After someone subscribes to your email list, their email address serves as the foundation for all further data collection. As they click, open and navigate across your site we gather more data, and what we learn about them quickly compounds.
But the problem with all this information is that there is so much of it. We need a way to tell what the important and revealing actions are — adding an item to a shopping cart, or browsing within a certain category of products or services. First defining and then organizing this information necessitates what I refer to as the Semantic Layer.
The name sounds a bit abstract, but it’s really very simple: it’s simply a description given to a certain action. It’s also very simple to implement. Every good EMS allows you to add additional information about individual subscribers through the use of what are often called “Tags” or sometimes “Custom Fields”. These are short descriptive labels that can be attached to an email address in your database. It’s a very simple, but very powerful feature that gives us a lot of control over how we manage things.
The Semantic Layer allows us to go from a single basic piece of data (an email address), into a complex profile about a person consisting of specific actions they have (or haven’t) taken. It’s in using this additional data that we can then segment and personalize the offers we send. The below image shows how just an email address can slowly be built into a high-information profile about a a customer:
Stage I: The process all starts with an email address — the new entrant to our ecosystem. When we set up the Awareness Automation in the Dispersal stage, we built an environment where this email address is less likely to unsubscribe or ignore our emails.
Stage II: As part of that Awareness Automation, every email we sent tracked when a subscriber interacted with it: opens, clicks, unsubscribes, spam complaints, etc. This actually serves our first stage of data collection: Email Interactions. Now you have an email address associated with certain actions, giving some basic insight into the interests of this individual. Already, it’s much more than just an email address.
Stage III: This is where the Semantic Layer begins. Rather than being a catalog of interactions, this new layer we’ll build describes the meaning of those interactions. In describing what a certain action represents to us, we sort our users into valuable segments. I’ll give an example below.
Stage IV: In the final stage, we also collect events from outside the EMS. Now we can not only describe the meaning of interactions with emails we send, but the behaviors of our subscribers after they click a link from one of our emails. We can follow the interactions of a subscriber back to our site to see what products or services they show interest in, and again use the Semantic Layer to describe the meanings of those actions.
The point is that over time as we collect more data, this Semantic Layer becomes a source of great insight. We begin to open up the Black Box. As Kotler might describe, we will have data revealing both the Customer Journey (via engagement with certain emails and interactions on the site) and the Customer Avatar (via which offers are engaged with and what products they express interest in).
Crucially, this data opens up new creative and targeted marketing opportunities that wouldn’t have been otherwise available. Using an eCommerce example:
Without the Semantic Layer you can’t really get these insights - this information remains hidden inside the Black Box. But more importantly, without this information, you can’t really do advanced email marketing automation at all. Because not only does the Semantic Layer provide the data necessary to make strategic decisions — it also serves as the central switchboard for your automation. It’s what allows you to trigger and end sequences, segment and filter your subscribers, and generally ensure the right person is getting the right message at the right time.
So far, this might still all seem a bit abstract. But setting up an effective Semantic Layer is actually quite simple. As mentioned, your EMS should already have all the tools you need to set it up through “tags” or “custom fields”.
The problem is that most EMS typically provide very little instruction on how to get the most out of these features. Being simply an empty text field, they’re very flexible, which is both a good and bad thing. While it offers you complete freedom in how you use them, without a proper plan in place you can quickly create a big mess. Without a strategy, you’ll find one day your tags simply overwhelm you. Your subscribers will quickly gain a lot of different tags and you won’t remember the function of any of them.
So you’ll need a strategy for how you organize everything. Again, nature provides us with a solution.
Scientists long ago came up with a taxonomy to name, describe and classify the diversity of life. A taxonomy is simply an organized system for naming living things. But it’s very powerful: with a naming structure that shares a common pattern, scientists from across the world can understand the relationships between different organisms, while also inferring their important characteristics.
For example, from just two words, “sequoia sempervirens”, a scientist knows that the giant tree from the post on personalization is related to all other trees that also start with the first word (the genus), sequoia. Because of the way the system is organized, they can then check to see that a sequoia is part of a family of organisms called Cupressaceae, the cypresses, which in turn is part of a broader class known as Pinopsida, or conifers.
Since they have knowledge of how the system is organized, these two Latin words carry a lot of meaning for those who know how the underlying system of naming works. By using a common syntax - a structured way of using words - we build relationships and importance for the things we apply them to. Without a proper taxonomy, it would be very hard to organize the approximately 1.2 million species of which we have current written knowledge. The information would be unstructured and naming organisms in Latin would quite literally be meaningless.
