Signal vs. Noise: the Art of Adjudicating Conflicting Provider Data
The real issue isn’t how much provider data you have. It’s that no one is deciding which data to trust.
You have to choose from many options: your credentialing team pulls a provider’s address from one source. Your claims system has a different one. Your member directory has a third. All three came from legitimate, widely-used databases so it seems they cannot be wrong. But in reality, all three cannot be right.
As you can see from the example, this isn’t a data shortage problem. Health plans today have access to more provider information than ever — federal registries, state licensing boards, credentialing databases, claims history. But the problem is, these sources regularly contradict each other, and without a way to resolve those conflicts, bad information wins by default.
Why is my provider data inaccurate even though I'm pulling from multiple sources?
More sources means more disagreement, not more accuracy. Each database reflects a different moment in time and a different purpose. A federal registry captures what a provider reported at enrollment. A state licensing board reflects what was submitted at renewal. A credentialing database shows what the provider’s group submitted on their behalf.
The thing is, none of them talk to each other, and none of them are automatically updated when something changes. Now, the result: a single provider might have four different addresses across four systems, all technically “official,” none of them necessarily current. This is described in two words:
- The Noise: Discrepancies between these sources create “digital litter.”
- The Signal: The one accurate piece of information that actually leads a member to care.
Which provider data source is most reliable: NPPES, CAQH, state boards?
It depends on the field you are looking for which is exactly the problem. For Tax ID and NPI, federal enrollment data tends to be most reliable. For licensure status, state boards are the authoritative source. For practice location, claims history often reflects where a provider is actually seeing patients better than any self-reported registry.
No single source wins across every data point. Plans that treat all sources as equal end up with whichever record was updated most recently, but unfortunately that may have nothing to do with which one is accurate
Why do my provider records show different addresses depending on where I look?
Each source captures addresses for a different purpose. A federal billing registry might have the address where a provider receives payments. A state board has the address on file from their last license renewal. A national credentialing database has whatever the provider’s group submitted when they joined. These addresses can all be different, as well as all legitimate, for the same provider at the same time.
For a member trying to find care, only one of those addresses matters: the one where the provider is actually seeing patients. Getting that wrong means a member shows up at an empty office or a billing department, not an exam room.
Taking care of members is why you need a system that uses weighted scoring to decide which “signal” is the most trustworthy for a specific field.
What does conflicting provider data actually cost my plan?
If companies do not take care of these odd addresses, cost goes up in many different ways. Claims routed to the wrong address get rejected and reprocessed — adding administrative cost to every affected claim. Member complaints tied to directory errors require staff time to investigate and resolve. Plans with persistent directory inaccuracies face regulatory scrutiny under the No Surprises Act and state directory accuracy rules, where fines can reach $25,000 per violation per day in some states.
One analysis found that health plans spend an average of $2.50 to $5.00 in administrative cost for every claim that requires manual correction due to provider data errors. Across a large network, that adds up quickly and invisibly.
What is provider data adjudication, and how does it work?
Adjudication is the process of comparing conflicting data points across sources and deciding which one to trust, field by field, not record by record. A simple example: if four sources have four different addresses for the same provider, an adjudication system doesn’t just pick the most recent one. It weighs each source based on how reliable it tends to be for that specific type of information, how recently it was updated, and whether other signals (like recent claims activity) corroborate it.
The output is a single, resolved record that every downstream system can use with confidence.
Can you show me a real example of a data conflict being resolved?
Consider a provider who retired from private practice and joined a hospital system. Her old solo practice address still appears in a national credentialing database because the record was never updated. Her NPI registry entry shows a P.O. box used for billing. But her claims activity for the past 18 months all originate from the hospital’s main campus.
A system that adjudicates, will properly flag the credentialing address as stale (no corroborating signals), dismisses the billing P.O. box as a practice location, and confirms the hospital address based on consistent claims activity. The member directory shows the right place. The member finds her. Without adjudication, all three addresses compete equally, and the directory shows whichever one was entered last.
What should a PDM platform actually do?
Collecting data is the easy part, every platform does that. The differentiator is what happens when the data conflicts, which it always will. A strong provider data management platform should resolve conflicts automatically using logic that reflects how reliable each source actually is for each type of field. It should flag records where confidence is low rather than silently surfacing a guess. And it should update those resolutions continuously as new information comes in not just at the next monthly batch.
The goal isn’t the largest database. It’s the most trustworthy record for every provider, every time a downstream system asks for it.
See how ProviderLenz resolves conflicting provider data before it reaches your claims system or member directory. Explore Curatus.