The Stages of Personalization

Our Stages of Personalization eBook shows you the path to delivering a more personalized experience for your customers.


The Stages of Personalization

Our Stages of Personalization eBook shows you the path to delivering a more personalized experience for your customers.

Personalized messaging is the backbone behind some of the most successful marketing campaigns of the modern era. Cadbury’s chocolate matching campaign. Coca Cola’s “Share a Coke” packaging in which the company replaced “Coca Cola” from one side of the bottle with “Share a Coke with {Person’s Name}.” Starbucks’ entire mobile app.

Although thought leaders in the marketing industry have been preaching the benefits of personalization for years, there’s been very little in the way of hard data around how effective personalization can be -- and as a result, many mobile marketers don’t know what they’re leaving on the table by not doing it well.

Taking this into account, we decided to analyze 23 billion push notifications sent by Localytics customers during 2018 and let the data tell us exactly what personalization can do for marketing campaigns. Using a K-Means Analysis (also called K-Means Clustering), we uncovered the natural progression of content personalization. Our work revealed that you can put a brand’s marketing into one of four progressively more sophisticated groups based on what methods the brand uses to personalize their messages. We’re calling these four groups the Stages of Personalization. We also discovered that the more sophisticated the Stage, the better the average message performance.

Most importantly, we found that moving from one Stage to the next can have a massive impact on average message performance. For example, the average message from brands in Stage 3 had a click-through conversion rate that was 68% higher than the average click-through conversion rate for a message from brands in Stage 2. We saw similar improvements across almost all other measurements of message performance, from open rate to subsequent app sessions.

Based on these findings, it appears as though marketers’ frequent assumption that it’s too time consuming to deliver a personalized experience causes them to miss out. We find that personalization actually offers a strong competitive advantage. These findings also align with consumer surveys: Per Salesforce’s State of the Connected Customer, 57% of customers are willing to provide their data in exchange for a more personalized experience. The problem is that only 22% of customers are satisfied with the level of personalization that brands are delivering, per Segment’s 2017 State of Personalization Report.

If your brand delivers a strong personalized experience, it will stand out to these unsatisfied customers. The first step in improving is understanding the personalization equation.


The Personalization Formula

The Personalization Formula

Far too much of the existing content on marketing personalization treats it as a black and white concept - it implies that a message is either personalized or it isn’t. Nothing could be further from the truth. In fact, there are several different ways you can personalize a message. To highlight this and emphasize that the more techniques you use, the more personalized the resulting experience becomes, we have created The Personalization Formula:

Audience + Message + Timing = Personalized Experience

Let’s quickly review the three parts of the formula, as you can’t start improving your brand’s personalization if you don’t know all the tools you have at your disposal!


Effective personalization is as much about what messages you don’t send as it is about what you do send. For example, if you know a customer is a fan of your jeans but you send them a message suggesting they buy a new pair 48 hours after they last purchased jeans from your store, that message is likely going to generate negative reaction from the customer.

As you can see, the Audience plays a critical role in any effort at personalization because it is where you determine who does and doesn’t receive your message. Generally speaking, most marketing automation tools give you the ability to use either Profile Data (who a customer is) or Behavioral Data (what the customer has/hasn’t done on your app or website) to define who should be included in your messaging Audience. The best platforms let you use both types of data, as this combination gives you the greatest ability to ensure the right people get your message. For example, to avoid the situation above, we would build an Audience of customers whose favorite product category is jeans (this is Profile Data) and who haven’t performed a purchase in the last 7 days (this is Behavioral Data).

Our analysis showed that messages whose Audience used both Profile and Behavioral criteria ended up having 2.2x higher conversion rates than messages whose Audience just used Profile criteria (6.0% vs. 2.7% respectively). So when you set out to improve the personalization and performance of your messages, don’t overlook the Audience!


It should probably go without saying that one very effective way to personalize your marketing is to adjust the message itself. There are several ways you can do this and we’ll review them right now.

When most marketers think of personalization, they think of being able to insert things like a recipient's name into their message. This is done with what’s known as dynamic content - content that changes for each recipient. It can be used to not only insert a recipient’s name, but also do things like mention a specific topic of interest, determine what pictures get shown, and much more. Dynamic content can be very effective - when you include it in a push notification, the average open rate jumps from 3.6% to 5.8%. However, it’s important to recognize that dynamic content alone is not enough to ensure a personalized experience. After all, going back to our example of the customer who received the promotion about jeans 48 hours after they last purchased jeans, do you think their reaction would have improved at all if we had mentioned their name in the message? Probably not.

Another often overlooked way you can better personalize message content is through A/B/n testing. Although A/B testing doesn’t allow you to change a message in real-time, it does give you the ability to learn over time what content a particular Audience prefers. By running A/B tests, you can learn what language, offers, and content an Audience likes to receive and use that understanding to craft the best messages possible on an ongoing basis.

