Unlocking the Potential of Personalization: A Four-Pillar Approach to Optimizing Data, Insights, Content, and Operations

Personalization has been talked about since the early days of digital marketing. And although there are a number of technical options for personalization available today, we see time and again that the existing potential is not being optimally exploited.

When we at fluent:cx talk to our customers about personalization, the first step is to work out the basics and the status quo of personalization. We do this across 4 pillars, which we see as the basis for personalization and the corresponding processes:

1. Data

The obvious first. Personalization needs data. No big surprise. However, it is not uncommon for the most obvious and simple data such as gender, date of birth, or explicit preferences to go unused, even though this is where the first tactics for personalization can be launched. Deeper insights can be found in transaction data from the store or behavioral data from web analytics.

2. Insights

Based on the existing data, additional insights can be generated in the next step. Transactional and engagement data can be used to derive customer lifetime value, personas, or RFM scores, which are often used in audience segmentation but also play an important role in content personalization.

3. Content

Every now and then, we’ve had the scenario where there was a lot of talk about data and rulesets for personalization, and in the end, it failed because of the content creation. The higher the level of personalization or the complexity of the rules, the greater the need for automated content integrations. A simple replication of content into the marketing automation platform does not bring the necessary scalability here, as these then still have to be integrated manually by the marketer. Integrations with product catalogs, automatic connection via JSON, RSS, etc. are an imperative step here, as this allows content to be included and played out in large quantities without a marketer having to intervene.

4. Operations

When the number of content variants becomes very high – i.e. close to hyper-personalization – the general way of working has to change as well. The perhaps sometimes still common process in which emails, push notifications or SMS are created manually with drag & drop is no longer suitable for this or does not scale. We have assisted many customers in this process and have seen, especially in scenarios of highly personalized and multi-lingual content, that the creation of emails, for example, is more likely to be done via data tables than via classic drag & drop of content building blocks.

With its flexibility, Salesforce Marketing Cloud offers plenty of options from simple personalization with dynamic content to real-time personalization and also supports with artificial intelligence.

Dynamic Content

Even the simple if-then rules of dynamic content can be used to deliver personalized, targeted content on digital channels. And while it doesn’t actually require a complex data foundation, we still often see that dynamic content is not used at all or not used extensively. Unfortunately, the content component or the processes are often the main bottleneck here. Marketers today have to manage multiple channels, and creating or managing additional content or setting up sets of rules is sometimes simply not possible in terms of time. Therefore, as shown in our 4-pillar model, it is important that the operational processes are adapted to automate the necessary steps as far as possible.

Scripted Content

The number of content variants that can be played out with Dynamic Content is ultimately limited, because there is a point at which the use of if-then rules is simply not sufficient. With the scripted content options of the Marketing Cloud, I as a marketer can create sets of rules that look at multiple data sources and thus significantly increase the number of content variants played out. I.e. I move from target group personalization (e.g. dynamic content based on gender or interests) to 1:1 personalization for each individual recipient of a message.

Einstein Content Selection

Einstein Content Selection (ECS) is where two worlds meet. Dynamic content on the one hand, artificial intelligence on the other. With ECS, marketers are able to extend their personalization logics to include performance/engagement data to continuously optimize KPIs. Because even though multiple recipients may have the same attributes, they still respond differently to content or incentives. As a marketer, I need to provide multiple variants of content here for a given category. At this point, ECS decides, based on AI, what content to present to the recipient. This is then based either on individual engagement, i.e. has the recipient already seen and reacted to this variant of the content, or on the performance of all recipients across the different variants. Continuous optimization of KPIs with a few configurations!

Marketing Cloud Personalization (aka Interaction Studio)

Last but not least, the beautiful buzzword real time. Content for emails or push notifications is created at the time of sending and is then static in the recipient’s inbox. With the help of Marketing Cloud Personalization, I can play out content for both the website and emails in real time, i.e. when the website is visited or the email is opened. What is also worth highlighting about Marketing Cloud Personalization is that it covers anonymous visitors on the website. Marketing automation platforms, yes, focus on already identified (email address, device ID, etc.) users, but across nearly all industries, the percentage of anonymous visitors to the site is between 80 and 90%. This potential is perfectly exploited by a Real Time Interaction solution like Marketing Cloud Personalization.

“From small beginnings all things spring.”

As you can see, there are many options available to you with Salesforce Marketing Cloud when it comes to personalization. We initially conduct discussions with our customers on this topic using the 4-pillar model. We must first understand what data is available or can or should be used. We want to make sure that we implement the existing database directly in the first personalization use cases – even if these are initially smaller dynamic content rules – instead of starting directly with larger integration projects. And from then on, it is a matter of introducing continuous improvements in order to move step by step towards 1:1 personalization.