Kumo.ai: The One-Stop Shop for Predictive ML in Ecommerce

Author: Ivaylo Bahtchevanov

In today’s fast-paced world of ecommerce and online marketplaces, AI and predictive ML have become essential tools in allowing businesses to stay competitive and drive sustainable growth. Consumers have more options than ever on-demand, so winning them over means anticipating their needs and while delivering the best digital experience. Ecommerce businesses need a sophisticated ML practice in place in order to build a nuanced understanding of their customers, personalize their experiences, and make the best decisions to drive the business forward.

In the traditional way of performing ML, your team needs to create many different models to solve multiple problems, and each use case requires building extensive infrastructure, dedicated ML pipelines, production tooling, and maintenance efforts. As you start adding new use cases, you need to allocate more resources and time to build them out. Additionally, as new data on user behavior keeps coming in, significant effort is required to maintain all of your pipelines so that your models will reflect the most up-to-date state of the world.

We built Kumo to bring AI directly to your data and make it easy to solve these problems out of the box. Kumo is the one-stop shop for empowering the growth and marketing teams for ecommerce. Customers connect their data to Kumo, and then Kumo builds a graph that learns relationships and interactions based on user activity, transactions, marketing data, and more. Kumo uses state-of-the-art ML specifically optimized for your business data to arm your growth and marketing teams with the tools you need to improve your business.

Below are just a few of the many problems ecommerce businesses solve with Kumo.

Identifying customers with the highest potential CLTV

For any B2C business, it’s important to focus on the users that have the potential to spend the most on your platform. Your data contains interactions from many users, but most of them will never become valuable or highly profitable customers. Understanding where the most relevant users for your business are early on in their interactions with you allows you to prioritize where to spend limited marketing dollars. Developing customer engagement strategies with these users will help you unlock higher profitability through a superior customer experience.

Kumo leverages signals from your entire enterprise graph to predict future behavior of users – including actions critical for moving your business forward. The platform can quantify the future spend for any given user over any specified period, automatically taking into account key drivers such as average order values, purchase frequency, customer retention rates, among other factors. In addition, the platform offers powerful scenario-planning tools that allow you to see how future customer spend will change as you implement different strategies for engaging your users.

Managing the end-to-end customer lifecycle – from acquisition to engagement

Acquiring new customers is usually done through paid advertising. Kumo helps you identify which leads to focus on based on downstream value and likelihood of conversion, which drastically decreases the cost to acquire users (CAC). Additionally, creating highly targeted and personalized marketing strategies will make those outbound efforts more effective, leading to a higher percentage of conversions.

In addition to acquiring new users, your business depends on making sure those users are able to complete critical actions that move your business forward. Converting users to loyal, paying repeat customers means optimizing around a set of metrics that include click-through rates, minimal shopping cart abandonment rates and bounce rates, repeat purchase frequency, and more.

Each of these steps in the funnel can be optimized by personalizing the user experience and then combining with the best organic outreach methodologies – typically done through in-app or platform notifications. By sending the right message at the right time, notifications can be seen as an extension to your platform, funneling the users exactly where you want them to go based on where they’re at.

Leveraging the best channels for a response, finding the optimal volume and timing for each message, and identifying the most relevant products and respective summary lines to resonate with each user are all factors that will make users more likely to respond and dramatically increase the return on investment to bring them onboard.

Using Kumo’s platform, you can consolidate the end-to-end ML process and build your entire outreach strategy in a single sitting. Kumo’s predictions not only tell you which set of notifications are best for each user, but also run through a set of potential actions to understand how your users will behave based on different notifications strategies and timing.

Improving Loyalty and Retention

There’s often a sense of urgency in proactively identifying churn risk since there’s a disproportionate cost to re-acquiring the customer relative to re-engaging them (on average re-acquiring is 5 times more expensive than retaining a customer). Improving retention early on even marginally can lead to drastic improvements in profitability, and can help you convert low-engagement customers into loyal, high-spend customers with the right approach. Similarly, the success rate of selling to a customer you already have is 60-70%, but this drops down to 5-20% when selling to a new customer.

Predicting churn can refer to identifying high impact events, including membership cancellation, uninstalls, unsubscribing, or deactivating. It can also refer to predicting significant changes in activity that signal reduced engagement – i.e. declines or drops in views, clicks, or purchases.

The flexibility of the Kumo platform allows you to specify what you consider to be a churned user in the context of your business by specifying the activity you want to track. Given that definition, you can then predict which users will churn at what time, identify the leading causes per user, and then build a highly effective retention strategy.

Simply put, you can ask questions such as: “which of my users will churn in the next week / month, and what are the actions we need to take to retain those users?” And then subsequently convert them into loyal, highly profitable customers.

Personalize the shopping experience

The most effective way to create an excellent shopping experience is through personalized product recommendations. Kumo gives you the ability to recommend any products or content to each user, based on individual preferences at a given point in time. Traditional ML will analyze previous browsing and purchasing behavior, so predictions are limited to what information you’ve collected on the given user. Kumo takes this one step further by leveraging the full context of your enterprise graph, turning relationships and interactions into useful signals. Kumo can identify what a user will purchase based on what other similar users are buying. The platform also overcomes the cold-start problem by leveraging context from other users to power recommendations for new users who have no historical precedent on your platform.

These recommendations can come in the form of a curated home screen selection, direct outreach, or highly effective search experience that quickly surfaces the most relevant items for a user given their query.

Increase cross-sell and upselling

With no additional overhead, Kumo can identify for each user complementary products or services based on what they’re already looking at buying. You can create personalized product bundles with dynamic pricing Combining these recommendations with the ideal notifications strategy discussed above will drastically increase the average spend per customer and ultimately provide a superior experience.

Increase customer satisfaction (CSAT) and Net Promoter Score (NPS)

Providing a highly personalized and timeline experience (product recommendations, notifications) will significantly improve the overall user experience, contributing to a higher satisfaction score.

Another example of how customer satisfaction can be improved is to directly power proactive customer support. By delivering predictive ML rooted in your business data, Kumo can enable chatbots (or support agents) with analytics. As a result, you can offer real-time assistance, resolve queries, and provide personalized support. This proactive approach not only enhances the overall customer experience but also reduces response times and increases the chances of resolving issues before they escalate, leading to improved CSAT and NPS.

Improving Gross Merchandise Value (GMV)

By identifying the best customers, improving your ability to engage with them to move them through the funnel, accelerating cross-selling/up-selling while optimizing product bundles, and personalizing the end-to-end experience, you can effectively build a loyal customer base at scale and increasing the demand for goods and services on your platform.

How to get started

If any of the above is relevant to your business, reach out directly to try out the platform!