How identity resolution works and what most guides leave out

Think about a lost and found bin at the end of a long season, the kind that piles up at a ski lodge or a summer camp. Every glove, water bottle, and jacket in there is missing its owner's name tag. Someone has to look at a worn-out mitten, cross-reference it with a stack of found-item reports, and figure out whose kid it actually belongs to. Get the match right, and the item goes home. Get it wrong, and you've handed someone else's coat to a stranger.

That's essentially the job identity resolution does with your customer data. Every time someone browses your mobile app, visits your site from a work laptop, or clicks an email link from their phone, they leave behind a data point tied to a different device ID, a piece of information that's missing its name tag. Identity resolution is the process of matching those fragments back to the same person, so your team isn't treating one loyal customer like five different strangers.

Brands invest in identity resolution because disparate data points create real business costs: wasted ad spend on people you've already converted, customer experience gaps when support doesn't recognize a returning shopper, and reporting that undercounts how well your marketing is actually working.

Key takeaways

• Identity resolution matches fragmented customer data, like an email on one device and a cookie on another, to figure out when they belong to the same person.
• It relies on two matching approaches, deterministic and probabilistic matching, that trade off certainty for coverage.
• Companies typically build identity resolution through a customer data platform or an in-house, warehouse-native setup, and each comes with different costs and tradeoffs.
• A lot of the identity resolution market runs on data that wasn't collected with the visitor's direct consent, which is a compliance question worth asking any vendor upfront.
• Match rates vary widely by traffic source and vendor, so "it works" for one brand doesn't guarantee it'll work the same way for yours.
• Maintenance is an ongoing cost, not a one-time setup fee, since identity graphs decay as people change emails, devices, and phone numbers.
• Many brands buy identity resolution to answer a measurement question (is my marketing working?) that a marketing mix model can answer without touching personally identifiable information at all.

What identity resolution actually is

At its core, identity resolution connects disparate data points, like a device ID, an email address, and a loyalty account number, into a single, unified customer profile. Instead of your data teams staring at ten different records that might all be the same shopper, identity resolution rules do the work of stitching those data points into one entity. That single, unified profile becomes the reference point every other system checks against.

The output is usually called a unified customer profile or a unified customer view. Instead of customer records sitting in separate silos (your email platform, your ad platform, your point-of-sale system), you get one connected record living inside your identity graph that reflects the same person's full set of customer interactions, pulled from multiple data points rather than one isolated system.

This matters most when your customer touches your brand across channels. Someone might browse your mobile app during a commute, then complete a purchase from a laptop that evening, leaving behind different IP addresses each time. Without identity resolution, those look like two different sessions from two different people. With it, your team gets comprehensive customer profiles that reflect reality instead of a fragmented view.

How identity resolution works

Every identity resolution system relies on some combination of two matching methods: deterministic and probabilistic matching. Most vendors blend the two rather than leaning on just one, and understanding the logic behind each helps you evaluate any tool or vendor pitching you a unified view of your customer.

Deterministic matching

Deterministic identity resolution connects records using verified data points, like the same email address, a phone number, or a login ID, that unambiguously belong to one person. If two records share the same email address, deterministic matching considers that a confirmed link. This is the most accurate matching method available, but it only works when you already have that verified data point in both records.

Probabilistic matching

Probabilistic identity resolution takes a different approach. Rather than requiring an exact identifier, it uses statistical algorithms to weigh signals like IP addresses, device IDs, browsing patterns, and timing to estimate the likelihood that two records belong to the same person. It's useful when you don't have a shared identifier to work with, but it comes with a tradeoff: the resolved identities you get back are confidence scores, not certainties.

Here's a quick side-by-side to keep the tradeoffs straight:

Most identity resolution strategies lean on both methods together, using deterministic matching wherever possible and falling back on probabilistic methods to fill in the gaps.

