Tapping the e-Commerce Opportunity with Applied AI  During and Beyond COVID-19
Updated on

Tapping the e-Commerce Opportunity with Applied AI During and Beyond COVID-19

One certainty of the COVID-19 pandemic is that buying behavior is shifting towards e-commerce. Studies conducted in February 2020 suggested that,

US retail would grow by 2.8%, but new data shows that there will be a 10.5% decline in brick-and-mortar in 2020, while e-commerce will grow by 18% in 2020.

As a result of this, we expect that companies that create a strong digital experience will be able to balance the risk and overcome the economic downturn by getting closer to customers and understanding their changing needs and behaviors, while maintaining “social distancing”.

It is also critical for companies to understand that these tactical solutions, implemented to get through immediate crisis, should be designed with a bigger aim in mind – building towards a broader digital strategy.

This creates the need for well-designed e-commerce AI architecture with real-time customer personalization capabilities that will allow companies to connect physical shopping experiences with online shopping behavior on any relevant digital channel.

Delivering customer delight with enhanced e-commerce AI architecture

1. Identify and leverage incremental data streams:

These may include owned customer behavior data, social media data, demographics data and traditional primary research-based stated data.

2. Utilize behavioral science that influences Digital Sales:

Identify and establish key customer segments based on respondents’ attitudes, habits, need states and purchase occasion behaviors. Before identifying the segments, take care to bucket products into three categories based on their consumption trends in the Covid-19 era:

  1. Accelerated: Such as grocery and food items
  2. Emerged: All PPE items
  3. Declined: Such as apparel

3. Develop new customer personas:

Go beyond the broad segments and identify micro-segments which are now more actionable and are based on real-time consumer behavior. Identify micro-segments from standard segmentation using transactional POS and social data and other macro level variables. This will help track how customers move among segments over time, increase conversion rates, and predict future profits the customer will generate. Key buyer personas to look for in the COVID-19 era include:

  1. Worrier: Panic buyer. Needs very high attention with immediate product recommendation.
  2. Deal seeker: Buyer who is price sensitive, looking to find the Best Deal. Needs right promotions/discounts through the right channel based on propensity scores.
  3. Attribute-Anchored: Buyer whose shopping is driven by a single attribute. Needs a personalized experience with cross-sell and up-sell offerings.
  4. Ambivalent: Buyer who completely abandons transaction during checkout. Needs personalized experience on the platform.
  5. Brand seeker: Buyer whose shopping is driven by preferred brand. Needs personalized experience until delivery happens.

4. Map online customer journeys:

Map purchase journeys and life cycles across identified personas. Identify opportunities to optimize journeys to drive higher customer conversion and engagement.

5. Build an AI-led product recommendation engine:

Design, develop and implement recommendation engines in real-time to optimize cost and generate revenue while driving traffic, delivering relevant content, engaging shoppers, converting shoppers to customers and increasing number of items per order, increasing average order value, etc. This involves:

  1. Offer bundling models: Association models to identify what offers and product combinations are purchased most frequently by targeted customer segments. Perform this exercise for each product-offer combination for each interested segment or across all customers.
  2. Response modeling: Propensity models to understand likelihood to respond to offers or products favorably. Perform this for each product bundle for each segment or across all customers.

Some key steps to follow:
• Generate a list of products/offers that are likely used by the micro-segment.
• Correlate right event with right offer in the event analysis.
• Generate rules based on business inputs for each recommended offer.
• The offers that have high profitability or business focus will be weighted higher for selection.
• Release offer via “ideal” channels based on customer profiling.
• Record offer acceptance/rejection response to re-train the model.

For new to file customers, identify “look-alike” customer groups using a look-alike classification model. This personalization strategy should be extended across multiple channels and devices even when the customer is not active on any platform.

6. Focus on holistic customer experience:

While AI-led product recommendations help identify conversion opportunities across customer segments and buyer personas, the focus should be on delivering the best possible end-to-end customer experience along with it.

Employing AI-led Personalization and Marketing Automation with Course5 Intelligence

With our AI-driven insights and tools, we at Course5 are constantly exploring new opportunities for businesses to harness emerging technologies and restore business growth.

Our Applied AI team is working with global clients to train AI and advanced analytics models with domain-specific data and build solutions, such as —

  1. Omnichannel personalized product recommendation systems
  2. Marketplace optimization solution
  3. Digital media optimization solution
  4. Search-based demand forecasting (forecasting demand based on online search behavior of customers)

Anshul Singh

Anshul Singh

Anshul Singh is an analytics consultant, product manager, and business analyst with 3 years of industry experience. Anshul specializes in the digital domain, driving various...

Read More    Read More