Prior authorization automation that approves without delay where criteria are clearly met, gathers the clinical documentation that consumes most of the elapsed time, and routes every case that could result in a denial to a qualified clinical reviewer.
Prior authorization is the most scrutinized workflow a health plan operates. It carries regulated turnaround requirements, sustained legislative attention, measurable provider abrasion and a member harm narrative that follows any error into public view. The risk is asymmetric: a wrongly automated approval costs money, and a wrongly automated denial costs considerably more than money. Everything on this page follows from that.
The fastest authorization is the one that was never required. The second fastest is the one approved automatically because the criteria were plainly met.
A clinical reviewer looking at a complete case with clear criteria usually decides quickly. The days that accumulate before that point are spent establishing what documentation is required, requesting it, waiting, receiving something incomplete, requesting again, and reassembling the case each time somebody picks it up. That is administrative work, it is most of the cycle, and none of it requires clinical judgement.
The same pattern produces the friction. Providers experience prior authorization as repeated requests for information they believe they already sent, and members experience it as delay with no visible reason. Both perceptions are usually accurate.
The determination is a fraction of the cycle
Reviewers decide quickly on complete cases. Time is consumed by assembling the case, and speeding up the review does not address it.
Pending additional information is not a status
The system should know what is missing, who can supply it, whether it was requested, whether anything returned, whether it matched the right case and how much time remains.
Volume nobody has reviewed
Requirements accumulate and services can remain on the list long after the evidence supporting review has changed.
Criteria applied inconsistently
The same case can reach different reviewers and different outcomes. That is a quality problem before it is a compliance one.
Turnaround requirements that differ by everything
Standard, expedited, line of business, service type and state can carry different clocks. Managing them in reporting rather than routing is how breaches happen.
Abrasion that never enters the business case
Provider effort, repeat contact, appeals generated and network relationship cost are real and almost never measured against programme savings.
The highest-value automation is deciding not to require the authorization.
Before automating a workflow, review whether it should exist. Services where approval rates are consistently high, where criteria are rarely the deciding factor, or where review has not changed an outcome in a meaningful number of cases are candidates for removal from the requirement list or provider exemption.
A request resolves in one of three ways. It does not require authorization, which the system can determine. It clearly meets criteria, which the system can approve. Or it does not clearly meet criteria, which is a clinical question and goes to a clinical reviewer with the case fully assembled. The third path is never automated regardless of how confident the system is.
| Step | What Happens | Automatable |
|---|---|---|
| 1. Intake and Classification | Request received across channels, classified by service, member, product and urgency, with the applicable clock started immediately. | Yes |
| 2. Requirement Check | Determine whether authorization is actually required for this member, product, service and provider, including any exemption. | Yes, and it removes work entirely. |
| 3. Completeness Assessment | Establish what documentation the criteria require and what is present, before anything is requested from the provider. | Yes |
| 4. Documentation Retrieval | Retrieve available clinical documentation from connected sources before asking the provider for it. | Yes, and this is where most of the elapsed time is. |
| 5. Criteria Application | Compare the assembled evidence against the applicable criteria and establish whether they are clearly met. | Yes, as an assessment. |
| 6a. Clear Approval | Where criteria are plainly met, approve, notify and update downstream systems. | Yes |
| 6b. Anything Else | Route to a qualified clinical reviewer with the case assembled, the criteria applied and the gaps identified. | Never Automated |
| 7. Notification and Downstream | Determination communicated to provider and member, and authorization written to the systems that will adjudicate the claim. | Yes |
Retrieve before you request
The most common avoidable delay is asking a provider for documentation the plan could have retrieved itself. Before any information request is issued, the system should exhaust what is available through connected sources, prior submissions on the same member, and documentation already attached to related cases.
The capabilities below are deliberately administrative. Each removes work from a reviewer, a provider or both, and none of them makes a clinical decision. That boundary is the design, not a limitation of the technology.