This is actually the best way to manage your own Semantic Layer. Mercifully, we don’t need anything on the scale of the taxonomy of living organisms I described above. But it’s pretty easy to come up with a taxonomy of your own. All you need to do is create some labels to describe the sets of information that appear frequently.
Here are some examples of tags using a proper tagging taxonomy:
Something like the above is what you're aiming for. It’s kind of like the [genus]—[species] structure in sequoia sempervirens. For example, you can set it up so that every time someone subscribes to your list they get a tag with the starting syntax “LEAD”. Similarly, the word “STATUS” can be put at the front of any tag where you deal with the management of that subscriber — unengaged, highly engaged, solution aware, etc.
To recap: To improve our conversions we need to focus on personalization: and personalizing is really just about improving the relevance and timing of a given email.
The only way to do that is by understanding our customer. The Black Box and Kotler’s Customer Journey and Customer Avatar guide us in knowing where to look to do so. To actually collect and organize this data we must build the Semantic Layer. We can then use this to segment, personalize, trigger automations and improve conversions.
The following section will give an overview of the mechanics underlying the engine of the Recruitment stage: what I call the Semantic Automation.
With a broad understanding of Semantic Layer and the Black Box components, it’s time to cover exactly how they work together. Taken in isolation these two concepts may seem unrelated. But in this section we’ll reconcile them to show exactly how they work together.
By understanding what behaviors a subscriber takes on their way to purchase, you can remove any roadblocks that may be hindering that.
This clarity will reveal which actions are important to describe with the Semantic Layer. Over time this will lead to us learning more about our customers.
We’ll use an eCommerce store as the example that follows. The reason for this is because in eCommerce, the Customer Journey tends to follow a pretty predictable formula: while details may vary slightly, in general a visitor browses the site, chooses a product they’re interested in, adds it to their “cart”, then goes to a “checkout” page where they complete their purchase.
For this reason, it’s the most straightforward type of online business to use as an example. This doesn’t mean you can’t build the sequences of the Recruitment stage if you’re not an eCommerce store. The broader concept applies whether you’re a blog, software company or any other type of small online business looking to get the most out of email marketing and automation. You just need to make sure you have an understanding of your Customer Journey before you build the following sequences.
You may already be familiar with the idea of “Cart Abandonment”. Put simply, an abandonment is a type of automation used in eCommerce that tracks when a visitor to a site has clicked the “Add to Cart” button for a product but has not followed through on their purchase. This sequence will then automatically trigger, sending one or more follow up emails that intercept the abandoning customer and save the conversion.
Cart Abandonment sequences are famously effective at improving conversion rates — but it’s helpful to understand just how effective.
The amount of revenue loss that occurs due to customers adding items to carts and not completing the purchase is massive. eCommerce data science company Barilliance reported the average rate of Cart Abandonment in 2019 was ~77%. This means more than three quarters of potential online retail sales in 2019 were lost due to customers deciding not to purchase, even after signaling strong interest in a product.
With Statista.com showing total 2019 online retail sales to be valued at roughly 3.5 billion US dollars, this means something in the range of 5-10 billion dollars of hypothetical online retail sales were lost that year — all of it added to an online cart (a strong signal of purchase intent if there ever was one) but never actually followed through.
While those are large figures in aggregate, the amount of potential lost sales for individual store owners is just as huge. Assuming a similar average Cart Abandonment rate of ~75%, an eCommerce store with a yearly turnover of $1,000,000 would be missing out on around $3,000,000 of potential revenue just from abandoned carts alone.
So the numbers make it clear there’s a large amount of revenue loss that can be avoided. And the way to avoid these losses has proven again and again to be via these email marketing abandonment sequences.
In fact, the effectiveness of Cart Abandonment email sequences is one of the most well-studied aspects of online retail and the results it yields are relatively consistent. Barilliance states in their 2020 email marketing statistics report that the average conversion from a Cart Abandonment email was 18.6%. This means that, on average, one in every five Cart Abandonment emails you send will result in a conversion.
When applied to the above example of a store with a turnover of $1,000,000 and an abandonment-attributed loss of $3,000,000, this would represent $558,000 of recaptured additional yearly revenue from a single automated sequence.
