Case Study: Gap, Inc. Improves Marketing Results with a Personalized Approach
A conversation with Noam Paransky, SVP of Digital at Gap Inc., about their strategy and technology
Savvy marketers know that today’s consumers are picky—they no longer respond to batch-and-blast marketing campaigns directed to a broad swath of buyers. They want to connect with brands who are as authentic and unique as they are. They look for personalization in every interaction. With that in mind, large companies with broad consumer appeal have a challenge to overcome.
The digital team at Gap, Inc. (which includes the following brands, Gap, Banana Republic, Old Navy, Athleta, and Intermix) are doing just that. I interviewed Noam Paransky, the Senior Vice President, Digital of Gap Inc., to discuss how they’ve personalized marketing with a sophisticated digital marketing tech stack. Their approach enables the team to target customer segments and deliver personalized marketing messages, which makes their customers feel as if they are interacting with a distinct, unique brand rather than a giant company.
The cornerstone of this strategy is their own customer data platform (CDP). This brings together all customer data into a unified customer profile. By using a combination of first-party and third-party data in its CDP, Gap Inc. can use this information for personalized marketing across all systems and digital advertising channels to target their customers. Then, using dynamic content optimization technology, Gap Inc. is able to further customize and personalize their messaging and creative based on real-time data about the viewer when the ad is served.
Clearly, the process and strategy that Gap Inc. has developed are innovative and pretty impressive. You can learn more about Gap’s approach in my interview with Paransky below.
The Interview
Veronika Sonsev: Why is personalized marketing important for Gap Inc. brands, which have such broad consumer reach?
Noam Paransky: Regardless of the size of the brand, customers are seeking relevant points of connection. Given the large and diverse customer base that we serve, personalization is critical to creating relevance.
Sonsev: What kind of data do you use in marketing personalization?
Paransky: Our data science team built a proprietary CDP leveraging both third-party and internally developed components to store our first-party data from all available sources. The CDP is essentially our single source of truth about our customers. We augment our first-party data with third-party data like demographics, interests, and life stages to further segment our audience before syndicating these audiences to advertising platforms like Google and Facebook.
Past purchase behavior only tells part of the story. We have a data science team that uses the combined first-party/third-party data to help us predict customer propensities and behaviors before targeting them in digital channels. After we create these hybrid first-party audience segments. We can leverage the advertising platforms to create lookalike audiences, enabling us to market to new prospects much more efficiently.
Sonsev: Once you identify target segments, how do you deliver the right message?
Paransky: It’s a combination of art and science. For each audience, we need a strategy of what we are looking to accomplish. Then, we develop a library of assets and messaging we believe is relevant to that strategy. Then we use capabilities like dynamic content optimization (DCO) to help us personalize the content we show consumers and machine learn into the winning messaging. We also leverage tools like Persado [an AI solution that optimizes text copy against a customer’s preferred emotional sentiment or ‘voice’] for any associated headlines and descriptions. We started by utilizing them for email subject lines. Then, expanded the use of this tool to other channels, including personalized messaging in select use cases on our websites.
Sonsev: What’s the role of AI and automation in personalized marketing?
Paransky: Gap Inc. has tens of millions of customer records in our CDP, which is too much to segment manually. We leverage AI in the CDP to help with customer resolution, segmentation, and clustering.
Additionally, we use AI and automation in managing advertising bid management and content optimization.
In the past, we ran an in-house deterministic matching routine to create a unified view of a customer across channels and transactions. The model took hours, if not an entire day to run. This resulted in mediocre results and the ability to only run the model once a week. Now, we are partnering with an AI start-up called Amperity which allows us to execute near real-time probabilistic customer matching. We see much better match rates with high degrees of accuracy – and the ability to resolve in real-time allows us to immediately change the marketing strategies and messages to the customer. Whereas, before we could have spent nearly an entire week serving an irrelevant and possibly confusing message to the customer.
Sonsev: Is there a point when too much targeting yields diminishing returns?
Paransky: At Gap Inc., we develop a hypothesis as to which audiences have exhibited purchase behavior signals and seek to target those consumers. We then examine the results of our targeting efforts and refine our strategy. If we don’t get a meaningful lift from an audience sub-segment, we will fold it back into a broader audience. Given we are relatively early in what seems like an infinite journey, we think we are a long way from seeing diminishing returns.
Sonsev: What kind of results are you seeing from your marketing personalization efforts?
Paransky: There are a lot of moving pieces in the journey of optimizing our digital marketing activities and related spend making it difficult to attribute exactly how personalization is contributing. Our best read is that our digital marketing uplift efficiency is approximately 50% higher than it was last year.
Sonsev: How do you measure performance and incrementality?
Paransky: Measuring incrementality is very difficult — we’re still in the crawl stage and getting closer to walking day by day. We use a hybrid of mixed media modeling [a technique which helps quantify the impact of several marketing inputs on sales] and multi-touch attribution in monthly modeling cycles. We then use what we describe as proxy metrics [a technique that uses the mixed media modeling/multi-touch attribution variable coefficients like impressions and clicks to attribute revenue to a specific marketing tactic] to manage the channels as effectively as possible in real-time.
The current reality is that there are no real-time tools to perfectly measure attribution and incrementality. The key for us is to understand the value of the channels and execute campaigns in the right channels. Doing this effectively, we need to balance targeting efforts throughout the customer journey, not skew towards lower funnel marketing tactics.
For an organization like Gap Inc., it can be tempting to simply broadcast your marketing messages across all marketing channels. However, Gap Inc. knows that consumers come in all shapes and sizes and they’ve built a marketing machine that can take on this challenge head-on and data first.
Sources:
https://www.makethunder.com/what-is-dynamic-creative-optimization-dco/
https://towardsdatascience.com/market-mix-modeling-mmm-101-3d094df976f9?gi=f42c9713f792