Reduce and Intake
Intake, Requirement and Exemption Determination
Normalize requests from portal, API, fax, phone and electronic transactions into one case with the clock started at receipt, then test whether authorization is genuinely required for this member, product, service and provider.
Clinical Document Intelligence
Extract and interpret clinical records, notes, imaging reports and supporting documentation, with every extracted value traceable to its source page for reviewer verification.
Completeness Assessment
Determine what the applicable criteria require and what has been supplied, so a single specific request is made rather than a sequence of partial ones.
Assemble and Assess
Documentation Retrieval
Gather available clinical evidence from connected sources, prior submissions and related cases before any request reaches the provider.
Criteria Application
Compare assembled evidence against the applicable criteria and present which are met, which are not, and which cannot be established from the available documentation.
Case Chronology
Organize submissions, requests, responses and prior authorization activity into a usable timeline.
Case Assembly for Review
The reviewer receives the request, evidence, criteria applied, gaps identified and relevant history, so review time is spent on judgement rather than assembly.
Auto-Approval Within Configured Criteria
Approval issued without delay where criteria are clearly and objectively met, within limits you configure. Approval is the only determination the system issues.
Manage and Close
Turnaround Management
Clocks applied per line of business, case type and urgency, surfaced in prioritization and escalated before a deadline is at risk.
Missing Information Workflow
The specific gap identified, outreach initiated and tracked, returned information matched back to the case, completeness reassessed and unresolved items escalated, with the clock visible throughout.
Provider Communication
Status visibility, single consolidated information requests and determination notification designed to reduce enquiry volume.
Downstream Write-Back
Authorization recorded in the systems that will adjudicate the claim.
Pattern and Volume Analytics
Approval rates by service, criteria that rarely change an outcome, providers with consistently high approval rates, and requirements worth removing entirely.
What CaliberFocus does, and does not do?
We do not build systems that issue denials, and this is not a configurable setting. Adverse determinations are made by qualified clinical reviewers acting under their own authority, with the case assembled and the criteria applied for them. We will also recommend removing services from your authorization requirement list where the data supports it.
The best starting point is a service category where the criteria are objective, the documentation is retrievable and most requests are approved anyway. Those are the cases where automation removes real work and carries almost no clinical risk, and they are frequently also the cases that should be reviewed for removal from the requirement list entirely.
| Category | Why It Fits or Does Not | Where to Start |
|---|---|---|
| High-Approval Routine Services | Objective criteria, retrievable documentation, most requests approved. | Best starting point, and the first candidates for removing the requirement altogether. |
| Imaging and Diagnostics | Criteria are usually explicit and evidence is structured. | Strong. Auto-approval where criteria are met, clinical review for the rest. |
| Elective Procedures | Criteria explicit, documentation substantial but retrievable. | Strong for retrieval and assembly. Determinations remain clinical. |
| Specialty Drugs | Complex criteria, step therapy history, frequent policy change. | Assembly and criteria application help considerably. Approval automation only where criteria are unambiguous. |
| Behavioural Health | Sensitive, criteria involve judgement, documentation is confidentiality-constrained. | Administrative support only. Handle documentation under stricter rules. |
| Urgent and Expedited Requests | Compressed clocks, higher consequence of delay. | Prioritization and retrieval matter most. The shorter clock raises the cost of any error. |
| Anything Likely to Result in Denial | The consequence of being wrong is a member harm event. | Clinical reviewer, always, with the case assembled for them. |
Gold Carding Is an Automation StrategyGold Carding Is an Automation Strategy
A provider whose requests are approved at a consistently high rate over meaningful volume is generating administrative cost on both sides for no decision value. Exempting them, or applying a lighter pathway, removes the work rather than accelerating it and improves a network relationship at no clinical cost.
A wrongly automated approval costs the plan money and is recoverable. A wrongly automated denial delays or prevents care, generates an appeal, damages a provider relationship and may become a regulatory matter or a public one. Those two errors are not comparable, and a design that treats them as a single accuracy problem has misunderstood the workflow.
