The 2019 Digital Marketer’s Complete Guide to Ecommerce Personalization

Despite many consumers wanting and expecting more relevant and personalized experiences, a recent study by Pure360 suggests that most brands are still relying on only basic forms of ecommerce personalization.

According to a Deloitte paper, this is creating a gap between customer expectations and the experiences brands are delivering.

Part of the issue is that personalization is hard. The messages too often spoken about personalization are simplistic or catastrophic. It’s either something dramatic and urgent like, “if you’re not doing one to one personalization, your competitors will zoom past you.” Or it’s too simplistic, like “personalize post-click landing pages with company names.”

These types of messages, however, focus on the tip of the iceberg rather than the process you need to develop to make personalization work. The focus should be on strategy, and how personalization can be a tool used to improve the user experience and lift business metrics.

What is ecommerce personalization?

Ecommerce personalization is the art and science of delivering specific shopping experiences to targeted visitor subgroups to increase conversions and revenue, as well as improve the user experience:

It’s like any form of personalization. You have to have a few things in place:

1. An ability to collect data on visitors (behavioral, transactional, demographic, etc.)
2. An ability to analyze that data to find segments that may respond more favorably than the aggregate pool of visitors to a specific experience
3. An ability to deliver an experience to that segment in real time

While it’s mostly used, at this point, by ecommerce marketers, others such as engineers, product managers, customer success professionals, and sales professionals can also use personalization to increase ROI.

Walk before you crawl: Strategic planning

Before we dive into inspiring examples and actionable tips, it’s important to preface the whole “personalization” thing with a caveat: it’s no silver bullet.

There are countless articles that recommend personalizing emails with the recipient’s name or pushing a different CTA to some segment just because you can. While these tactics could be useful, looking at personalization this way is missing the forest for the trees.

Rather, you should view it as a tool that extends from your optimization process. Sometimes it’s optimal to have a universal experience for all visitors. Sometimes you find an exploitable user segment to target personalized experiences to. But as Andrew Anderson suggests, you shouldn’t just personalize experiences because you can.

Here’s why…

Think of each action you take as a cost, even if it’s only an opportunity cost, and with each action you have an expected value of ROI. Action for action’s sake drives up the cost of your program with no real eye on return, which is a real strategic dilemma.

Framing personalization as a tradeoff between costs and rewards forces you to consider both the feasibility of an experience (can it be delivered effectively?) and the potential impact (if it is indeed successful, what is the upside?). When you arrive at those answers, it also forces you to ask, “is this the best use of our resources or is there a more valuable experiment to run?”

Andrew Anderson asks, “Are you just serving up an experience because you can? Or have you done the active acquisition of knowledge that shows not only that it improves performance, but that it is the best way to increase performance.”

In addition to the ROI considerations, with each additional personalization rule you set up, you incur a marginal level of organizational complexity. Meaning, each additional experience you deliver is an experience you must manage as well.

Matt Gershoff, CEO of Conductrics, puts it well:

While targeting can be incredibly valuable, many in the industry haven’t fully grasped that targeting ALWAYS leads to greater organizational complexity, and that greater complexity means greater costs.

“Complexity,” he says, “is the flipside to targeting.”

So with that in mind, you have two challenges to care for:

1. How do you choose to deliver experiences?
2. How do you manage those experiences operationally?

How to choose to deliver ecommerce personalization

There are two primary methods to discovering segments for personalization:

1. Business rules
2. Machine learning rules

In the first case, an analyst or marketer makes a decision whether or not to set up a targeting rule. This decision can be made by various means. Sometimes, it’s as simple as “I think we should give mobile users a different form than desktop users.” Sometimes, you can see that a certain segment under or over performed in an A/B test. Sometimes it’s a combination of data (qualitative and quantitative) and intuition.

The second case relies on machine algorithms to predictively surface exploitable segments. In this case, a software solution will track, analyze, and suggest segments that seem to behave differently than others. In one such example, Conductrics, an experimentation and personalization platform, can show you, based on experiments you’re running, if there are any device/behavioral/demographic segments worth looking into:

There’s no right or wrong answers as to which method you use, though it’s likely you’ll need to start with business logic, simply due to the expense incurred via machine learning software or development. Often there will be low hanging fruit in the beginning stages as well, much of which can be picked off with simple decision rules.