One final way you can personalize a message is by adjusting the channel you use to deliver it. Many brands do this by asking customers to specify what channels they prefer to be contacted on. Another way to determine the right channel is based on the content itself. For example, a breaking news story should probably be sent as a push notification, where it will be seen while the news is still relevant. In contrast, a message asking for a user to confirm their contact information is up-to-date should probably be sent via an in-app or Inbox message. Sending such a low-priority message via push will probably come across as aggressive and impersonal.


Anyone who has been woken up at 3:00AM by a push notification can tell you that it is not an experience that feels personalized in any way, shape, or form. There are several options for determining when and where a user receives a message. Some of them are used to avoid sending a message at the wrong time and others are about sending the message at the perfect time. Let’s take a look.

In order to avoid having your push notification turn into a 3:00AM wake up call for an unfortunate recipient, most mobile engagement platforms offer the ability to send a message by timezone. This allows you to specify a desired send time, say 8:00PM, and the platform then sends the message to users in each timezone when the local time is 8:00PM.

When it comes to sending a message at the right time, a great way to accomplish this is with triggered messages. Triggered messages get sent either immediately after a user performs a specific action or after a specified delay. For example, an entertainment app that wants to show a thank you message to a user who just invited a friend to check out the app can build a message that gets sent when the user clicks the send button.

Another way to optimize the timing of your message is with location-based messages. By joining the digital and physical worlds, location-based messages in many ways represent the pinnacle of what’s possible when it comes to personalization. These messages get sent when a user enters or exits a specific location (for example, a retailer sending a message reminding a customer that they have unspent credit as the customer walks by one of the retailer’s stores). Of course, like all personalization, there’s a line between beneficial personalization and creepiness. Nowhere is that line finer than with location-based messages. The key to staying on the right side of that line is putting yourself in the shoes of a customer and asking yourself, “How would I feel if I received this message out of the blue?”

We can now move on to the next step: understanding the four stages of personalization. Each brand resides on an arc from “no personalization” to “highly personalized.” By understanding where you currently are, you can get a better sense of what opportunities are available for you to improve the effectiveness of your marketing efforts.


What are the four stages?

What are the four stages?

Stage 1: The Beginner Stage

Brands usually start out broadcasting their campaigns, meaning every customer receives the same message. Messages in this stage are similar to billboards, as the message is shown to every possible recipient and it isn’t personalized for the viewer in any way. Many brands like broadcast messages because they are efficient - a large number of customers can be reached with minimal effort. However, there are some significant drawbacks to broadcast messages that need to be considered. Over time, many users get turned off by broadcast messages and will disable notifications or even uninstall your app to put an end to what they can perceive to be persistent, irrelevant marketing. In addition, broadcast messages do a poor job of driving changes in user behavior, with an average click-through conversion rate of just 0.44%. In many cases, a more targeted campaign might end up reaching fewer customers, but paradoxically drive a higher number of total conversions thanks to the higher conversion rates personalized messages enjoy.


Stage 2: The Intermediate Stage

In Stage 2, there’s a dramatic change as brands almost completely abandon broadcast messages and start to take their first steps in delivering a personalized experience by using an Audience for almost every message they send. In Stage 2, the Audiences tend to only use one type of data. Some brands in Stage 2 build their Audiences using only Profile data, while other brands use only Behavioral data. We’ve mentioned these two types of data already, but let’s do a brief recap.

Profile data is a snapshot of who a user is. Think name, age, gender, etc. Some aspects to remember:

  • Captured using SDK or pulled from other systems (like a CRM)

  • Provides a snapshot of who the user is at the current moment

  • Can be harder to obtain as a user generally has to take the time to provide some of this information themselves (e.g. while creating an account)

Behavioral data is a snapshot of what a user has done inside your app or website. Think of it as the actions a user has taken: opened app, read article, upgraded subscription, purchased product. Remember:

  • Captured using SDK or Events API

  • Provides information on what a user has done in the past

  • Generally easier to collect as there’s no onus on the user to provide this information manually

Okay, with that recap out of the way, let’s get back to discussing Stage 2. As we said above, brands in Stage 2 tend to build Audiences that use either Profile data or Behavioral data, but not both. Brands at this Stage are starting to get familiar with the idea of segmenting customers, but they are not yet comfortable using data from multiple different sources to build Audiences. Brands that build Behavioral-only Audiences tend to segment users based on “events” like the number of sessions a user has had, the content the customer has viewed inside the app/website, or the last time a user opened the brand’s app. Brands that build Profile-only Audiences tend to segment users on “attributes” like customer type (free, paid subscriber, premium subscriber, etc.), customer interests (favorite category, favorite artist, etc.), or basic demographic data (age, gender, hometown, etc.).

