Identity: Navigating its Complexity in CX

The convergence of AdTech and MarTech is just one of the many reasons why the topic of identity is complex for any organization to tackle.

I've observed numerous conversations where two sides discuss the identity topic using a different language without realizing it. Preventing fraud in loyalty programs or limiting ad spend on Facebook are both critical, requiring a nuanced approach to addressing identity, but in vastly different ways.

Whether you're engaging in internal discussions, strategizing with partners, or evaluating new vendors, I'll provide you with a framework comprising five pillars to drive better outcomes in your CX use cases requiring identity.

1. Users: Anonymous ⇔ Authenticated

Anonymous Users: These are users who have not disclosed their identity to the brand. They may be interacting with the brand for the first time or have had previous interactions.

Anonymous Identities can include 1st-party cookies, 3rd-party cookies, Mobile Advertising IDs (MAIDs), or an IP Address.

Authenticated Users: These are users who have directly identified themselves to the brand. Identities used for authenticated users, such as an email address or phone number, directly link a record to an individual, classifying these identities as PII.

Before digging into any identity conversation, it's crucial to clarify whether the pain points you are addressing relates to anonymous users (particularly relevant given the ongoing discussion around the deprecation of third-party cookies) or authenticated users.

The following four pillars will serve as new layers atop the Anonymous ⇔ Authenticated foundation.

2. Channels: Paid Media (Advertising) ⇔ Owned Channels (Marketing)

Paid Media (Advertising): You may aim to target users directly on paid media channels (e.g. Walled Gardens such as Facebook, Google), through a DSP (e.g. The Trade Desk, Criteo) or through a Data Onboarder (e.g. Liveramp, TransUnion).

• Anonymous targeting on paid media often involves the process of a cookie sync, historically accomplished with 3rd-party cookies. More effective solutions are achieved through 1st-party cookies.

• Authenticated targeting on paid media is accomplished using hashed PII, an onboarder to anonymize the identity, or the new Conversion API built by Walled Gardens.

Owned Channels (Marketing): This encompasses both inbound channels (web, mobile app) and outbound channels (email, SMS, direct mail).

• Anonymous users can only be targeted via inbound channels, with personalization driven by the 1st-party cookie stored on the browser.

• Authenticated users can be targeted through the channels associated with the type of identity you have access to, such as an email for an email address, an SMS for a phone number, etc. 

3. ID Source: 1st-party  ⇔ 3rd-party

1st-party: The identity is collected (email) or created (1st-party cookie) by the brand directly.

3rd-party: You are accessing an identity from an external provider such as the RampID from Liveramp or the FabrickID from TransUnion. 

4. Technique: Deterministic  ⇔ Probabilistic

Deterministic: This technical term simply means that an exact match between two values is required to consider that two records represent the same individual.

• Anonymous users are identified by linking a record based on an exact match in a brand-owned or vendor-provided ID Graph.

• Authenticated users are deduplicated based on exact PII value matches, such as two records with the same email address.

Probabilistic: As the name suggests, this technique seeks enough similarities to link individuals.

• Anonymous users are recognized by combining multiple signals such as IP address, device, browser, etc. This technique is also referred to as fingerprinting, with both Apple and Google taking steps to prevent it.

• Authenticated users are deduplicated via probabilistic techniques that combine different approaches, such as fuzzy matching (evaluating similarity between values).

5. Latency: Batch  ⇔ Real-Time

Batch: This involves resolving identities across an entire table of records or a subset of it. The objective is to match, deduplicate, and merge records. Both deterministic and probabilistic techniques can be used.

Real-Time: In this scenario, the aim is not to deduplicate multiple records but to identify one specific user. This is typically achieved via exact matching against a brand-owned or vendor-owned ID graph.

Identity is multifaceted

🔑 Identity is multifaceted, and navigating through its complexities requires a nuanced understanding. 

With the framework provided, we now have the tools to gather more context and evaluate suitable solutions.