Ask buyers who own an identity-resolution tool what they'd tell their past self and the theme is verification, not features. Whether the match rate holds up on their own data, whose graph is really doing the matching, what the pipe actually costs, and whether their CDP already did most of it. The vendors differ; the regrets converge on trust and the meter. Here they are, in the order buyers hit them — for the identity question inside a customer data platform specifically, see the related lesson in what CDP buyers wish they'd known.
The seven lessons
Match rate is the whole game — validate it on your own file, and expect it to vary
Everything else is secondary to whether the tool actually resolves your people. Buyers warn that the match rate on the slide is a marketing number, not a measured one: a digital-media company flagged uneven availability from a geographic standpoint that constrained its global use, with scale dependent on derived IDs. Others found during validation that vendor match numbers didn't line up with their own internal identity figures. Run the vendor's match against a known slice of your file, compare it to what you already resolve, per region, and make them explain every discrepancy before you sign.
Named in this context: LiveRamp · Amperity · Acxiom
The "identity graph" is often a white-labeled third party — you may be paying a markup
One of the quieter surprises in the interviews. More than one buyer discovered mid-evaluation that a vendor's proprietary-sounding "identity graph" was a rebranded third party — a capability they could have licensed directly, without the intermediary's margin. In a category that sells the graph as the moat, it's worth asking the blunt question: whose data actually powers the match, and can we go to the source? Treat "it's our proprietary graph" as a claim to verify, not a given.
Named in this context: LiveRamp · Merkle · Epsilon
Deterministic vs. probabilistic is the accuracy-vs-scale tradeoff — know which you're buying
Buyers repeatedly bump into the same tension: deterministic matching on hard identifiers is precise but thin, while derived, probabilistic IDs extend reach but cost accuracy — and leaning on them for scale is a limitation buyers name directly. The rule they land on is to match the method to the use case: deterministic for measurement, suppression, and anything you'll defend to finance; probabilistic where scale matters more than precision. Most real stacks blend both, so ask any vendor exactly where that line sits in their graph, rather than accepting a single blended "match rate."
Named in this context: LiveRamp · TransUnion · Neustar
Pricing is usage-based and opaque — every pipe and onboard meters
The bill surprises buyers more than the product does. The category leader charges each time the data pipe is used, with additional fees for identity resolution on top — a structure buyers describe as "expensive and opaque," and a barrier that sends budget-conscious teams looking for alternatives. Because the cost scales with destinations, onboarding volume, and resolution, the sticker that fits today's usage doesn't fit your ambitions. Model the total against your real onboarding volume and destination count, and ask specifically how identity fees stack on top of the pipe, before you commit to an annual.
Named in this context: LiveRamp · Epsilon
Privacy compliance is the point now — buy the external-ID and clean-room model, not raw PII movement
The value proposition has shifted from "match more emails" to "collaborate on data without moving raw PII." Buyers in regulated industries specifically praise the external-ID approach — de-identified data and an ID for targeting instead of raw customer records — because it keeps them compliant, and a global pharmaceutical company cites exactly this as the reason the tool works for healthcare. As third-party cookies fade, compliant identity and data clean rooms are where collaboration is heading. Buy for that model, not for a pile of matched identifiers you'll struggle to use compliantly.
Named in this context: LiveRamp · InfoSum · Habu (clean rooms)
Don't buy standalone identity if your CDP already does it — or bolt one on when it can't hit your rates
Identity resolution overlaps the CDP, the data onboarder, and the ad platforms, so the first question isn't "which vendor" — it's "what do we already resolve?" Many buyers find their CDP or onboarder covers the core need. But the reverse happens too: one retailer bolted a dedicated identity vendor onto its stack precisely because its CDP couldn't hit the identification rates it needed for anonymous traffic. Quantify what your existing stack resolves today, isolate the specific shortfall — anonymous visitors, offline, cross-device, householding — and buy only that gap, not a second copy of a capability you already pay for.
Named in this context: Amperity · LiveRamp — see also the CDP lessons
Reporting and global coverage are the soft spots — pressure-test both before you commit
Even buyers happy with the core matching flag two recurring gaps: reporting that falls short of expectations, and uneven international availability. Identity vendors optimize for the match, not the dashboard, so if your team relies on the tool for reporting, confirm the depth is there — and if you operate globally, confirm coverage in your priority regions rather than assuming the domestic match rate travels. These are the checks buyers wish they'd run in the trial instead of discovering the limits in production.
