Which claims passed every rule but paid wrong under the plan or contract?

QNXT
Facets
Snowflake
IBM ODM
SharePoint
SQL Server

Flume: Secure AI for healthcare. Knowledge you own.

2,000 databases31,000 schemas1,200,000 tables

30,000,000 columns6,100,000,000 claims4.2 PB of data

Answer · 3 of 3 checks passed

41,206 claims passed every rule and still paid above contract.

QNXT6.1 billion claim lines, read in place
Contracts2026 rates, amendment 4
IBM ODMRule PAY-114 applied the expired fee schedule

Open a ticket
Automate it
Build an app
Make a dashboard
Convert ODM rules to code

READ IN PLACE, READ-ONLY.TRAVERSEEVIDENCEACT

Learns your business. Then does the work.

Flume reads any data where it sits.

[ Customers ]

Health plans, TPAs, providers, and consulting firms build on Flume.

Read the impact studies

[ Studies → ]

Large payer01 / 03$13MYearly cost of the stack it retiredRead the impact study →

From your systems to the finished work.

Flume connects your systems, learns what the data means, checks every answer, governs access, and delivers the work.

The Flume platform

Works alongside your systems. They stay as they are.

ContextBase

Waking up…

Ask the hardest question.
See the evidence.
Everything gets smarter.

authorization_caseclaim_headerinquiry_caseplan_design

EVIDENCE SO FARQueries0Sources0Rows0Analyzed0Knowledge0

Overturned appeals still read as deniedDemo data · names are invented

Which denied authorizations still paid this plan year?

872 claims paid on a denied authorization this plan year, $2.0M.

1,284 claims matched a denied authorization. 412 had the denial overturned on appeal, so they come out of the count.

What I applied

The four endorsed pitfalls on these tables, including “Claims paid on an authorization that was denied”.

What I did not check

Expired authorizations still marked Approved. I can run that next.

in 212.4k · out 9.8k · cache read 640.2k

Save as knowledgeCopy

Asked: “Which denied authorizations still paid this plan year?”

Ask a follow-up · sends when this answer finishes

GraphPlain

Capture contextfrom this answer

CanonicalPitfallNoteCaveatRuleMethodology

ANCHORED TO (FROM THIS ANSWER)

authorization_caseinquiry_caseclaim_headerauth_status_history

DETAILS

Denials overturned on appeal still read as Denied in authorization_case. Exclude claims with an appeal_overturn row in inquiry_case, then report both numbers.

TITLE (optional)

Overturned appeals still read as denied

Team · recall on

FILE UNDER GROUPING (OPTIONAL)

Authorizations

WORKSPACE

Payment integrity

CancelSave

[ Explore more ]

See the map, then the work it powers.

ContextBaseThe map of how your systems really work.See how it works

WorkspaceContextBase, put to work.See the workspace

Getting an answer is easy.
Getting one you can use is the work.

TRADITIONAL6 people

TicketApprovalFind ownerWhich tableJoinsPipelineBuildAnswer13 WEEKS

TIME TO DECISIONCOSTUSABLE OUTPUTANSWER, A QUARTER LATE

GENERAL MODEL1 person

AnswerWrong answerRe-askWrong answerRe-askGive up→SECONDS

TIME TO DECISIONTOKENS BURNEDUSABLE OUTPUTNO ANSWER

FLUME1 person

AskResolveCheckDeployMINUTES

TIME TO DECISIONCOSTUSABLE OUTPUTCHECKED

Where healthcare data finally agrees.

Whether you run the plan, deliver the care, or advise either, you work from the same checked data.

For payers and TPAs

Answers from claims, eligibility, and accumulators where they live, checked and kept live plan year after plan year.

See Flume for payers and TPAs

claim_linemember_spanprior_authorization

DataPropertiesDiagram

prior_authorizationdemo_payer_dw › authDemo dataShowing 1-40 of 318,000 rows · 9 cols9/9

