Understanding AI in Commerce to Engage Your Customers

Are you finding it increasingly difficult to keep up with maintaining new features required inside your business that are also continuously expanding sales and building relationships with your customers? In the dynamic world of e-commerce it used to be enough for a site’s content to simply be searchable– Can customers find your products? Next came increasingly comprehensive SEO– Are your items properly search enhanced with the correct keywords, titles and meta-data? Then, focus shifted to the site’s social presence– Do customers like your items? Do you respond quickly and honestly to issues and concerns? Next, mobile-enhanced websites became a necessity to reach customers where they want you to be.

Sometimes, new features are easy to implement (like full site SSL,) while other features take more effort (like making a site work well with mobile). More than ever, customers are shaping our businesses. Now necessary, the latest e-commerce feature desired by forward-thinking customers is the effective use of artificial intelligence.

Today’s artificial intelligence (AI) can be referred to by many names: Machine Intelligence, Machine Learning, Deep Learning, Natural Language Processing, Supervised Learning, or Unsupervised Learning. In general AI uses large data sets to create maps that link one type of data to another that can be particularly useful in commerce focused organizations:

  • A customer may post an item on social media, and AI can create a map to identify whether the customer is angry, frustrated, happy, or confused. This type of AI mapping, known as “sentiment analysis,” uses a type of supervised learning called the “bag of words” model.
  • When a customer browses a site and purchases items, their clickstream can be mapped to customer specific product recommendations. If a customer desires scrapbooking supplies there is no point in showing her fabric swatches, give her what she wants!
  • Automatically generate relevant captions for customer supplied photographs and may identify the site’s product(s) in customers’ photos to leverage social marketing and community building opportunities.
  • Improve commerce sites is to help marketing teams identify how the site can best group products or create bundles that leverage customers’ natural browsing and purchasing patterns. AI is able to update the site’s dictionary by discovering missing search terms automatically without human help.

Salesforce has developed a number of AI tools across its portfolio, ranging from automated lead prioritization in Sales Cloud, to image classification in Marketing Cloud, to identifying interests and sentiment in Community Cloud. AI tools currently available in Commerce Cloud include:

Commerce Insights
Commerce Cloud analyzes order history and identifies how often items are purchased together. Marketers can then use the data to improve site design, build bundles, and develop promotion strategies to improve the average order value. Although this type of analysis can be applied to e-commerce sites of any size, it is especially powerful for sites with a large product catalog.

Product Recommendations
Product Recommendations are a valuable / necessary e-commerce feature and have been known to increase sales by 10% for some sites. Within Salesforce Commerce Cloud, Einstein Recommendations systematizes the steps necessary to enable effective product recommendations on a site. The steps required are: turning on automatic data analysis, updating the site code to use recommendations; using a Chrome plugin to validate the recommendations, enabling recommendations on the live site, and A-B testing on the live site to verify the recommendations’ values.

Predictive Sort
Predictive Sort uses individual customer specific clickstreams to create customized product sorting in search results and product grid pages. The steps to implement Predictive Sort include minor site code changes, scheduling of order and catalog import jobs; updating existing sorting rules as necessary; enabling Predictive Sort in production, and A-B testing on the live site.

Search Dictionaries
Einstein Search Dictionaries are similar to Product Recommendations and Predictive Sort but does not require any code or developer assistance. The Einstein Search Dictionaries (ESDs) require site search term approvals by the marketing team. ESD is built upon the data feed process described above and requires the product and orders to be analyzed by the automatic processes.

Can you imagine the possibilities for engaging your customers? If you’d like to explore how AI can help you meet both customer expectations and your objectives using Salesforce, learn more today.

Previous Article
Getting Started with Preference Centers
Getting Started with Preference Centers

Preference centers can significantly help businesses cater to a customer’s individual needs and interests w...

Next Article
Preference Centers 101: Customer Marketing Solutions
Preference Centers 101: Customer Marketing Solutions

Preference centers are a valuable tool in the belt of a digital, cross-channel marketer. Simply put, your p...