Quick sidenote: If you are currently segmenting users using only demographic data, we strongly encourage you to look at what other Profile data you can use that might be more accurate. We have yet to see a situation where demographic data, like gender or age, alone predicts the interests of a group with any large degree of accuracy.

In addition to using Audiences, Stage 2 sees the first significant use of A/B testing. The use of A/B testing helps brands better understand what messages resonate with their Audiences, which allows them to deliver more enjoyable, relevant notifications over time. By using Audiences and A/B testing, brands in Stage 2 are taking the first big steps in delivering a personalized experience, but there’s still a lot of opportunities left for improvement.


Stage 3: The Advanced Stage

After working through profile-only and behavior-only targeting, brands can progress to Stage 3 using one of two methodologies:

  • The Combination Methodology: The majority of brands progress to Stage 3 by making their Audiences more robust. These brands move from Audiences that only use Profile or Behavioral data to Audiences that utilize both types of data. By making use of all the data that’s available to them, brands using the Combination Methodology can send far more compelling and individualized messages, thanks to the granular Audiences they have created. To understand what makes Behavioral + Profile Audiences so powerful, let’s look at some examples: A news app might promote their paid subscription without annoying casual readers by targeting the campaign only at customers that currently have a free subscription (Profile data) and who have read at least nine articles in the last week (behavioral data). Alternatively, a retailer might make sure their Audience of customers interested in clothing is as large as possible, yet still targeted, by including users who have said clothing is their favorite category in the store (Profile data) and also include users who have viewed at least three items of clothing in the app over the last week (Behavioral data). These are just two examples, but hopefully we’ve conveyed just how much more flexibility you have when it comes to ensuring the right message gets to the right person when you make use of Profile and Behavioral data in your Audiences.

  • The Profile Audience + Dynamic Content Methodology: Brands using this approach stick to using Profile-only Audiences, but make up for the limited personalization of this approach by complementing it, with the use of dynamic content, which allows you to customize the content of a push notification for each recipient. For example, a retailer might have an audience of people who have unspent store credit. Using dynamic content, the retailer could send a message that mentions the recipient’s name and the amount of store credit they currently have. As another example , a media company could send a message encouraging users to upgrade to a premium sports subscription and include a mention of their favorite team in the message. The pairing of Profile Audiences and dynamic content yields a strong personalized experience, but in general, the average performance of these messages isn’t quite as strong as the performance of messages sent by brands using the Combination Methodology. For example, compared to the average message performance in Stage 2, using the Profile Audience + Dynamic Content Methodology results in increasing conversion rates by 17%, but using the Combination Methodology results in increasing conversion rates by 31%.

No Matter which methodology is used, brands in Stage 3 continue to use A/B testing to optimize their messages and some even start to experiment with more advanced ideas like message timing, using triggers to initiate engagement, and location-based messages. However, brands in Stage 3 do not make widespread use of these capabilities. For that, we’ll have to look at Stage 4.


Stage 4: The Expert Stage

In Stage 4, brands fully embrace Profile + Behavioral Audiences as well as the use of dynamic content. In addition, brands in Stage 4 make prodigious use of triggered and location-based messaging, not to mention continued use of A/B testing. All of these factors combine to deliver a potent personalized experience that sets these brands apart from its competition. By making maximum use of all the personalization tools at their disposal, brands in Stage 4 enjoy unequaled customer loyalty and brand reputation.

While there are probably some brands in the world in Stage 4, we sense there are very few and we haven’t seen any reach this stage just yet.  This isn’t surprising given Stage 4 is a rather sophisticated stage. As we continue to work with clients, we expect to see Stage 4 customers become more common in the coming months and we’ll be excited to share additional insights as more customers progress to this Stage. Speaking of which, let’s take a look at the breakdown of where our customers fall along the Stages of Personalization.


How far along are brands right now in these stages?

How far along are brands right now in these stages?

As you can see from the graphic below, the majority of customers reside in Stages 2 and 3. We believe this distribution is largely reflective of the wider market, which has a slightly larger proportion of brands in Stage 1 and less than five percent of brands in Stage 4.


As you can see, the largest chunk of customers reside in Stage 2, and none have reached the mastery of Stage 4 yet.

The primary takeaway goal for this eBook is helping your brand understand where you reside now and how to get to a higher stage of personalization. Before we do that, though, let’s consider the incentive to progress: What happens when brands are more effective at personalization? What are the results they can expect?


Personalization is going to help your engagement (and your business!)

Personalization is going to help your engagement (and your business!)

As we mentioned in the introduction, moving from one stage to the next can have a massive impact on average message performance.


The chart above shows how the average message performance changes between the various stages. As you can see, there’s a substantial increase in all metrics between Stages 1 and 2 and the metrics increase even more between Stages 2 and 3. This tells us that moving to a higher Stage can result in more effective and compelling messages, but why?

