Unlocking identity resolution with Snowflake and Braze

TL;DR

Braze and Snowflake offer complementary tools to help marketers with identity resolution, supporting more accurate message targeting and deeper personalization.

Key takeaways

• Snowflake offers four distinct tools to support identity resolution, including deterministic matching and LLM adjudication
• The Braze Data Platform offers a collection of identity resolution capabilities designed to collect ongoing engagement data and resolve identities before triggering personalized cross-channel campaigns

In order to deliver experiences, marketers need a complete view of the customer, one that unifies digital and off-line transactions from both known and anonymous touchpoints. But for many brands, achieving this unified view can introduce significant complexity and overhead.

When a single customer journey spans multiple devices, browsers, and sessions, one person can look like many, with duplicate records creating flawed audiences, misdirected messages, and inaccurate data. By leveraging identity resolution, marketers can connect fragmented customer data signals (from different channels, devices, browsers, and touchpoints) into a single, accurate profile of one real person.

Most brands have already invested heavily in cloud data warehouses like Snowflake as their operational source of truth. The data, compute, and governance live right inside Snowflake. So it makes sense to resolve identities where the data resides, rather than adding an additional platform layer in your martech stack. And when Snowflake’s data foundation with native identity resolution pairs with real-time activation and orchestration in Braze, it makes it possible for marketers to:

• Reduce customer acquisition costs (CAC) by helping minimize redundant ad spend aimed at re-acquiring existing customers who are already in your database.
• Protect their brand experience by enabling frequency capping and suppression rules to apply across resolved profiles and avoiding the dreaded moment where they email a discount to someone who just paid full price yesterday.
• Help mitigate compliance risk by enabling consistent unsubscribe, opt-out, and consent updates across duplicate records.

Unifying the customer view with native identity resolution in Snowflake

Getting identity resolution right can be the difference between serving up broken vs. brilliant experiences. Imagine a (fictional) consumer named Michael Jones who signs up for a "find my style" event online, and later makes an in-store purchase, sharing new preferences and context each time. If a brand doesn't have identity resolution, marketers will see two strangers instead of one loyal customer—and Michael will be treated like a first-time visitor every time he engages.

To avoid that outcome, Snowflake makes it possible to create a single view of the customer natively in their data warehouse, using four different methodologies to unify first-, second-, and third-party user data and then syncing the final user identity using Braze Cloud Data Ingestion (CDI), supporting personalized customer engagement at scale. First, deterministic matching starts with the clearest signal available in the user data. When two profiles share an exact match through reliable pieces of data such as an email address, a phone number, or a loyalty ID, the records are merged into a single user identity. In Michael's case, his event registration included his color and season preferences and his in-person checkout included the list of merchandise he purchased. And because he shared the same phone number in this interaction, both profiles could be merged to create one complete view of Michael’s preferences.

Of course, not every match is that clean, and that is where fuzzy matching picks up the slack, catching small inconsistencies that deterministic matching cannot, such as typos, formatting differences, and name variations (e.g. "Bob Smith" and "Robert Smith" with the same phone number or "Cathy Lee" and "Kathy Lee" with the same email). Some user data includes weaker signals that are related to known users but does not contain enough information to be resolved using deterministic or fuzzy matching. In these cases probabilistic or ML-based matching resolves profiles based on first-, second-, and third-party data collected within the Snowflake environment (for example, shared addresses, cookie data, device information, etc. are matched to relevant users in this step).

To resolve the most ambiguous cases, Snowflake Cortex AI leverages LLM adjudication. Using a structured prompt containing two user records' attributes side by side, Snowflake Cortex AI returns an output along with a confidence score and its reasoning. For example, it might compare two profiles that have identical names and addresses, however one record specifies “Jr.” as a suffix. It might reason that these are father and son profiles and attribute a confidence score to the output. It offers an explainable way to resolve identity even in situations that don't fit neatly into the earlier steps.

Activating user profiles in Braze through Cloud Data Ingestion

When identity resolution lives in isolation, brands suffer from operational friction and missed revenue opportunities. Companies waste significant acquisition dollars paying to re-acquire high-intent or existing customers simply because their identities are fragmented in separate systems.

You can't connect in the moment with your customers without the full picture in place. Braze keeps customer context connected and current in real time by quickly syncing fresh user profiles from Snowflake using CDI. You do not need to secure resources to build complex ETL pipelines or wait for manual data uploads to access the golden record in Braze. Inside Braze, the Braze Data Platform offers a collection of identity resolution capabilities designed to collect ongoing engagement data and resolve identities before triggering personalized cross-channel campaigns.

The Braze Data Platform offers flexible user identification that captures both anonymous and known user profiles across channels, carrying past engagement history over automatically as soon as an anonymous user identifies themselves. Then automated Identity Resolution de-duplicates records native to the engagement engine so messages can land on a unified profile instead of in front of the wrong recipient. The Braze Data Platform collects ongoing engagement signals, such as email captures or app downloads, enabling you to progressively enrich the user identities for customer engagement.

Closing the loop with Snowflake Data Sharing

Customer engagement is a data-rich process and activation is only part of the story. Marketers depend on the data generated through the engagement process to learn more about their customers in order to tailor experiences that map to the changing user preferences. The data generated inside Braze (such as email opens, app clicks, message suppressions, and campaign conversions) is continuously streamed back into Snowflake in real time via Snowflake Secure Data Sharing to further enrich the “golden record” for each customer as well as to power business intelligence at scale.

This bidirectional data flow creates a complete feedback loop to power unified business reporting. Data science teams can join live marketing interaction events with additional data sources such as point-of-sale (POS) records, offline store transactions, customer sentiment, customer lifetime value (LTV) metrics, revenue, profits, and more inside Snowflake. By bringing engagement performance and financial metrics together in the warehouse, marketing leaders can prove clear campaign ROI to executive stakeholders within the exact same data environment.

Final thoughts

Getting the identity resolution right is the key to delivering unified, informed, and relevant experiences at scale. Brands can achieve this within their data warehouse and engagement platform without onboarding an additional solution.

Interested in learning more? Dig deeper into Snowflake’s identity resolution capabilities here.