Confidence does not create authority. How certain the system is and whether the system is permitted to decide are different questions, and the second one is settled first.
| Outcome | Who Decides | What the System Contributes |
|---|---|---|
| Not Required | System, against configured rules | Determines the service does not require authorization for this member and provider. |
| Approved, Criteria Clearly Met | System, within configured limits | Applies objective criteria to assembled evidence and issues the approval. |
| Approved, Judgement Involved | Clinical Reviewer | Assembles the case, applies criteria, identifies gaps, recommends. |
| Additional Information Needed | System may request, reviewer may direct | Determines precisely what is missing, having exhausted retrieval first. |
| Denied or Partially Denied | Qualified Clinical Reviewer Only | Assembles evidence and applies criteria. Issues nothing. |
| Appeal Outcome | Authorized Reviewer under Appeal Authority | Prepares the case and tracks the statutory clock. Issues nothing. |
Auto-approval limits you set
Which services, criteria, evidence standards and financial thresholds qualify for automated approval are configured by clinical leadership.
Criteria applied, not interpreted
The system establishes whether objective criteria are met. Where interpretation of clinical context is required, that is reviewer work
The reviewer sees the evidence
The system establishes whether objective criteria are met. Where interpretation of clinical context is required, that is reviewer work
Override capture as quality signal
Reviewer disagreement is recorded and analysed. Consistent patterns indicate criteria or extraction problems.
The review point is designed
Define who receives the case, what they see, what they can do, what escalates, how disagreement is recorded and what happens next.
Clinical governance owns the rules
Criteria configuration, auto-approval scope and escalation policy sit with clinical leadership, with change control and review cadence.
No confidence score makes a denial automatable.Collide
This is not a threshold that could be set high enough. An adverse determination affecting access to care carries authority that attaches to a qualified reviewer acting under their own licence and judgement. A system can assemble the case perfectly, apply the criteria correctly and be right about the outcome, and it still cannot be the one that issues it.
Every capability on this page that removes real time depends on retrieving clinical documentation without asking the provider for it. Where that access does not exist, the automation reduces to faster classification and better queue management, which is worth having and is not the value being described.
Utilization management platform
Case creation, status, criteria, review workflow and determination recording, with write authority for approvals within configured scope.
Core administration platform
Member, benefit, product, provider and authorization data, plus write-back so an approved service does not later deny for a missing authorization.
Clinical documentation sources
Provider systems, health information exchange, prior submissions and attached documents. The single most valuable connection on this page and usually the hardest to obtain.
Provider channels
Portal, API and electronic transaction intake, plus fax and correspondence, normalized into one case regardless of how it arrived.
Criteria and policy sources
Medical policy and criteria content maintained as versioned assets, since a criteria change alters what the system may auto-approve.
Claims and downstream systems
Authorization written where adjudication will look for it, with the linkage verified rather than assumed.
Retrieve before requesting, always
Exhaust connected sources, prior submissions and related cases before any information request reaches a provider.
Confirm the downstream state
An approval is complete when the authorization exists in the system that will adjudicate the claim.
Provenance retained through extraction
Every value used in a criteria assessment is traceable to the document and page it came from.
Integration failure becomes visible workflow work
A document that cannot be retrieved or an authorization that cannot be written creates a visible exception with owner, attempts and time remaining.
Sensitive documentation handled under stricter rules
Behavioural health, substance use and other protected categories are segmented at ingestion rather than after assembly.
Turnaround requirements vary by line of business, case type and urgency, and they run regardless of whether documentation has arrived or a reviewer is available. Managing them in a report is managing them after the fact. They belong in the routing logic, visible on every case from the moment it is received.
