The 3 eras of Personalization

From Web Personalization to Data Warehouse centric ecosystems

Intro - A myriad of companies for one purpose 

Every showcased company engages, to some extent, in personalizing the customer experience.

These vendors may belong to 'Suites', position themselves as a 'Customer Engagement Platform'*, a 'Cross-Channel Marketing Hub'**, or specialize in 'Customer Journey Orchestration', among various other sophisticated descriptions.

Web Personalization has little data & context available, while platform such as Real Time Integration Management or CDP, are deeper integrated in the customer’s ecosystem and have access to 360° visitor’s profile.

*CDP: Customer Data Platform (Segment, RudderStack, Mparticle, Hightouch, etc.) 

**CCMH by Forrester, Cross Channel Marketing Hub (Braze, Insider, Bloomreach, Moengage, etc.)

In essence, the definition of ‘Personalisation platforms', ‘Real Time Interaction Management’ or ‘Customer Journey Orchestration’ hinges on their integration within your ecosystem aka available context for activation and the breadth of channels through which you can engage with your customers. 

What I define as context is all the user’s data that you could potentially leverage to build segments and activate them:

Context: Various contexts as exemple, sorted from most anonymous to most detailed. 

When it comes to channels, marketers universally recognize and refer to this aspect, making it relatively straightforward: What are the various touchpoints available to interact with your customers?

Channels: Various channels, sorted by ease of access

3 eras of Personalization

Based on this, I wanted to illustrate the evolution of this market through 3 eras and practice.

disclaimer: this is based on my personal opinion and perception.

Before delving into the first era, let me offer a brief guide to help you interpret the visuals:

Vendors addressed the challenge of "messaging delivery" through various approaches. 

I've labeled the initial phase as the "Personalization Platforms Age." Each era will be outlined following the same principle describing a set of actions:

• Something is occurring before the data activation (collecting data, building segments, etc.) 

• Then the activation occurs with a dedicated tool (or the same, depending on the era)

• Aftermath you can monitor the result and eventually reuse this data.

Data Activation Legacy - Personalization platforms Age

At the “Personalization platforms Age”, data is mostly collected by the Personalization platform itself, eventually leveraging integrations to add additional targeting criteria and unlock new use-cases. Segments built based on the data collected.

The activation is performed within the product integration perimeter: 90% of the time, this is limited to the website. Some instruments allow sending personalized emails. Later the capability to run personalizations on apps was added - but late, after the rise of the next era.

Lastly the data collected is sent back to the platform for analysis & reporting and eventually further targeting.

“Personalization platforms Age” illustration

Cons:

• Unichannel - Website (Desktop + Mobile), app joined too late 

• Temporality: Short

• Very incomplete user profile

• Absence of Identity resolution

• Limited scenarization capabilities

• Disconnected from ads & acquisition

• Siloed data, not SSOT

• Limited Flexibility / Vendor Lock

• Limited to Growth / Marketers teams

Pro:

• Fast (no activation latency), easy, tailored to customer expectations back then.

Data Activation transformation - CCMH* / Packaged CDP** Age

The second era offers major improvements: 

• It combines different instruments tightly interconnected together 

• It offers the capability to interact with users through many touchpoints. 

If we take a look at the flow, 

1-The data is collected  from touchpoints by CCMH SDKs, but CDPs are often used due to their capacity for enhancing data quality and resolving identity concerns across numerous touch points. This leads to the creation of an almost comprehensive 360° customer profile, encompassing details like viewed ads, preferred products, and comprehensive contact information.

2-Following this, within the CDPs, segments can be constructed and the audience can be transferred to the CCMH (or even constructed directly there). The activation platform then takes charge of executing the campaigns. This might involve actions like discontinuing a Facebook ad (post-purchase), dispatching a WhatsApp message, sending an email with personalized recommendations, or push notifications.

3- Subsequently the platforms collects data again to measure the impact, regardless of the touchpoints. You can analyze the result in-depth onProduct Analytic platforms for instance. 

CCMH / Packaged CDP age illustration.

Cons

• Siloed data, the activation tool is not your SSOT

• Limited Flexibility / Vendor Lock

• Limited to Growth / Marketers team

Pro

• Omnichanel - Website, apps, SMS, emails, etc.