The losses recaptured with these sequences are impressive and should be considered for that reason alone. But the more important and underappreciated aspect is that they can also help reveal our Customer Journey.
The “Cart Abandonment” you are most likely familiar with is just one form of this sequence. It’s actually part of a broader category of sequences, which I’ll here refer to collectively as Abandonment. In fact, there are actually four main types of abandonment sequences:
Together these four sequences will track the Customer Journey of a new subscriber — from exploring types of products (Category Abandonment), to selecting a specific product (Product Abandonment), to expressing discrete product interest (Cart Abandonment) all the way to final purchase objections (Checkout Abandonment).
Together, these sequences collect the data that allow you to personalize your emails. They are like cameras placed to take a snapshot at the parts of your store where the most critical actions take place. A camera at the category level, a camera at each product, and at the cart and checkout. You can then combine these individual snapshots with the engagement in your emails to take your Semantic Layer to a whole new level.
It’s only by combining all four types of this sequence that we can get the data from all the stages of the Customer Journey, illuminate the Customer Avatar, and finally send targeted and personalized offers to our subscribers.
These abandonment sequences are broken down into two broad categories, Purchase and Browse Abandonment:
The first sub-category of abandonment is concerned with retrieving lost sales. Let’s look at these two sequences and how they work together to bridge the gap between different parts of the purchase process.
a. Cart Abandonment
The Cart Abandonment sequence is triggered when a prospect has viewed a product and added it to their cart but has left the site before they completed their purchase.
For many store owners, this sequence is their first taste of “marketing automation”. It’s also the most highly tested form of abandonment we’ll look at here, with the stats I referred to above all referring specifically to this variant. The ROI is great, they’re increasingly easy to set up, and they just about guarantee conversion increases for stores with moderate to high sales volume.
To quickly recap, a typical Cart Abandonment sequence looks something like:
b. Checkout Abandonment
There is often confusion between Cart Abandonment and this second type, Checkout Abandonment. I often hear store owners referring to each of these interchangeably, which is an understandable confusion. The differences are subtle, but important.
While the former has shown interest in a product by adding it to their cart to be purchased, someone abandoning the checkout has added the product to their cart, and actually gone all the way towards making payment on that item, then stopped.
The difference between the two sounds small but there’s a huge gap between flippantly adding something to your basket and actually taking steps to transact for that item.
Imagine going to the grocery store and picking something up from the shelf then realizing you don’t need it and putting it back. That’s similar to Cart Abandonment. Now imagine taking a cart full of items to the register, then walking away from your trolley when the cashier asks how you wish to pay. This is the equivalent of Checkout Abandonment.
Thinking about it in this way, the difference between purchase intention at either stage should be a bit clearer. The first example had a relatively weak intention in comparison to the second.
To illustrate further, Checkout Abandonment typically goes something like this:
The difference between Cart and Checkout Abandonment can be seen in Steps 3-6 above. Checkout Abandonment can provide valuable data about your audience. You must uncover why they didn’t go all the way to purchase: was the card declined? Did they realize they can’t afford it? Thought better at the last minute? A link on the site enticed them away from completing the purchase? Solving these issues will invariably lead to better conversions and more revenue in the long term.
While the above two forms of abandonment are concerned with improving conversions for those with an intention to purchase, the next two are slightly further from the sale. While the Cart and Checkout Abandonment sequences were concerned with retrieving lost sales, this second category of abandonment instead focuses on collecting the data about what a visitor is interested in so it can be used to personalize emails.
c. Product Abandonment
Product Abandonment tracks what products a site visitor views. Using this information, more targeted offers can be sent to those subscribers classed as Product Aware from the Dispersal stage. The strategy is to send increasingly targeted and personalized offers, thus improving the likelihood of conversion.
As opposed to a Cart or Checkout Abandonment sequence, a Product Abandonment plays out more like this:
d. Category Abandonment
Zooming out further to our final type is Category Abandonment. Like Product Abandonment explored above, this sequence is also concerned with gathering visitor data to allow for more personalized offers. The mechanisms underlying both Category and Product Abandonment are very similar. However, it’s important both are used in order to get the best snapshot of how visitors are interacting with your store.
The Category Abandonment sequence often looks something like this:
The biggest objection I usually hear from store owners is that setting up so many Abandonment sequences will surely annoy their subscribers. If every interaction on the site triggers a sequence of emails following them up, surely this will be overkill?
