behaviors outside of one category can be predictive of intent within another) to connect seemingly unrelated data points. Similarly, we know that an increase in alcohol consumption and restaurant spend are just some of the indicators among men that they are likely to get engaged!
V: So, how can we take this information and then harness it?
BG: When we understand the interactions and patterns between habits and buying behavior and how they impact the moments that matter, we can use that data to improve the context and relevance of advertising. This transforms the traditional approach to advertising — interruptive push messages — to a new ad engagement model where you’re delivering the right message to the right people at the right moment in time.
MG: It’s also about knowing when not to talk to someone. For example, a decision that happens once every 5–10 years, like buying a car. Why spend ad dollars advertising a car to someone who has no interest in buying one in the next few years? While you definitely want to stay relevant over time with that customer, their spend should proportionately increase as they move closer to pre-intent. Once a potential consumer knows they need a car and what they want, it’s probably too late. They’re now just looking for information for the negotiation process, not the actual choice of automobile.
V: So, what does that mean for marketers? How can we use data to drive more of their brand growth?
BG: The power of data allows you to not only drive brand growth, but to drive brand growth at the right time. Knowing when to engage your audience is key. I can’t say that enough.
MG: Yes, we’re now at the point where the right data at scale (combined with improved technology and big data capabilities) is having a positive change on the effectiveness of traditional TV advertising. Based on the combination of millions of unique triggers, it’s allowing us, as partners, to deliver highly relevant messages to consumers when they want it, need it and usually before they know it. And then measure its effectiveness. It’s a big step forward.
