Marketing Evolution - From Product to Customer
Marketing as a discipline has gone through a whirlwind of changes since its explosion at the start of the industrial revolution. In its most primitive stage, marketing was centered around generating interest and enthusiasm for a product or service among the masses irrespective of customer identity. The 1980s saw the dawn of customer identity. Enterprises began to take an active interest in understanding consumer needs and targeting their pitch to a subsection of the masses that they believed to be their core customer base. In this period, enterprises began to exploit media based advertising such as radio and television. Tele-advertising took off and has over the years grown to be one of the strongest means of marketing.
With the arrival of the internet age, there was a sudden advancement in technology that gave way to new marketing channels called Digital Marketing. In today’s world, where most users have a fingerprint online, corporations have the scales tilted in their favor. The mass consumer market is available for marketers to target and advertise online. The open secret to a successful marketing campaign remains the same - it lies in the enterprise’s ability to identify the right customer to market the right product. However, the underlying problem of no longer having to target customers as segments can now be resolved.
With the usage of a customer’s online behavior and preferences, we can now make marketing strategies and decisions targeted toward individual customers through Hyper Personalized Marketing. It allows enterprises to tailor marketing content to each consumer to maximize conversion rates.
Apollo Program - Hyper personalized marketing
As per a study conducted by SalesForce, 51% of consumers today expect that in the near future, companies they interact with will anticipate their needs way ahead of time and accordingly make relevant recommendations. To get to that level of personalization, one must first understand a customer’s identity. What kind of background does the customer come from? How old is she? What does she do for a living? Hyper personalized marketing considers a customer’s real time behavioral data and browsing information and has the potential to generate more leads for the enterprise.
Apollo Program, a technology-led audience insights company based out of New York City, turns entire data lakes into invaluable data assets for their clients through the use of artificial intelligence. Apollo’s business is centered around its patent pending technology that captures, organizes, and structures consumers’ digital behavioral signals to provide a competitive edge to its clients. Leveraging the billions of data signals that it captures, Apollo creates customer centric insights for its clients using which they can make fully informed strategic decisions.
Apollo Program had the vision to create an AI-based hyper personalized marketing platform that can help marketers tailor content based on individual preferences and achieve true hyper personalization. To realize this herculean task, Apollo Program partnered with Quantiphi, a category defining applied artificial intelligence company.
A holistic cloud based solution
As a part of this solution, each website being visited by a customer was classified into a certain broad category based on keywords contained in the web page link to better understand the web page’s offerings and value. For example - any webpage that contains keywords associated with the pyramids information would be classified under the “Arts and History” category. Any customers associated with such websites would be deduced to be interested in Arts and History subjects. Based on the strong correlation between a web page’s offerings and a customer’s key interests, a marketer would be better placed to make important marketing strategy decisions.
To realize their vision of achieving hyper personalization, Apollo Program and Quantiphi leveraged the AWS cloud platform.
Quantiphi was able to help break down the problem for Apollo Program into the following:
Cost-effective and fully managed ETL pipeline
- Quantiphi used services such as AWS Glue, AWS Athena, AWS Lambda, CodePipelines for scheduling a daily batch process and used Cloudwatch alarms, SNS alerts for data governance
- 60% reduction was achieved in monthly costs by implementing lifecycle policies on intermediate and raw tables, purchasing reserved instances for existing infra, and identifying non-active instances and stopping them
Creating a feature-rich Consumer Data Platform for both internal and external use
- Quantiphi developed a self-service Consumer Data Platform which automates tasks such as the creation of consumer pools with multiple combinations of pixels or non-pixel metrics, creation of customer assets such as Accounts, Brands, Projects, and Pixels
- Quantiphi developed a reporting feature in a Web App that provides automated consumer behavior insights and visualizations from consumer pools
Connecting Digital Asset Management & Consumer Data Platform (CDP) with web-scraping technology
- Quantiphi developed a web scraping technology to extract text, meta tags from millions of URLs to analyze the web content consumption of consumer pools across content categories (such as Arts & Entertainment, Politics, Food & Beverages, etc). We evaluated different series of instances, keeping network and capacity in mind, to optimize the scraping speed
- This web scraping technology is integrated with the Consumer Data Platform and allows for scraping of images, speech, and video
Adding intelligence to the Reporting of data pull requests (to be integrated with CDP)
- Quantiphi was able to provide consumer behavior insights by analyzing the consumption of consumers by content categories, domain, geography, device, demography, and time
- Quantiphi designed customer segments from their demographic and psychographic behavior. Analyzed the product and psychographic preferences of these customer segments using techniques such as market basket analysis, clustering, etc. Provided recommendations of the target audience and suitable mediums of online advertising using the customer segments and insights from their behavior
- Using NLP, Quantiphi derived keywords, named-entities, language, and sentiment from the web scraped data. For sentiment analysis, Quantiphi trained, evaluated a 3-class model using BERT, and achieved an accuracy of 90%
Revising the existing Pixel Server to make it more scalable, robust, and secure
- Quantiphi proposed a revised solution architecture for Pixel Server which reduced the Pixel fire response time to microseconds. We reduced the downtime and risk by implementing the blue-green deployment technique to move Pixel traffic to new infra
- Pixel Server can now integrate with multiple data partners, handle user opt-out requests, and scale to international geographies
With AWS cloud services, Quantiphi and Apollo Program were able to build a scalable, robust, hyper-personalized consumer insights platform that allows marketers to attract and convert high quality leads, improve brand perception and build deeper customer loyalty by interacting directly with the end consumer. Although we are still quite far from the idea of being able to anticipate the consumers' needs beforehand, this solution is a strong leap in that direction.