Yes, it would. The main purpose of setting up these sequences isn’t to bombard your subscribers with emails every time they interact with your site. Notice how in the Product and Category Abandonment stages in particular, a follow up sequence is not triggered at the last step? This is an important detail.
Not all of these sequences need to be necessarily attached to email follow ups. The Cart Abandonment is of course going to send follow up emails after it’s triggered. But the others may execute without your subscribers ever even knowing. Instead, the Browse Abandonment sequences fire and run silently in the background, helping keep customer data up to date and informing the rest of your automation system what should be sent to this subscriber in the future.
As we’ve covered, the Semantic Automation for eCommerce is composed of four automations: Category Abandonment, Product Abandonment, Cart Abandonment and Checkout Abandonment. Now we’ll look at how to build each of these inside your EMS.
Just as with the implementation section from the Dispersal stage, the exact methods for how to build these sequences will vary greatly. It depends largely on which EMS you choose to use. For example, setting up the sequences in ActiveCampaign is different to doing so with Mailchimp or Klaviyo. While the individual steps themselves may be different, the underlying mechanics for how these sequences work are all the same.
Just as with the Dispersal stage, it’s the strategy that’s important, not the minor details. What I want to get across is a high-level understanding of the mechanics behind each of these sequences. By understanding what’s happening and why, you’ll be better informed to set these up for yourself, no matter which platform you ultimately go with. Let’s look at how to actually build each one in more detail.
As the value of these sequences becomes increasingly apparent, users have begun to expect this as a basic function and increasingly as a non-negotiable. EMS developers have responded in kind, and the most popular solutions now have these sequences built in. Even free solutions with relatively basic features such as Mailchimp now have native support for setting up Cart Abandonment, with integrations to match a wide variety of different eCommerce content management systems. The result of this being a core feature is that it is usually trivially easy to set up, sometimes with guided templates and instructions for how to do so.
So the first step is to either select the option equivalent to “Create abandoned cart sequence”, or saving that, simply creating a new blank automation template where you’ll build your own from scratch.
If you’re building your own, the first thing you need to do is select the trigger that executes the automation. In the case of Cart Abandonment, this trigger will be the following conditions set as true in your EMS automation builder:
In practice, you probably won’t even need to define those variables yourself. You can just use the “out of the box” solution. Once you’ve set up Site Tracking, most EMS have a pre-built option for an abandoned cart start trigger. Selecting the “user abandoned cart” (or similar) starting trigger for your new automation is probably the best option if it’s available.
Now you’ve set up site tracking and you’ve selected a starting trigger. The following is a template for how to design an effective Cart Abandonment email follow up sequence. Again, your EMS may already have a template they recommend you use and doing so is at your discretion. But in general, a typical sequence of this type follows the formula of “ROI” – Reminder, Objection, Incentive:
Sometimes marketers begin the Cart Abandonment sequence with an incentive, but it’s unnecessary. Wait until you have first been reminded and then handle any remaining objections before offering a discount or free shipping. Many times, the initial simple reminder will be enough to get the customer to convert.
When implemented, an idealized Cart Abandonment sequence has steps that look something like this:
The next automation we’ll build is Checkout Abandonment. As already mentioned, this is often confused with Cart Abandonment. However it’s worth looking at what conditions actually comprise the triggers for your default “Abandonment” sequence inside your EMS. So while many EMS provide “Cart Abandonment” out of the box, sometimes the conditions that trigger this sequence are actually the conditions that should be used to trigger an abandoned checkout.
The result is many store owners build what they think is a Cart Abandonment sequence but turns out to actually be Checkout Abandonment. I must stress that the gulf that exists between the two is vast. Setting up both of these is essential to improve conversion rates. Not only do you miss out on a full picture of the customer journey, but the number of visitors abandoning their carts is far higher than those abandoning at checkout stage. So by only setting up Checkout Abandonment, store owners are unknowingly missing out on a huge number of potential sales.
In contrast to the starting conditions of Cart Abandonment, Checkout Abandonment begins with a trigger more like “user has visited page: (checkout url)”. This is an important distinction because while the trigger for Cart Abandonment was an item being added to the cart, this doesn’t matter so much for this sequence. They can’t access the checkout unless they have an item already in their cart to begin with.
