Imagine a customer buys a jacket online, returns it in a store, and later makes another purchase using their loyalty card. The e-commerce platform has their email address. The point-of-sale system has the return. The loyalty program has their member ID. If those records stay separate, the customer’s purchase history tells three different stories.
Identity resolution helps a business determine which records belong to the same person. That matters when a cashier looks up a purchase, a CRM team counts repeat customers, or a loyalty program credits points. The hard part is deciding which clues are strong enough to connect records and what to do when the evidence is incomplete.
What is identity resolution?
Identity resolution is the process of linking records or interactions that refer to the same person across systems. It can connect a customer’s e-commerce account, loyalty membership, and in-store transactions to a common customer ID. That ID gives connected tools a consistent way to look up the person and build a unified view of their activity.
The resolution process answers whose record is this? The customer profile answers what do we know about this person? For example, once the system links a store purchase to the right customer, a profile can show that purchase alongside previous orders and loyalty activity. The business still has to decide which details to display when source systems disagree, such as which address to use when two records list different ones.
For most retailers and hospitality operators, a practical starting point is the first-party data they collect directly, such as account logins, reservations, and loyalty scans. Each data point has a scope. A member ID identifies a membership, a device ID identifies a device or browser in the relevant system, and an email address can belong to a person or be shared by a household.
How does identity resolution work?
Identity resolution works by preparing identifiers, comparing records, and linking the ones that meet a match rule or confidence threshold. The inputs and matching methods depend on what the business collects and what it needs the result for.
Tokenized data
Tokenization replaces a sensitive value with a substitute identifier. For example, a business could use a token in place of an email address so a connected system can reference the record without receiving the original address. Two systems can compare tokens directly when they use a compatible scheme that produces the same token for the same input. Otherwise, they need a way to translate between the identifiers.
Tokenization can limit where personally identifiable information is exposed. The match still depends on what the token represents. If two people share an email address, a token for that address carries the same ambiguity.
Offline data
Offline data comes from interactions such as purchases at a register, bookings over the phone, and hotel check-ins. A store transaction may already be in a digital point-of-sale system. What matters for matching is whether it contains a usable identifier, such as a loyalty number scanned at checkout. Some providers also link offline identifiers to online activity through a common ID.
Suppose a customer provides a membership number when making a phone reservation. The reservation system can use that number to connect the booking to their account. If the booking only contains a surname and party size, keep it separate until a stronger identifier becomes available.
Resolution
Start by making comparable fields consistent, such as formatting phone numbers with country codes and removing stray spaces from identifiers. Then compare the records and link accepted matches to a common ID. An identity graph can store the relationships among that ID, source records, and identifiers such as account IDs and device IDs.
A link can affect more than two records. Imagine two customers log in to separate accounts on a shared tablet. If the graph treats the tablet as proof of a single identity, it can pull both accounts into one profile. When evaluating an identity resolution solution, ask how it handles conflicting account IDs, how you can see why a link was made, and whether you can undo a bad merge.
Deterministic matching
Deterministic matching applies a defined rule. For example, an online account and a store transaction can be linked when both carry the same unique loyalty ID from the same program. Another rule might require two fields to agree before linking records. The team can trace each match back to the rule that produced it.
The tradeoff is coverage. A strict rule can miss the same customer when their email changes or an older record lacks a member ID. A loose rule can link two people who happen to share a phone number. Test each rule against records your team can verify before using it to update profiles or balances.
Probabilistic matching
Probabilistic matching weighs several clues to estimate how likely two records are to refer to the same person. Two records might have slightly different spellings of a name, an exact phone number match, and the same postal code. A model can give each clue a different weight based on how well it distinguishes a genuine match from a coincidence.
Set thresholds according to what a wrong link would do. A CRM team might review uncertain matches during a one-time cleanup. If a link would let someone access an order history or spend loyalty credit, the business can require a verified identifier. Track both kinds of error in your checks. A missed match fragments the customer’s history. A false match can put someone else’s activity in their profile.
What’s the Difference Between Identity Resolution and Entity Resolution?
Identity resolution usually refers to matching records about people in customer data and marketing. Entity resolution is the broader term. It can apply to people, businesses, products, or other things that appear under different identifiers in different systems. Vendor terminology varies, so check what the product is designed to match.
A retailer might use identity resolution to link a shopper’s loyalty membership to an online account. That same retailer might use entity resolution to connect a supplier’s product code with the internal SKU for the same item. Both tasks involve matching records. Each needs its own rules for what counts as the same thing.
How different industries use identity resolution
Retail and e-commerce
Retailers can link store and online activity so staff can recognize a returning customer and CRM teams can count purchases across channels. Imagine a customer signs in online and later scans their loyalty card at the register. If both interactions are linked to the same membership, the purchase can appear in one customer profile. A cash purchase without a usable identifier can stay unlinked.
Leat’s identity resolution feature describes recognition through online logins, wallet pass scans, and identification at a physical location. If you’re evaluating that setup, confirm which identifiers your e-commerce and point-of-sale connections pass through.
Healthcare
Healthcare organizations use patient matching to link records within and across health systems. The US Office of the National Coordinator for Health Information Technology describes matching across several demographic fields, including name, date of birth, phone number, and address.
For example, a clinic might need to associate a lab result with the correct patient record even though the lab uses a different record number. A wrong match could place the result in another person’s chart. An uncertain match warrants verification before the records are linked.
Travel and hospitality
A hotel can use a verified membership ID to connect a direct reservation with a guest’s past stays. That gives the front desk a way to recognize a returning guest and see relevant preferences, such as whether they’ve previously requested an accessible room.
The person who books, the person who stays, and the person who pays can be different. For example, an office manager may use the same email address to book rooms for several employees. Keep the booking contact and guest identities distinct so stays and loyalty benefits go to the right person.
Media and entertainment
A streaming service can link activity on a phone and a television when a subscriber signs in to the same account on both. That can support a feature like resuming a show on another device. In a shared household, the service can use the selected viewer profile for viewing history while keeping the subscription attached to the account. Treating every session on a shared television as one person’s activity can mix up recommendations.
conclusion
Start with one customer experience your team can test, such as crediting an in-store purchase to an existing online loyalty account. Check which identifier both systems capture, test how it behaves with shared contact details and missing IDs, and decide who can review or reverse a link. Once those checks work, apply the same process to the next channel. Some records will stay separate until you have enough evidence to connect them.