Turnaround management
- Per workflow: applicable request category, timing rule, start event, current state, time consumed, time remaining, escalation threshold, owner and completion event, determined at intake and visible throughout
- Prioritization driven by time remaining rather than by receipt order, with escalation before a deadline is at risk
- A manual fallback for any component whose failure would put a regulated deadline at risk
Clinical governance
- Policy separated from workflow execution. Timing requirements, criteria and rules held as configurable, versioned, approved, effective-dated and auditable assets, so a requirement change is a configuration release rather than a development project
- Auto-approval scope defined and reviewed by clinical leadership, with change control and a stated review cadence
- Consistency of outcome measured across reviewers and over time, as a quality programme rather than a report
- Adverse determinations attributable to a named qualified reviewer acting under their own authority
Auditability
- For any case: when the request arrived, what came with it, what was missing, what outreach occurred, what returned, what evidence was presented for review, who determined it, when, what communication was issued, whether downstream systems updated, and whether it completed within the applicable requirement. That is an audit trail. A list of system events is not
- Every extracted clinical value traceable to its source document, so an assessment can be verified rather than accepted
- Information requests logged with what was asked, why, and what had already been retrieved before asking
- Override and escalation patterns retained and analysed, since they are the first indicator of a criteria or extraction problem
Operational control
- Auto-approval can be paused, narrowed by service, stepped down to act-on-approval or draft, or routed entirely to review when policy changes, accuracy deteriorates or an integration destabilizes, without disabling unrelated workflows
- Six named owners per workflow: business, clinical, operational, technical, data and compliance oversight
- Monitoring of outcome rather than execution, since a case can process correctly and still reach the wrong state
- Provider abrasion tracked as an operational measure: information requests per case, repeat contact and appeal volume
Measure abrasion or you will only see the savings.
The business case for authorization automation counts reviewer time and cycle time. The cost sits in provider effort, repeat information requests, appeals generated and network relationships, none of which appear in that calculation. A programme that halves internal cost while doubling provider friction has produced a result the plan will pay for later, in contracting and in reputation. Track it from the first workflow.
Authorization programmes are usually reported on turnaround time and cases processed. Both improve without telling you whether the work was necessary or whether providers experienced anything different.
| Category | What We Measure | Why It Matters |
|---|---|---|
| Volume Removed | Requests eliminated by requirement review and provider exemption | The only outcome that reduces cost on both sides permanently. |
| Elapsed Time | Time to determination, and specifically time spent awaiting documentation | The second number is where the cycle actually goes. |
| Retrieval Effectiveness | Documentation obtained without a provider request, and information requests per case | The single largest lever on both cycle time and abrasion. |
| Reviewer Time | Time per case spent on judgement versus on assembly | The clinical capacity measure, and the one reviewers will judge it on. |
| Timeline Compliance | Turnaround performance by case type, and cases identified at risk before breach | The exposure, and why the clock belongs in routing. |
| Abrasion | Repeat contact, appeal volume from authorization decisions, provider enquiry volume | The cost that appears in contracting rather than in operations. |
| Consistency | Similar cases reaching similar outcomes, and override rate on assembled recommendations | The quality measure most likely to be examined. |
Two findings are likely and both are uncomfortable.
Some services on your requirement list will not survive an honest review of whether the authorization changes an outcome. And clinical documentation access, not automation capability, will be the constraint on how much time you can remove. Both are better established before the programme is scoped than after it has been sold internally on a different premise.
We will analyse that category: how often the review changes the outcome, how much of the cycle is spent awaiting documentation, how much of that documentation was retrievable, and what auto-approval scope the criteria would objectively support. In several categories the honest recommendation is to reduce the requirement rather than automate the process, and that finding is worth more than the automation would have been.
Bring us a specialty or clinical setting where clinicians are spending too much time creating notes. We will assess where ambient documentation fits, what must remain clinician controlled, how it should integrate with your EHR, and how to measure whether it is actually reducing burden.
One conversation with people who have run these deployments, and a written readiness view you can use with or without us.
- AI Agents and Workflow Automation
- Voice and Conversational AI
- Document AI and Intelligent Processing
- Generative AI and Enterprise Copilots
- AI Strategy and Governance
- HCC and Risk Adjustment Analytics
Security & Compliance
