Overview of customer identity and resolution

Identity resolution identifies the events on a customer's journey and creates a record of that journey. These events, such as email clicks, webpage views, products added to cart, orders placed, and support tickets submitted, can be captured across channels and devices.

Optimizely Data Platform (ODP) uses many identifiers to develop customer profiles, including email address, browser cookie, and even custom identifiers like a help desk ID, account number, shopping cart ID, and more.

The benefits of identity resolution include:

• A better understanding of how customers engage with your brand across channels
• Insight into what makes for more relevant customer messaging and communications
• Delivering a consistent brand experience across all channels
• A more accurate count of your unique customers
• New opportunities to drive revenue

How identifiers work

ODP generates default identifiers for all accounts, and integrations with another platform generate custom identifiers. A low- or high-confidence level further distinguishes these identifiers and identity resolution.

• Default identifiers – Email addresses and browser cookies
• Custom identifiers – Content that identifies a customer in another system, such as your support or e-commerce platform.

Identifier confidence

All identifiers have a confidence level to determine the identifier's role in the identity resolution process. 

High confidence identifiers are unique to an individual customer and you should use these only when you have a high degree of confidence that the identifiers represent a single customer. High-confidence identifiers merge events and profiles during the identity resolution process. Examples of high-confidence identifiers include a customer's email address, e-commerce ID, and service system ID.

Low confidence identifiers are unique to a device or browser and may resolve to multiple customers in certain situations. You should use these when there is a weak degree of confidence that the identifiers represent a single customer. Low-confidence identifiers only merge events and profiles during the identity resolution process if no high-confidence identifiers are involved. Examples of low-confidence identifiers include a customer's VUID, other browser cookie IDs, and push tokens.

ODP or a third-party integration sets an identifier's confidence. However, you can manually set a custom identifier's confidence when using a custom integration. Consult the ODP developer docs, your Customer Success Manager, and your developer to correctly configure any custom identifiers and understand how they impact your account's identity resolution process.

How resolution works

Identity resolution forms a complete record of a customer's journey and interactions with your brand. Identifiers trace events across multiple channels and devices back to a common individual. Instead of creating numerous customer profiles, ODP can merge related information. 

When you include matching high-confidence identifiers with events, they are connected. The presence of these identifiers proves the linkage between the existing profile and new events. 

When you include low-confidence identifiers, the new events are only linked with existing information if no high-confidence identifiers are available. Low-confidence identifiers can also move from one profile to another if the same values are associated with non-matching high-confidence identifiers. When this movement occurs, any previous events related to the identifier remain on the original profile.

Example scenarios

High confidence + Low confidence – Imagine you have an in-store kiosk. Your customers can use the kiosk to sign-up for exclusive store discounts with a web form. Each submission results in an event that includes an email address and a browser cookie (VUID). The email address is considered a high-confidence identifier, so it is used to link this new event to existing profiles when they are available or create a new one if not. Meanwhile, the browser cookie, which is considered a low-confidence identifier, moves from one profile to the next with each submission.

Low confidence + Low confidence – Imagine you have an anonymous visitor on your site today who browses some products then leaves. ODP tracks the behavior and associates it with a unique customer profile and browser cookie. Two months later, that visitor comes back on the same device and browses some more. ODP then makes the connection between the two visits using the browser cookie and merges the events on the unique customer profile because no other high-confidence identifiers were provided.

Example 1 (Shared Device)

Your customer's object looks like the following:

You receive an event:

This event seems to indicate that Jane is using a browser formerly associated with John. Because `Cookie ID` is a Low Confidence identifier, instead of merging these customer's together, the cookie is moved to associate it with Jane.

If ODP merged these customers together, then every single person that shared that computer would be merged into one, so the `Cookie ID` was moved.

Example 2 (Integration/App)

Your customers object looks like the following:

You receive an event:

This event seems to indicate that John's Shopify account has a secondary email. Because Shopify ID is a High Confidence identifier, ODP merges the customer records.

Common challenges

Email forwarding – When a customer interacts with an email, their action associates with unique tracking parameters, and the event is attributed to their profile. If you send an email to john@optimizely.com, the links in that email have specific parameters to assign the subsequent events to the existing john@optimizely.com customer profile.

This behavior can get complicated if the message is forwarded to other individuals because the email still includes the original tracking parameters. For example, if John forwards the email to jane@optimizely.com and Jane follows a link, John's profile will contain Jane's actions until Jane clears her tracking cookies or is associated with her own unique customer profile.