To find the answer, we need to look at how the average performance of an unpersonalized message compares to the average performance of a highly personalized message. In this case, we looked at the average performance of messages that didn’t contain any dynamic content and were sent to all users (a broadcast audience) and compared it to the average performance of messages that contained dynamic content and were sent to a Behavioral + Profile Audience. These were the results:


As you can see, the highly personalized messages drastically outperform the unpersonalized messages. This explains why the average message performance for each subsequent Stage is better than the one that preceded it - as your messaging becomes more personalized, the average message performance improves as well.


How to move towards these results

How to move towards these results

Start by focusing on the Audience

Depending on which stage you’re currently in, focusing on the Audience might mean building your first Audiences, while for others, it might mean starting to combine profile data and behavioral data in your Audiences. No matter where you are today, your long-term objective should be to create these Profile + Behavioral Audiences, as they will help you obtain three distinct benefits:

  • You can increase audience size without diluting the relevance of the message: Consider a travel app that wants to engage customers interested in New York City. They can build an Audience that includes users who performed event “Hotel Search” at least 1 time in the last 7 days where search location filter was “New York City” (behavioral data) or users whose favorite destination is “New York City”(Profile data).” Although these groups seem to be very different, everyone is interested in updates about travel deals in New York City. By including them both in one Audience, the app can send a highly relevant message while still reaching the most people possible.

  • You can build audiences from a combination of app and non-app data: Because Profile data can come from non-app sources such as a CRM or DMP, Profile + Behavioral Audiences are a great way to combine behavioral data from inside the app with Profile data from other systems. For example “Users who performed event “viewed product” at least 3 times in the last 7 days where the category was “Clothing” (behavioral data) and who have unspent store credit (Profile data from CRM system).

  • You can build more targeted audiences: Such as: “Users who performed event “read article” at least 5 times in the last 7 days (behavioral data) and users whose subscription status is one of “free” or “basic.” (Profile data).

Combining profile and behavioral data is one of the most effective ways for a brands to improve the personalization of their marketing.

Next up is Messaging.

Embrace the power of dynamic content

As noted above, dynamic content allows marketers to customize the content of a push notification for each recipient. Some things you can do with dynamic content include:

  • Have a message greet each recipient using their name

  • Mention a customer’s favorite category in a message

  • Display a message in a different language, depending on the language settings of a device

  • Show different rich content or provide different deep links depending on a user’s last purchase

  • Completely change the content of a message based on a customer’s subscription status

Let’s say you know a user is named “Todd” and he has a fondness for buying khakis from you in the past. With dynamic content you can easily send a push to him, calling him Todd and referencing khakis.

Learn with A/B Testing

Marketers have been using A/B testing for decades to improve the efficacy of their messages. When it comes to personalization, A/B testing can once again be a marketer’s best friend. By running tests involving different types of content, you can understand what messages each Audience prefers to receive.

Don’t forget the importance of timing

Timing can have a big impact on how personalized your message appears to the end user. As we mentioned before, a push that wakes the recipient up at 3:00AM is never going to be seen as personalized. Pay attention to basics like timezone-based sends and also explore how triggered campaigns and location-based messaging might unlock new possibilities to deliver messaging experiences that improve the lives of your customers.

The Time Management Factor

How long does it take to move between stages?

It honestly depends on what marketing platform you’re using. However, for most teams on most platforms, increasing the personalization of your messaging shouldn’t be a massive time commitment.

For example, we calculated what it would take for a brand using Localytics to move from Stage 1 to Stage 2. We determined that building 6 basic Profile-Audiences and 3 A/B tests would be a good way to start the transition. In Localytics, building a new Audience takes about 5 minutes and incorporating an A/B test adds between 5 and 10 minutes to the process of building a messaging campaign. So all told, creating the Audiences and A/B tests would require about 60 minutes of work in total.

The situation is similar for brands looking to progress to later Stages - when you stop and actually calculate the amount of time required to improve the personalization of your marketing, the time commitment is quite reasonable. Unfortunately, far too marketing teams conclude that delivering a personalized experience is more time consuming than it really is and never even give it a shot. Instead, we encourage you to start slow, test the waters with a few campaigns and we expect you’ll discover your new campaigns not only deliver better results, but that they do so for a surprisingly reasonable level of effort.

How Localytics helps

One of the most powerful aspects of Localytics is our platform’s ability to deliver messages that are personally relevant to the user. We have the tools, and we love teaching you how to use the tools to better your mobile marketing and keep customers coming back for more (and more, and more, and more...).

If you’d be interested in talking about how exactly to implement profile and behavioral data, do A/B work, and create dynamic content -- all for amazing returns and results -- we’d love to show you how it works in the back-end and how easily you can be on the road to personalized experiences from your brand!