• Tightly connected with ads & acquisition

• Advanced scenarization : more data, more touchpoint, more capabilities

To improve

• Near 360° user profile - not 100% integrated to your ecosystem (call center, logistics, finance, etc.) but important data is there.

• Identity resolution integrated - the efficiency varies a lot depending on vendors and you don’t have the control over the logic.

Data Activation now & future - DWH & Composable CDP age

We will take a look at how the MDS is deeply impacting  data activation markets as well. This era is built around cloud storage (DWH), that’s the core of the MDS.

Usually the data flow is the following:

• The data is collected (from any service) via composable CDPs (or packaged ones - we will be back to this later). And then loaded into the DWH. (click) from any service. Data can come eventually from other departments or systems (finance, logistic or CRM, content platform, etc.) 

• Data modeling, segmentation, audiences, everything is done and transformed within the DWH.

• Once it is ready, you need to pull these audiences (nothing more than a list of ID) out of the DWH in order to ingest them in your Data Activation tools. This is done by reverse ETL such as Census, Rudderstack or Hightouch or Segment itself.

• The activation will occur as in the previous era, except that it is powered by the SSOT.

• The data collected after exposure is sent back to the DWH. Then you can build reporting on Product Analytics platforms or perform very advanced BI with all the data already stored in the DWH.

DWH and Composable CDP age

Pro

• Any-channel - Website, apps, SMS, emails, call center, CRM, offline, etc.

• Temporality: Unlimited

• 360° user profile unlocked : all data stored in the DWH

• Custom build Identity resolution : if you don’t go for Packaged cdp, it can be challenging, but you keep control of it.

• Tightly connected with ads & acquisition

• Advanced scenarization: sky's the limit 

• Non-siloed data, DWH is the SSOT

• Unlimited flexibility, limited vendor lock (plug the tools you want, change vendor)

• Growth / Marketers / Product / Data / Finance / Exe

Con

• Pricing if poorly operated (compute costs)

• Technical barriers (building, maintaining and operating a complex stack)

To Improve

• Treatment delay - decreasing   : there is a delay (all data flow) but tends to decrease quickly. Still you can’t display a pop-in in real time. But CCMH will do that for you.

Conclusion - waiting for the 4th Age

As we draw conclusions on the exploration of the Modern Data Stack and its application in storing and activating marketing data and events, we recognize it as a transformative force in data management and customer engagement strategies. The Data Warehouse emerges as a robust Single Source of Truth (SSOT), effectively breaking down silos and enabling the amalgamation of diverse data sources to construct comprehensive customer profiles.

The promise held by this approach is substantial: by centralizing data storage, organizations gain clarity and coherence in their data narratives, facilitating more informed decision-making and more targeted marketing actions. The ability to activate this data—whether it be through targeted email campaigns, push notifications, or the dynamic creation of real-time segments—is not just a strategic advantage but a necessity in the fast-paced, data-driven marketplace.

However, we must temper our enthusiasm with a pragmatic acknowledgment of the challenges inherent in this model. Synchronizing a multitude of tools to work in harmony with a data warehouse can be an intricate and, at times, an expensive endeavor. There may be substantial upfront costs and resource commitments required to establish such an integrated system.

Additionally, while the Data Warehouse excels in centralizing data, it currently lags behind packaged Customer Data Platforms (CDPs) in real-time data activation capabilities. This latency is a crucial consideration for organizations seeking to engage with their customers at the moment, as immediacy often defines the success of digital marketing efforts.

In conclusion, while the Modern Data Stack holds a promise for the future of marketing data utilization and customer profile development, it is not ideal - yet. Its implementation should be carefully considered, with a thorough cost-benefit analysis and a strategic plan to overcome its current limitations. Only then can businesses truly leverage the full potential of the Modern Data Stack, ensuring that their marketing efforts are as efficient, effective, and forward-thinking as the technology they employ.

New vendors are streamlining data management by embedding Product Analytics and CDPs directly into Data Warehouses, yet sub-second 'real-time' activation remains elusive. Industry giants like Snowflake and Databricks are optimistic about overcoming this hurdle. If they succeed, the next transformative phase —4th Age, delivering instantaneous customer Personalization —may be just around the corner.