Named in this context: LiveRamp
How buyers talk about the tools
The landscape is led by a clear marquee. LiveRamp is the most-rated identity and data-onboarding platform in the corpus by a wide margin — buyers praise its resolution, onboarding scale, and privacy-compliant external-ID model, and flag its usage-based pricing, uneven global match rates, and thin reporting. Acxiom and Amperity sit alongside it, the latter often as the CDP-native option buyers weigh against a standalone tool. Merkle and Epsilon anchor the agency-and-data end, with Epsilon drawing the softer marks. The through-line: buyers rate these tools on trust in the match and clarity of the meter, not on feature counts — which is exactly why the lessons above are about verification, not capability.
The stories behind the lessons
The counter-current: the match is rarely the whole problem
The buyers who trust their identity setup did the unglamorous work before the contract. They validated match rates against their own resolved identities instead of the vendor's slide; they asked whose graph was really doing the matching; they mapped the overlap with the CDP and onboarder and bought only the gap; they modeled the usage meter against real volume. The buyers still frustrated — a match rate that didn't travel, a graph they overpaid for, a bill that scaled with ambition — tend to have skipped the verification and bought the pitch. The lessons above are cheap diligence; unwinding an identity contract mid-flight is not.
What this means for your evaluation
Four checks before you sign. First, make match rate a measured test on your own file, per region, and score the discrepancy explanations — not the headline. Second, ask whose graph powers the match and whether you could license it directly, so you're not paying a markup on a rebranded third party. Third, map the overlap with your CDP, onboarder, and ad platforms, and buy only the specific gap. Fourth, model the usage meter — pipes, destinations, and identity fees — against your real volume, and buy the compliant, clean-room-ready model rather than raw identifier movement. For the identity question inside a CDP specifically, see what CDP buyers wish they'd known; for where the customer-data stack is heading, why companies rip out the CDP they have.
Common questions
What's a good match rate for identity resolution?
The one you verified against your own data, in your own regions. There's no universal "good" number — match rates vary by geography, by how much you depend on derived IDs for scale, and by how clean your input data is. Buyers warn that vendor-quoted rates didn't line up with their own internal identity numbers during validation, and that availability is uneven internationally. Before you sign, run the vendor's match against a known slice of your file, compare it to your resolved identities, and make them explain every discrepancy. Treat the headline rate on the slide as a marketing number, not a measured one.
Deterministic vs. probabilistic identity resolution — what's the difference?
Deterministic matching links identities on hard, known identifiers — a hashed email or login — so it's precise but only covers people you can directly key on. Probabilistic ("derived") matching infers a match from patterns like device, IP, and behavior — it extends reach but trades accuracy, and buyers note that leaning on derived IDs for scale is a real limitation. The rule buyers land on: deterministic for measurement, suppression, and anything you report to finance; probabilistic where scale beats precision, like prospecting. Most stacks blend both — ask any vendor exactly where the line sits in their graph.
Do I need identity resolution if I already have a CDP?
Often not as a separate purchase — but sometimes yes, to fill a gap. Identity resolution overlaps the CDP, the onboarder, and the ad platforms, so before buying a standalone tool, map what your stack already resolves. Many buyers find their CDP covers the core need. But the reverse happens: one retailer bolted a dedicated identity vendor on precisely because its CDP couldn't hit its identification rates for anonymous traffic. Buy the gap, not a second copy of a capability you already pay for — quantify what your CDP resolves, isolate the shortfall (anonymous traffic, offline, cross-device, householding), and buy only that.
How much does LiveRamp cost?
Buyers describe the pricing as usage-based and "expensive and opaque" — charged each time the data pipe is used, with additional fees for identity resolution itself, rather than a flat license. The cost scales with your ambition: more destinations, more onboarding, and more resolution all meter, and it's a barrier buyers on tighter budgets cite when they shop alternatives. If you're evaluating, model the total against your actual onboarding volume and destinations, ask how identity fees are charged on top of the pipe, and pressure-test the annual commitment against real usage. The platform is well-rated for what it does; the surprise is in the meter, not the match.
This is the aggregate. Your stack is specific.
Evaluating an identity-resolution or data-onboarding vendor right now? Do a 15-minute interview about your own data and use cases and get this personalized — what peers with your stack chose, where each tool's match rates and pricing bite, and which of these lessons apply to your shortlist.
Get my personalized briefNo password needed · your interview is anonymized before it ever informs a page like this one.