In any case, setting up Checkout Abandonment is relatively straightforward. The first step is to create a new blank automation template. With the automation builder open, the first thing you need to do is select a trigger that executes the automation. In the case of Checkout Abandonment, this trigger will be the following condition as true:
After setting up the new automation and the starting trigger, the next step is to design the steps of the automation itself. We’ll just reuse the same emails we used in the Cart Abandonment sequence. The goal of the two sequences is the same - catching a lost sale and converting it. The distinction is that they catch the abandonment at different points of the customer journey.
So what we’ll do is simply add a condition to make sure subscribers aren’t being sent the same emails twice. We’ll make it so if someone adds to cart and abandons, they’ll get sent the Cart Abandonment sequence, but if they don’t abandon there and instead make it all the way to Checkout Abandonment, it will trigger at that point.
So again, a typical Checkout Abandonment sequence follows the formula of “ROI”, or, Reminder, Objection, Incentive. When implemented, an idealized Cart Abandonment sequence has steps that look something like this:
The final two automations are similar with regards to setup.
Again, some EMS will provide a “Browse Abandonment” feature similar to a default Cart Abandonment. This can sometimes cause confusion as the exact trigger is unknown - is it Product or Category Abandonment? It can be hard to tell. A lot of the time, this trigger is something like “Viewed product”, and is rarely anything to do with the category level of the browsing stage of the buyer’s journey. So the best way to be sure is to set this up manually.
As with the other automations, the first place to start is creating a new blank automation. From there, you’ll add your starting trigger.
For Product Abandonment, these triggers can be:
For Category Abandonment, these triggers can be:
You might be wondering what emails you should send those subscribers who meet Category or Product-level abandonment triggers? Surely sending email follow ups to these subscribers will be a bit too much? I’d agree with that observation. Sending too many follow up emails from browsing activity, rather than improving conversions, can of course come across as creepy.
While we definitely want to use the data gathered from the browse abandonment sequences to inform our offers for that subscriber, we don’t want to overtly send follow ups as with the Cart and Checkout Abandonment sequences. Instead, these two sequences will be used to help build the Semantic Layer. Once data about browsing activity has been sent back to the EMS and the behaviors satisfy the above triggers, the sequence will simply apply a tag which progressively builds up a profile about that subscriber.
Let’s look at how we use the data gathered from the browse abandonment sequences to build highly personalized emails that improve conversions.
Now that the sequences are set up, how do we actually capitalize on this data we’re collecting? How do we personalize emails to improve conversions? Broadly, there are two main strategies for personalizing product-aware emails.
The most straightforward way to personalize the customer journey is by building multiple Product Aware sequences for individual categories or products.
For example, Let’s say Subscriber X visits your site from an email you sent. While on the site, the subscriber explores the “Shirts” category, and spends significant time on a specific product, “White Shirt”.
If you have the abandonment sequences set up correctly, the above behavior will be tracked and sent back to your EMS. After a while you may discover a large number of subscribers are being tagged “INTEREST: White Shirt”. So you can then confidently create a dedicated Product Aware sequence of 3-5 emails for White Shirts.
When that tag is added to the subscriber, it also triggers the sequence. This personalizes the customer experience using the Semantic Layer. By using the tags from the Browse Abandonment sequences, we trigger these more targeted sequences which show more relevant and timely information, and thus have a higher likelihood of conversion.
A couple things to note about these sequences:
Another way the Semantic Automation is useful is for sending once-off campaigns. This is important for when you create sales, promotions or discounts. Instead of sending a discount on a specific product to your entire list, you can send it only to those subscribers who have expressed interest in that product or category.
The reason you may want to do this is because it will limit the amount of irrelevant offers you send to your list, while also maximizing the conversions you receive from each campaign you run. You’ll save those who aren’t interested from being exposed to “spam”, but for those who actually are interested it will again appear like more value.
To do this, simply use the segmenting features inside your EMS to only send campaigns to users with a certain tag.
What’s next? Now you’ve built a system for recruiting new subscribers into your email database environment, and you’ve enabled the ideal conditions for them to establish themselves as new customers.
Now that you’ve built the foundations for an environment where buying behavior flourishes, the next step is to enhance the conditions for further growth and enable those who have already purchased to do so again (and again).
In a future post we'll take a look at how nature again has lessons for us in how we can find extremely profitable customer segments in our list, and a series of strategies we can use to make every conversion as profitable as possible.
"There's no clearer guide to getting maximum results and impact from email.
This book will change the way you think about email marketing automation in your business"
— Nir Eyal, author of Hooked: How to Build Habit-Forming Products and Indistractable