Why businesses invest in identity resolution

Modern identity resolution goes well beyond simple cookie matching. Connecting those data points gives your team deeper insights into how customers actually move between channels and a fuller view of the customer along the way, and the belief is that real business value shows up in a few concrete places:

• Personalization and customer engagement strategies: Use purchase history and browsing patterns to build relevant customer engagement strategies instead of generic, one-size-fits-all messaging.
• Fraud detection: Run fraud detection by connecting suspicious account activity across devices, catching duplicate signups from the same customer or stolen credentials that a single, isolated record would miss.
• More accurate attribution: Get a clearer, more accurate attribution of which channels a specific customer actually engaged with before converting.
• A complete picture for support and sales: Reps get a complete picture from an existing customer profile instead of guessing from one ticket or one order.
• Cleaner reporting: Reduce duplicate customer records and improve data accuracy across every system that touches that customer, from your CRM to your ad platforms.

That connected data becomes the foundation for richer user profiles that update as your customer behavior changes. Effective identity resolution takes some ongoing work to manage data across every source feeding the graph, but the effort is worth it for the right use case. Get the data quality right, and those user profiles translate directly into a smoother customer experience, all rolling up into one unified customer view your whole team can actually trust.

The two ways companies build an identity resolution strategy

Once a company decides it needs identity resolution, the next decision is how to implement identity resolution across its stack. Businesses generally take one of two routes.

Customer data platforms (CDPs)

A CDP is purpose-built software, one of the more popular identity resolution tools on the market, that ingests, normalizes, and matches your customer data across systems on your behalf so your team isn't maintaining ten versions of the same customer profile, often without ever touching your own data warehouse. You feed in data sources like your email platform, your ad accounts, and your CRM, and the CDP handles the matching logic behind the scenes to build a single customer view. This route trades a bigger monthly bill for a faster time to launch, since you're not building matching logic or identity resolution tools from scratch. (If you're in the market for one, check out our list of identity graph companies to narrow down your list of potential partners.)

Warehouse-native, in-house builds

The other path is warehouse native identity resolution, built directly inside your existing cloud data warehouse. Data teams write custom SQL to define their own matching rules against customer identifiers and user IDs already sitting in that data warehouse. This route gives you more control over exactly how matches get made, but it also means your team owns the ongoing work of maintaining and updating that logic as data sources change.

Neither path is inherently better. A leaner data team with a straightforward tech stack often leans toward a CDP for speed. A larger organization with dedicated data infrastructure and unique matching needs might prefer the control that comes with warehouse native identity resolution. The right call usually comes down to how much engineering bandwidth you have, and how customized your matching rules need to be to get a real view of the customer.

Questions to ask before you buy

Yes, understanding how identity resolution works is important if you're thinking of signing a contract with a company, but you'll get more from a deal if you go a little deeper. Before you sign, there are a few questions worth asking that don't get nearly enough attention.

Where did this data actually come from?

Not every identity resolution vendor sources data the same way. Some genuinely resolve identity using data you collected directly, first party data your customer handed over when they signed up for your list or made a purchase. Others resolve anonymous website visitors by matching a hashed identifier against a third-party database of previously collected emails and phone numbers, party data your specific visitor never gave your business directly.

That distinction matters. If a vendor can tell you a name and email address for someone who's never filled out a form on your site, it's worth asking directly where that match came from and whether the underlying data was collected with consent. This applies whether you're evaluating a standalone vendor or one of the larger customer data platforms, since even well-known customer data platforms can source party data differently depending on which module you turn on. It's a compliance question, and the answer varies a lot by vendor and by which state or country your customer is in.

How much will match rates actually vary?

Vendors love to lead with an average match rate, but the real answer is that resolved identities depend heavily on your traffic. A B2B SaaS company with high-intent organic traffic might see strong results. A brand with mostly paid social traffic and a lot of one-time visitors might see far lower match rates from the same vendor. Ask for benchmarks from businesses that look like yours, not just an industry-wide average.

What's the real cost of data security and maintenance?

Standing up an identity graph isn't a one-time project. Those identifiers age quickly: people change email addresses, phone numbers, and devices constantly. That means maintenance work is ongoing, not a line item you check off after launch. A few things to budget for:

• Ongoing vendor or platform fees, on top of any initial build cost
• Ongoing data team hours to manage duplicate data and fix data quality issues as they appear
• Periodic audits of your data sources to confirm you're not carrying data redundancy that inflates your customer counts
• Security reviews to confirm sensitive, personally identifiable information (PII) is stored and accessed the way your policies require

Is entity resolution the same thing?