To dive deeper on this, Andrew Anderson has an excellent methodology for finding good ecommerce personalization opportunities.

How to manage ecommerce personalization experiences

At the simplest level, each time you set up a targeting rule, you incur a cost (design, product development, or just the time required to set it up). The benefits are usually incremental, so at the very core, you need to make sure there is ROI on each rule you set up.

In addition, managing all the experiences you set up has a complexity cost, because it makes it more difficult to run future experiments and sometimes your website can have unplanned mismatches. You also have to track down specific experiences in the case of customer support tickets.

Finally, make sure your personalization rules continue to perform well and generate ROI. Most of the time, you can use similar methods that people use to assure A/B tests are still valid over time. You can:

• Retest the experience
• Keep a holdout set that always gets the control experience

This way, you can keep track of the perishability of an experience over the long run.

4 Ecommerce customer journey points to personalize

While there are many possible pathways to personalization, some are more common or more easily implementable in ecommerce personalization. In my opinion, anyone can and should look into the following areas:

1. PPC post-click landing pages
2. Product recommendations
3. On-site promotions
4. Programmatic email marketing

1. PPC post-click landing pages

When you create true PPC-driven ecommerce post-click landing pages, directing highly-targeted traffic at a specific page designed with one offer (and no distractions like navigation), you have a lot of potential for personalization.

You can look at the pathway from the pre-click experience (the advertisement) to the post-click landing page (the post-click landing page) almost like a separate universe — similar but separate from your general website experience.

When viewed this way, almost all PPC post-click landing pages should be personalized to some extent.

At the most basic level, you should be matching the imagery and copy (aka advertising scent) from your ad into the post-click landing page experience. Doing this helps create a cohesive user experience without having the broader context of the rest of your website.

As an example, look at this search ad from HootSuite:

With its corresponding post-click landing page:

This is excellent for many reasons:

• The headline is clear and relevant (the keyword phrase is included)
• The copy is broken down into bulleted points with helpful imagery
• CTA buttons use contrasting colors
• Customer badges further down the page provide much needed social proof

But saliently, the ad matches the post-click landing page experience. You don’t need to navigate the rest of Hootsuite’s site to figure out what’s going on because the ad-to-page experience says it all.

As an example, check out this Facebook ad from Four Sigmatic:

I’ve purchased with them before, so their ad is all about getting me to become a repeat customer, particularly by using a winter sale promotion. As such, the ad copy is all about savings, and we can see there is strong message match once you click through:

PPC post-click landing pages are an ideal form of personalization since you’re targeting a subset of users based on a group of common characteristics. (In this case, they all use the same search query so you’re targeting based on URL parameters).

When you begin to build PPC post-click landing pages at scale, you can use ecommerce personalization software like Mutiny to programmatically change the post-click landing page copy and design based on visitor characteristics, such as company size, keyword targeting, or technology stack.

Since marketers have the ability to hyper-target their ads and high control over traffic sourcing with paid ads, you have a world of personalization options open. There are a ton of great examples of this in action.

Any visitor characteristic you can learn before or during their visit is something you can use to create more relevant experiences (which translates to better quality scores, lower CPC, and higher conversions).

2. Product recommendations

Amazon has really set the tone for a lot of ecommerce best practices. From their ease of purchase (one click) to their shopping efficiency (Prime shipping), it’s a pretty good customer experience. They’ve also pushed the envelope on personalization.

While many skeptics still poke fun at Amazon when they deliver irrelevant product recommendations, you have to admit, most of the time they’re accurate. (Note: I’ve bought related books as well as Converse All-Stars, sandals, watches, and wallets in the past on Amazon. This is pretty well catered to my interests):

Most product recommendation modules are built on machine learning models that analyze past shopping behavior to cluster related products to what you’ve already viewed or purchased. Then they recommend things that you might also like.

While there are many SaaS products available, building product recommendation engines is also something you can do yourself, as long as you have an understanding of clustering algorithms and the programming capabilities to design them. Here’s a great walkthrough from Analytics Pros.

Also note that you don’t need to limit yourself to product recommendations based on past behavioral data. Think at a more meta-level: what type of recommendations would a user even appreciate?

Sure, some users would jump at the chance to view related products, but others might be interested in hearing about new offers. Some might just want to see what is most popular. Some might want to view the cheapest deals this month.