PA-26-0418331

97110

Approved

M-0048112

Therapy

2026-09-02

1487603921

PA-26-0418332

NULL

Pended

M-0120944

Imaging

2026-09-02

NULL

1932145870

PA-26-0418340

72148

Approved

M-0093310

Imaging

2026-09-02

2026-09-03

1356284470

PA-26-0418347

95810

Approved

M-0017785

Sleep study

2026-09-02

1609873345

PA-26-0418351

27447

Pended

M-0144029

Inpatient

2026-09-03

NULL

1740298816

PA-26-0418352

NULL

Pended

M-0144029

Inpatient

2026-09-03

NULL

1740298816

PA-26-0418366

70553

Denied

M-0061207

Imaging

2026-09-03

2026-09-04

1932145870

PA-26-0418370

97110

Approved

M-0102556

Therapy

2026-09-03

1487603921

PA-26-0418371

29881

Approved

M-0038841

Outpatient

2026-09-03

2026-09-04

1265530914

PA-26-0418385

NULL

Pended

M-0129930

DME

2026-09-04

NULL

1811047326

PA-26-0418390

43239

Approved

M-0075514

Outpatient

2026-09-04

1265530914

PA-26-0418392

95810

Approved

M-0011478

Sleep study

2026-09-04

1609873345

PA-26-0418404

72148

Denied

M-0150067

Imaging

2026-09-04

2026-09-05

1356284470

PA-26-0418411

97110

Approved

M-0088203

Therapy

2026-09-05

1487603921

PA-26-0418419

NULL

Pended

M-0042190

Imaging

2026-09-05

NULL

1932145870

PA-26-0418420

95811

Approved

M-0107735

Sleep study

2026-09-05

1609873345

PA-26-0418433

70553

Approved

M-0066514

Imaging

2026-09-05

2026-09-06

1356284470

PA-26-0418441

27447

Approved

M-0131872

Inpatient

2026-09-06

1740298816

PA-26-0418450

97110

Approved

M-0024419

Therapy

2026-09-06

1487603921

PA-26-0418458

NULL

Pended

M-0097763

DME

2026-09-06

NULL

1811047326

PA-26-0418462

29881

Approved

M-0056681

Outpatient

2026-09-06

2026-09-07

1265530914

PA-26-0418477

95810

Approved

M-0139012

Sleep study

2026-09-07

1609873345

SELECT CELLS FOR STATS

For providers

Revenue, cost, and care, answered from the systems health systems and provider groups already run.

See Flume for providers

Denials by Payer and Reason: 2026

Demo data

Live rev #5 · data from 3h · · built from Revenue cycle

Remit dateto

Claims denied: remit-date basis

$97.6Mdenied charges

8.8%denial rate

$38.8Moverturned on appeal

Denial rate by month

8.5%2026-08-0.2 pts vs 2026-07

0%3.75%7.5%11.25%15%

2026-012026-042026-07

month in progress

Top denial reasons

Eligibility not verified$26.4M

Prior auth missing$21.5M

Medical necessity$17.6M

Coding and bundling$12.7M

Timely filing$6.83M

Monthly detail

Month	Billed	Denied charges	Denial rate	Overturned
2026-01	$118.4M	$10.9M	9.2%	$4.10M
2026-02	$112.6M	$10.2M	9.1%	$3.90M
2026-03	$121.9M	$11.3M	9.3%	$4.40M
2026-04	$123.7M	$11.0M	8.9%	$4.30M
2026-05	$126.1M	$11.4M	9.0%	$4.60M
2026-06	$124.8M	$10.9M	8.7%	$4.40M
2026-07	$129.3M	$11.2M	8.7%	$4.70M
2026-08	$131.2M	$11.1M	8.5%	$4.80M
2026-09	$116.8M	$9.60M	8.2%	$3.60M

For consultants

Know what the client data supports in the first week, and bring the tools you built to the next engagement.

See Flume for consultants

Plan benefit clustering

Demo data

Groups plans with similar deductible, out-of-pocket, and coinsurance design. Filter by carrier and funding, then run.

CARRIERFUNDINGCLUSTERS (K)IN SCOPE417 plans

FEATURESInd. INN deductibleFam. INN deductibleInd. INN OOP maxFam. INN OOP maxInd. OON OOP maxCoinsurance

417Plans

5Carriers

120Employer groups

4Clusters

Cluster 1 is near-identical. 147 of its 148 plans, from 53 employer groups, sit within one $250 deductible step of the median. Consolidation candidate.

PLAN SIMILARITY MAP (PCA)

C1 · 148 plansC2C3C4PC1PC2

CLUSTERS

Cluster	Plans	Groups	Carrier mix	Consolidation
Cluster 1	148	53	A 71 · B 52 · C 25	High: near-identical
Cluster 2	121	43	A 64 · D 57	Medium: few designs
Cluster 3	96	35	B 40 · C 34 · E 22	Low: varied
Cluster 4	52	15	E 52	Low: varied

We will solve your hardest problems.

Choose a question with our team. Connect the data it needs. Review the answer and the work behind it.

Request a demo How we secure your data

Flume: Secure AI for healthcare. Knowledge you own.

Learns your business. Then does the work.

Health plans, TPAs, providers, and consulting firms build on Flume.

From your systems to the finished work.

Ask the hardest question. See the evidence. Everything gets smarter.

See the map, then the work it powers.

Getting an answer is easy. Getting one you can use is the work.

Where healthcare data finally agrees.

For payers and TPAs

For providers

For consultants

We will solve your hardest problems.

Ask the hardest question.
See the evidence.
Everything gets smarter.

Getting an answer is easy.
Getting one you can use is the work.