You'll sometimes see the term entity resolution used interchangeably with identity resolution, and technically, it's a bit broader. Entity resolution can apply to any kind of record matching, including business entities, addresses, or product listings, not just individual customer identities. A vendor that says entity resolution might be describing a much wider identity resolution system than the one you actually need, so if you see the term entity resolution in a pitch deck, it's worth clarifying whether they mean person-level matching or something broader.

Do you actually need identity resolution?

Here's the thing worth sitting with before you implement identity resolution: a lot of brands take on all that identity resolution work to answer a question that isn't really about identity at all. Fragmented data alone doesn't mean you need a whole new system. Brands usually want to know if their marketing is working, and that's a measurement question with more than one way to answer it.

Identity resolution earns its keep when you genuinely need to recognize and act on an individual customer, like personalizing an email based on customer behavior, flagging a returning high-value shopper to improve customer experience for your support team, or running fraud detection by connecting suspicious activity across accounts. Those are real, individual-level use cases, and no aggregate model replaces them.

But if what you're actually after is understanding whether your ad spend is driving revenue, whether your brand campaigns are paying off, or where to shift budget for higher efficiency, you don't need a name and email tied to every visitor to get that answer. A marketing mix model looks at your spend and results in aggregate, using statistical methods to separate what's driving revenue from what's just noise. It never needs to know who any individual customer is, which means it sidesteps the consent questions, the match rate variability, and the maintenance burden that come with an identity graph entirely.

Not every version of identity resolution requires the same investment either. If your customer data is already fairly clean and mostly first party data, deterministic matching alone might get you most of the way to a unified profile for the same customer without leaning on probabilistic matching or an expensive identity graph vendor. If your traffic is anonymous and cross-device, you'll lean harder on probabilistic matching, a bigger identity graph, and everything that comes with maintaining one.

The two approaches aren't in competition so much as they answer different questions. If your team's actual goal is measuring marketing effectiveness and making smarter budget calls, that's worth separating from the identity-level use cases before you invest in a whole new system.

Where Prescient comes in

If your team has been eyeing identity resolution because you want a clearer read on whether your marketing spend is paying off, that's exactly the gap Prescient's marketing mix modeling platform is built to close. Prescient measures the true impact of every campaign, including the halo effects that show up as branded search, organic traffic, and even sales on retail partners like Target, Walmart, and Amazon, all without needing to identify a single visitor.

Because Prescient works from your existing spend and performance data rather than person-level tracking, you get campaign-level, daily updated insight into what's actually driving revenue, along with tools like saturation curves and the Optimizer to guide your next budget decision. If you're trying to solve a measurement problem, not a personalization problem, it's worth seeing what the Prescient platform can do for your brand. Book a demo to see it in action.

FAQs

Is identity resolution legal in the United States?

Identity resolution itself isn't illegal, but how a vendor sources its data determines how much legal risk you're taking on. Using first party data your customers directly gave you is generally on solid ground. Matching hashed identifiers against third-party databases of purchased or scraped emails is a legal gray area that varies by state, and it's an area where privacy laws are actively evolving. It's worth having your legal team review any vendor's data sourcing before you sign on.

How is identity resolution different from cookie-based tracking?

Cookie-based tracking follows a browser or device across sessions on a single domain, but it breaks down the moment a customer switches devices or clears their cookies. Identity resolution aims to solve that specific problem by connecting the dots across devices and channels using other verified data points or behavioral signals. That gives you a more durable view of the customer instead of a fragmented one tied to a single device, whether you get there through deterministic matching or probabilistic identity resolution.

What's a good match rate for identity resolution?

There's no universal benchmark, since match rates depend heavily on your traffic mix, your industry, and the vendor you choose. A business with a lot of returning, logged-in customers will typically see higher match rates than one with mostly anonymous, first-time visitors. Ask any vendor for match rate examples from businesses similar to yours rather than relying on an industry-wide average.

Can small businesses use identity resolution, or is it only for enterprise?

Small businesses can use identity resolution, though the cost-benefit calculation looks different than it does for an enterprise brand. CDPs with lower-tier pricing plans have made this more accessible, but a smaller data team should weigh the benefits of identity resolution against the ongoing maintenance and data security work before committing to a full identity resolution strategy.