Why Identity Graphs (and Identity Resolution) Matter for Marketers

This was originally written and published on Medium.

Identity (ID) Graphs and Identity Resolution have been the buzz in marketing circles in recent times. Experts from all backgrounds and martech solution providers world over, are all reading, blogging, debating, podcasting, and putting out videos about ID graphs. The background, the rationale to it, the technical deep dive, the implications for marketers, …. all are fascinating reading for anyone interested or directly affected by it.

This blog is about my learnings and some interesting sources for anyone wanting to get an overview, or do a deep dive, on this topic.

Since Google announced cookie deprecation by H2 2024 (Google Chrome has ~60% of the browser marketshare), the marketing world has been scrambling to ready itself for a cookieless world. Cookies (tiny text files sitting on your device that track all manner of usage and behaviour attributes that help profile users for marketing purposes) have been the backbone of the entire martech/adtech ecosystem over the past decade, turbocharging an online advertising ecosystem (worth >US$ 250 billion in 2023) into a broader marketing ecosystem powering everything from online ads to ecommerce sales to 1st party CRM solutions to connected TV ads, and now offline Digital OOH programmatic. Once the cookies are gone, how will marketers engage their consumers and drive the ads, the sales, etc. mentioned above?

This is where ID graphs come into the picture. Simply put, it’s a database that

a) Hosts a plethora of data points about a consumer — demographic, behavioural, interests and preferences, predictive intent, purchase history, etc. This can be a mix of identifier data (e.g. Email) and non-identifier (e.g. IP address).

b) Enables you to organise the data into meaningful clusters, which in turn can be used to build a Unified or Single Customer View, the new Holy Grail of segmentation, personalisation and targeting.

ID graphs help identify relationships between real-world entities and seemingly unconnected data points to stitch together profiles from different systems and interactions, and de-duplicate profiles based on shared features. It is done by matching personal identifiers (PII data) to digital identifiers across devices and channels, enabling marketers to then target audiences with a greater degree of relevance and precision using a mix of deterministic and probabilistic models.

(Sidebar — B2B audience data could have a more deterministic skew vis-a-vis B2C audiences since there are more identifiers at play. Besides email, you could have company names, designations, tenure, etc. available to use. Plus, the audience tends to be more willing to keep the data updated and share more about their likes/dislikes, intent, etc. as a way to get relevant content, insights, and offers from B2B sellers.)

Benefits

a) A persistent ID across devices and channels, helping create a holistic profile of a consumer

b) Audience segmentation, personalisation, and targeting — analysing customer behaviour and path-to-purchase in realtime

c) Marketing attribution and performance optimisation

d) Enriching customer profiles and insights, by consistent data ingestion from various sources

e) Maintaining customer privacy and data security

Challenges

a) There’s no industry-accepted single ID graph or identity resolution standard. So marketers and data providers have to deal with multiple ID graphs, and devise ways to match identifiers across platforms i.e. move from a cross-device mindset to a cross-identifier mindset

b) Data sharing, or the lack of it. Between 1st party data with clients/publishers, walled gardens (FB, Amazon, etc.) that fiercely guard their data, and suspect credibility of data privacy/security practices, a lot of useful customer data sits in silos. This could be solved through the use Data Clean Rooms (WTF is a Data Clean Room) to merge data sets, but they are in their infancy as well

c) Adoption and integration of the right tech solutions, that will enable this to be done at scale and in real time. Legacy systems and ill-defined ownership of data/tech related investments within organisations make this an onerous task in most companies. Many marketers are both not ready for an initiative of this complexity nor able to rally all relevant stakeholders to setup, adapt, and fully maximise the value from ID resolution.

Key players

Example of Identity Solutions

UID 2.0 — Trade Desk

Ramp ID — LiveRamp

Epsilon CORE IDs — Epsilon PeopleCloud

ZeoTap ID+ — ZeoTap

Panorama ID — LoTame

Example of Build your own ID graphs

Amazon Neptune

We are at the cusp of all this accelerating, and being adopted, at a scale and pace unimaginable a year ago. Google’s announcement has kicked off a race to build, rollout, and gain widespread adoption of various solutions that can actively improve this need. It’ll be fascinating to watch how this evolves, and thereby changing the way marketing is done in the 21st century.

Here are some more resources you might enjoy learning more about this topic from.

Identity Graphs on AWS — Link

How identity graphs are built — the present and the future — Link

Identity Resolution Explained in Less Than 10 Minutes — Video Link

Identity Graphs — Overview and Key Concepts — Video Link