Of course, this is harder to predict, but that’s where a good data scientist brings value. If you can personalize product recommendations based on visitor characteristics, you’re certainly going to lift your conversion rates and be successful. Here’s an example of showcasing trending or popular products to a subset of visitors:

In addition to thinking about which product recommendation modules will work best, think about where they’ll give the biggest impact. One obvious place is the homepage, where you have a diverse set of visitors and a large part of them are browsing vaguely. This is a good place to surface targeted products or best sellers (or new offers).

After someone adds an item to their cart is a great time to show a “related products” module, which, in this case, acts as an upsell/cross-sell mechanism. You could even put a “related products” module on a product page, like Amazon does here:

3. On-site behaviors

When we think of ecommerce personalization, we often think of the simple things, like using UTM tags to personalize post-click landing page copy or using someone’s first name in an email newsletter. Product recommendations are becoming more popular, mostly due to a proliferation of SaaS tools that enable them (plus the inspiration of Amazon).

However, something underrated is altered experiences based on onsite navigation and behavior. Basically, you’re using things like mouse movements, pages visited, and a visitor’s progression through multi-step forms to deliver key messages and onsite targeting.

This is somewhat of a vague category of personalization, so I’ll walk through a few examples.

First, if you have a free shipping threshold, you can remind visitors with a message that indicates how much remains until they reach it. Or if they’ve reached it, you can give them a congratulations message. Example from Candle Delirium’s order process:

An easy one to implement is to personalize offers based on new versus returning visitors, which both have distinct shopping behaviors and needs. Many ecommerce retailers will offer some sort of discount to hook first time visitors into signing up for their email list:

Of course, many behavioral methods exist that marketers use to capture email leads, from targeting first time visitors to scroll triggered popups, exit intent popups, and more.

LawnStarter uses social proof to show visitors how many people have signed up for a given lawn care services within the last 24 hours:

One new way of delivering interesting and personalized messages is with live chat software or a chatbot. Most companies are still using the same chat message for all guests on their website, but with most tools, you can easily personalize the introduction message based on the URL, customer behavior, or other data you can collect from visitors:

You can also look into enterprise software like Granify that seeks to detect visitor behavior and use it to predict intent and various actions. For example, through a variety of touchpoints they claim to detect when a user is, for instance, paralyzed by choice, or is price sensitive. When they learn this, they can deliver modals or discounts or other interface changes to attempt to persuade the visitor to purchase.

4. Programmatic email marketing

One of the most common use cases for ecommerce personalization is another one that few people consider “personalization” — abandoned cart emails.

When you have someone’s email address, you can remind them that they haven’t finished purchasing yet. Often, this is super low hanging fruit and can give you an immediate ROI in your optimization efforts:

You don’t even need to keep behavioral emails exclusive to abandoned cart triggers. Here’s an email from Society6 that shows several items that I had just viewed but didn’t purchase:

Remember our chat about product recommendations? You can send those out via email as well, especially in post-sales emails:

Whereas most email marketers might stop at personalizing with your first name (“Hi {{Name}},), there’s a world of targeting you can do when you incorporate behavioral signals, contact property data, and firmographic data if you’re in B2B. In fact, the future of email marketing will likely be dominated by better-targeted personalization.

The thing with these types of emails is that they are clearer on the intent of the recipient. An email blast, by its nature, reaches a diverse audience with a variety of tastes and interests.

Programmatically personalizing emails based on behavioral characteristics can lead to much greater relevance, a better experience for your subscribers, and more revenue for you.

The top ecommerce marketers are working on campaigns like this, but this is table stakes in the B2B SaaS world, where it operates under the moniker of “marketing automation.” We build systems that attempt to deliver the right message to the right person at the right time.

Luckily, technology makes it increasingly easier to do this. We now have customer data platforms to centralize our data and easy ways to connect our marketing tools using solutions like Zapier. We can then operationalize that data using whatever messaging tool is available, such as Klaviyo or HubSpot’s email marketing service.

Conclusion

The examples here demonstrate what is currently possible with ecommerce personalization. Your only limits are your resources at hand and your creativity.

That said, look at personalization as an extension or a tool in your optimization toolkit. It’s not some silver bullet, and you won’t generate millions by putting someone’s first name on an email or a post-click landing page. Personalization should reflect an optimal user experience; thinking of it only in terms of content personalization will never bring meaningful returns at scale. Get more details in the Instapage personalization guide.