How Ecommerce Analysts Can Prep For The Holiday Season

“It’s the most wonderful time of the year,” sang Andy Williams about the Christmas season, but a “wonderful time” might not be how analysts working in ecommerce and retail see the incredibly busy and stressful season. 

According to the National Retail Federation “Overall, holiday sales represent about 20 percent of annual retail sales each year, but the figure can be higher for some retailers. Hobby, toy and game stores report the highest share at about 30 percent.” 

In addition, because holidays sales don’t come with a corresponding increase in fixed costs, they can be much more profitable for retailers. And given the sheer volume of purchasing during that two month period, a lot is on the line when Santa comes calling, especially in these COVID-19 times.

The stress is even higher this year, as the 2020 holiday season should prove to be one of the most consequential on record and, as I explain below, a data warehouse with strong data strategy can be the foundation for powerful customer experience initiatives and demand forecasting models that will make any retailer more profitable.

2020: A unique holiday season

The 2020 holiday season promises to be one of the biggest ecommerce events on record. Even if a COVID-19 vaccine is developed and distributed by Christmas, a highly unlikely event, shoppers will be leery of gathering in overcrowded malls. 

COVID-19 has given shoppers the perfect excuse to stay at home and shop online, rather than spending hours of their day driving to a shopping mall, circling parking lots while trying to find an empty parking slot, fighting with other shoppers over coveted toys, or trying to keep sane around a thousand strangers as little Jimmy goes off on one of his temper tantrums. 

The short-term mandatory behavior of isolation COVID has created might just spill over to become a long-term trend that radically alters the ecommerce and retailing landscape. What retailers and ecommerce companies do today could have lasting effects well into 2021 and, possibly, for years to come. 

Getting your data house in order

In their article "Getting Your Data House In Order," McKinsey & Company state, “Whatever one’s business, the realities of today’s marketplace mean that those with the best data systems and capabilities will win—and by an increasingly outsize margin.” 

But how do retailers build powerful data systems when their data warehouse is a jumble of various software products that do a variety of different things, including data mining, business intelligence, data virtualization, ETL work, digital marketing, and analytics? 

It starts with a shared strategy around your company’s data, one that goes beyond the analyst team (or a team’s analyst) so that everyone from Marketing to Finance knows which data to collect, why it’s being collected, and what it means. Taking the time to establish these “ground rules” yields crucial benefits to an organization, including: 

  • A common understanding of data, which helps users throughout the organization utilize it better.
  • Improved data quality, which makes the data more usable in general. 
  • A clear sense of what can be done with the data, from getting a 360-degree view of customers to protecting sensitive information.

Once the data has been quantified, tagged, and visualized, trends hidden within that data will be revealed. You won’t just gain important operational insight, but key customer insights that can shape holiday strategy from messaging to outreach.

Holiday season prep ecommerce analysts can do now 

Along with establishing a data strategy, retail analysts can get ahead of the holiday season by laying the groundwork for key metrics now. Robust customer segmentation, Recency Frequency Monetary (RFM), propensity to respond, lookalike audiences, A/B testing, and customer churn modeling can help ecommerce companies glean information from their data sets that will help make the busy holiday season a success. 

Customer intelligence

A deep dive analysis of current customer attributes can provide demographic and psychographic detail that could be used to target new customers. This is also incredibly useful in lookalike modeling, which can be done on platforms like Facebook, LinkedIn, Twitter, Instagram, among others. 

Likewise, RFM models allow retailers to get a deep understanding of who their best, worst, and most frequent customers are. This enables brands to focus their marketing and outreach on their most valuable customers and to make promotions count by targeting audiences that are most likely to spend. 

A propensity to respond model reveals how likely a person receiving a marketing offer will be to use that offer. For retailers who rely on direct mail, these models can reduce costs by identifying customers who are most likely to respond to a retailer’s offer. After segmenting out people who are unlikely to respond to an offer, the business can then focus only on those who are most likely to take up the offer, saving substantial money in the process. 

Churn prediction and prevention

In his article Smart Strategies Require Smarter KPIs, Michael Schrage argues just reducing customer churn is no longer good enough—companies now need to focus on “learning from churning” as a means of predicting and actively preventing customer loss.

Schrage suggests that, “Distinctions must be drawn between churn presumed (a customer who simply stops engaging, that is, no more visits, purchases, etc.) versus churn absolute (the customer who explicitly closes an account or discontinues a service).” 

For example, a customer who’s had a bad experience will need to be addressed differently than a customer who’s being wooed away by a competitor’s products. By figuring out why a customer might move on, retailers can create relevant save or win-back strategies.

Demand forecasting 

In his article Cross-Channel Predictive Analytics for Retail Distribution Decisions, James B. Coles notes that fashion products have short lifecycles with notoriously volatile demand, with items going from introduction through sold out and discontinued in as little as four weeks. “With such short lifetimes, it is critical to distribute inventory as accurately as possible to ensure maximum satisfaction of customer demand,” argues Coles.

Whereas many companies might see this rapid turnover as a curse, Zara sees it as a competitive advantage because constantly updated stock motivates customers to return once every two weeks to see what’s new. But meeting demand would be impossible without a strong data program underpinned by a powerful analytical solution that crunches all these inventory, customer intelligence, and demand forecasting numbers. 

In fact, Zara is famous for creating quick and accurate demand forecasts focused on periods of just 3-4 days. That enables the fast-fashion brand to keep every store stocked while avoiding costly inventory overruns at any one location.

While many—ok, most—retailers aren’t in a position to create this kind of fine-tuned forecasting, creating demand models that take the realities of their organization into account could be helpful in sustaining customer interest throughout the holiday shopping season that seems to be getting longer and longer each year.

Conclusion

A deep dive forensics into last year’s sales and marketing campaigns is a worthwhile endeavor to do before the autumn leaves fall. There’s no reason to wait until January to make improvements that might increase sales during the most important period, especially in this uniquely challenging year.

If retailers can get their data house in order, the analytics ROI will follow. And Black Friday, Cyber Monday, Singles Day in China, Boxing Day in England and Australia, and New Year’s Day everywhere will be something for retailers to celebrate as they recognize that the holiday season might just be analysts’s most wonderful time of the year.

Get a free consultation with a data architect to see how to build a data warehouse in minutes.
Request Demo
Read more in:
Share this post:

Work smarter, better, and faster with monthly tips and how-tos.