Customer Identity & Identity Graphs

Understanding who your customer is has never been more importnat for todays data driven companies.

At the same time it has never been harder.

Disperate technology platforms, marketing channels, identifiers, and an ever confusing regulatory landscape have left companies search for an easy way to answer a simple question.

Who is my customer?

Scroll down to see how Narrative can help.

Understanding Customer Identity

Companies often have some form of identity that helps identify their customers. It may be an email address, phone number, or even a cookie.

Identity is Incomplete

The challenge is that the identity that they have is often insufficient to undestand the customer across all of their device and marketing channels.

You always have blind spots that your own data won't be sufficient to fill in.

Identity Graph to the Rescue

An identity graph helps those companies create a fuller picture of their customers.

These solutions take your blind spots, make them visible, and give you a fuller picture of your customers.

The technology they use to do this is called an Identity Graph. Identity Graph's can be extremely powerful, but working with them is not always easy.

Graphs Are Complex

How an identity graph is built and consumed isn't as trivial as it seems.

We'll show you how Narrative helps you create a custom Identity graph and look like a hero in the process

Let's break down the various elements of a graph.


In an identity graph a Node one of the two basic building block. A node typically represents a tochpoint with a user.

It might be setting a cookie when they visit a website or getting their phone number when they dial into a call center.

A node is a piece of identifiable information that tells us something, but doesn't tell us the whole picture about the user.


Edges are the things that allow us to start building a richer view of the user.

It its most basic sense an edge in an identity graph tells us that two nodes are somehow related.

For example if someone told us that an email address and a phone number belonged to the same person it would allow us to draw an edge between the nodes for that email address and that phone number.

Edge Methodology

If someone was able to tell us about all of the edges, with perfect accuracy our jobs would be easy.

In reality, in order to create an edge we have to rely on different techniques. Those techniques are encoded as the edge methodology.

The different methodologies can get complex, but at a high level you can think of the methodologies as being "declared" or "inferred"

Declared edges come from where the user themselves, through some means, has told us about the connetion between two pieces of information.

Inferred edges are derived from information that we have that might point to two nodes being related, but we can not be completely sure taht we are right.

Inferred Edges

Why do we use inferred edges at all?

Relying only on declared edges creates a problem of scale. Users don't tell us everything we need to know to develop a full picture.

In order to create a fuller picture we use inferred edges.

How the inference is done can vary based on the available data and the company doing the inference, but a common technique is to use co-occurance.

Co-occurance means that nodes or identifiers are seen together. If everywhere we see one identifier we then see another identifier, even if we don't see them tied together in any meaningful way, we could start to infer they are related. That is often how inferred edges are created.


Use the checkboxes below to show how the makeup of the graph changes when we use different edge methodologies.

The more nodes we see in the graph, the fuller the picture of consumer identity we have.

Identity Types

We often think of identity as applying to a single person. But we may want to understand the identity of a household or even a business.

Both nodes and edges have a concept of identity type. An email address usually belongs to a single person so a node representing email will have an "individual" idenitty type.

Two emails from different members of a house may have an edge between them. Those two nodes now are related, not as the same person, but as the same household.

When building/using a graph it is common to define the type of identity you are interested in and have the resulting graph filtered based on that choice.


Use the checkboxes below to show how the makeup of the graph changes when we use different identity type

The more nodes we see in the graph, the fuller the picture of consumer identity we have.


We've shown how methodology and identity type can be used to filter the graph, but your optoins should not be limited to those two attributes

Narrative makes it possible for you to build entirely custom graphs on the fly by filtering on anything within the graph.

Only looking for mobile ids tied to users in your CRM? Great. We've got you.

Looking to combine identity graphs from 4-5 partners? No problem.

As we continue, just remember that Identity Graphs are complex and include a lot of underlying data. We make it easy for you to cut through the complexity and get what you need to run your business.


In a perfect world you would have all of the information you need to build your own identity graph. Most companies don't have that luxury.

It would be great if you could then find one partner that had everything you needed. That too isn't true in reality.

In order to get a full view of identity you often have to work with multiple partners across multiple ecoystems. Those providers all have slighly different technologies and ways of expression their graphs.

Narrative allows you to combine the information from those graphs into a single consolidated graph


One of the trickiest parts of working with an ID graph is making it from point A to point B.

This is called graph traversal. As you can see here not all nodes for a given user are directly related to all other nodes.

Traversing the graph to get from one node to another can be a complex process that involves a lot of steps. Thanksfully Narrative can deliver you a fully traversed graph, meaning we do the consolidation for you.


In our diagrams we've shown you a graph for a single user, but an identity graph for one user isnt' that useful.

An identity graph solution really needs to provide you with scale across millions or hundreds of millions of users.

This can be a challenge, not just from an information perspective, but the technology to do the graph building and graph traversal at scale is not trivial.

Narrative has nearly a decade of experience in working with graphs, so you can be sure that you'll get what you need when you need it.


Just because something is possible, doesn't mean it is easy. At Narrtive we focus on taking the complex and make it simple.

Our Identity solutions allow you to define exactly what you're looking for and have it delivered to you without having to understand all of the underlying complexity.

Most people don't want an "identity graph" they just want a better way to communicate with their users and potential customers.

We can help you do that.


< Back

Hi! I’m Rosetta, your big data assistant. Ask me anything! If you want to talk to one of our wonderful human team members, let me know! I can schedule a call for you.