/ Personalization

How Metisa personalization works (and how to make use of it)

Metisa does many things behind the scenes to make intelligent marketing decisions based on insights derived from data. But we made Metisa exactly to relieve you of having to deal with complex data science yourself. Our goal with Metisa is to make it easy for you to have the power of a data science team without investing in one.

If you'd like to dig deeper to understand how Metisa works, even if it's just for assurance that everything is going okay, please read on. Below we will go into detail about what data Metisa uses to discover insights, the data science behind it, and the main features you must know to fully leverage its power to drive sales.

An engine that learns

Running an e-commerce company means you are probably sitting on a treasure trove of data. Metisa analyzes data about your customers (name, gender), products (name, type, price, description), and purchases (what was bought, for how much, when) to discover valuable insights for your store.

This analysis starts when data is fed into the brain of Metisa, an analytics engine with machine learning abilities. (This data fitting process happens automatically with our Shopify and Magento integrations.) As Metisa makes sense of your data, it constructs a unique model of your store's "reality". This representation, known as a statistical model, is the wellspring of actionable insights.

Every night, Metisa re-runs its analysis (with the latest data) and reconstructs its model of your store. But here's the really powerful part - because of machine learning, Metisa is able to refine its model of your store as new data is added (some of which confirm and others refute certain hypotheses). This allows it to give you ever more precise insights on your store.

Insights to inform your marketing decisions

Insights can be accessed on the ‘Insights’ tab of your Metisa dashboard, helping you make better decisions on your marketing campaigns. You can use insights like the combined Customer Lifetime Values (CLV) by Acquisition Channel to confidently spend more marketing dollars on social or email marketing, for example.

All of the insights you see in the Insights section of your dashboard are statistical calculations. For example, to arrive at a Predictive CLV, we calculate several probability distributions, find the best fitting one to represent the churn model, and then feed it with each customer's unique information (age, gender, last purchase).

Metisa can actually do much more than discovering insights and visualize them for your use in informing marketing decisions. It also comes equipped with virtual limbs that can act on your behalf as you instruct.

Data-driven product recommendations

The insights can be used to engage with your customers intelligently and automatically. One way this is done by Metisa is through product recommendations.

Metisa's recommendation engine--similar to those used by international retailers like Amazon--makes intelligent predictions about what customers will like based on what they've loved.

Specifically, the recommendation engine is based on a customer-product matrix. Every product has a score, and customers belonging to a group will share a unique set of product scores. In general, products with the highest scores that a customer has not bought will be recommended to her. However, since an organic food store customer is much more likely to buy a product more than once than a fashion boutique customer, the engine also takes into account the nature of your store based on the products you stock.

With a distinct set of products ready to be recommended, personalized to every single one of your customers, Metisa can then put these products in front of them using Recommendation Widgets.

Recommendation widgets are designed in a customer-centric manner, to deliver recommendations in an unobtrusive and welcome way.

Based on your needs, you can choose to embed Recommendation Widgets on your site and/or in emails. Embedding involves anything between clicking a toggle button and copy-pasting a couple of lines of code into your site. Metisa takes care of the rest, like what to recommend, where to get the images to show, and getting the underlying JavaScript code to work.

Next, let's talk about what you can do with Emails in Metisa.

The right message for every customer segment

Because Metisa intelligently clusters your customers into segments based on their purchasing pattern (name and category of products bought, frequency), you can send emails to each segment. This, combined with the ability to embed recommendations within emails from the Metisa dashboard, makes it easy for you to do email marketing and makes it frictionless for your customers to buy more of the products they love.

Segmentation is achieved through the K-means statistical algorithm. Customers are grouped into K number of clusters based on their attributes like purchasing pattern, age, and gender. On top of smart segmentation, you can also create your own customized segments based on any insights you've arrived at independently, and send email campaigns to them from the dashboard.

Email campaigns sent are also automatically tracked by Metisa. Data like open rate and URL click rates are captured and can be accessed in the Report view under specific One-off Campaigns.

Now let's talk about one last important feature of Metisa - automated emails with product recommendations sent at precisely the right time.

Timely, automated emails

We understand that Email is still a strong channel for marketers; that's why we designed Metisa with lifecycle marketing principles baked into its core.

Each customer is understood as being at a distinct stage in his/her life cycle with your brand. The flow from start to exit is: Member -> One-time -> Active -> At-risk -> Lost.

Metisa stays on top of these 5 lifecycle buckets for you and can be instructed to act on your behalf (eg. send an email to win-back an At-risk customer) whenever a customer jumps buckets. Every night, your customers are reclassified into the various buckets, and Metisa keeps an eye out for any movement for you.

For example, when a customer makes her first purchase on your store (Member -> One-time), you can instruct Metisa to send an email with a welcome discount using an email template designed specifically for this scenario.

Similarly, you can setup Metisa to attentively spot customers who are slipping away from your brand (Active -> At-risk) and automatically try to win them back with a well-timed email.

With lifecycle marketing built into the core of Metisa, you can rest assured that you're sending the right message (and promotions and recommendations) to each customer at precisely the right time.

Let Metisa work for you

With Metisa, you can put to rest that nagging feeling that your data represents a heap of wasted opportunity.

Knowing how and when to engage your customers can mean the difference between losing customers and building a community of active customers. Metisa gives you insights powered by your data to keep your customers engaged and buying.

Many great features are free on Metisa, so you only need to upgrade when you need premium features. Start putting Metisa to use for your e-commerce store today! Start here.

Justin Yek

Justin Yek

Partner & cofounder at Altitude Labs, creator of Metisa, former investment banker, public speaker, hobbyist musician

Read More