A health plan executive recently told me something that stopped me cold: “We’re really proud of our retrospective review process. We’ve been doing it the same way since 2019, and it’s proven.”
Same way since 2019. In an industry where regulations shift monthly and technology advances weekly. That’s not proven. That’s fossilized.
The brutal truth about retrospective risk adjustment today? If you’re not evolving quarterly, you’re falling behind daily. And the gap between leaders and laggards isn’t measured in percentage points anymore; it’s measured in generations of capability.
The 2019 Playbook Is Dead
Remember what worked in 2019? Annual batch reviews. Offshore coding farms. Excel-based tracking. Keyword searching through PDFs. Manual MEAT validation. These weren’t just standard practices, they were best practices.
Now? Annual reviews mean leaving money on the table for 11 months. Offshore coding without AI pre-processing wastes 70% of human effort on mechanical tasks. Excel tracking can’t handle the complexity of modern documentation requirements. And keyword searching misses the clinical nuance that determines whether codes survive audits.
But here’s what really changed: the risk profile of staying static. In 2019, using outdated methods meant you were slightly less efficient. Today, it means you’re systematically vulnerable. CMS uses machine learning to identify audit targets. Competitors use AI to find codes you miss. The technology gap has become an existential threat.
I visited a mid-sized plan last month still using their 2019 processes. Their coding accuracy was 94%, which sounds good. Their competitor down the street? 98.4% accuracy with 3x the throughput. That’s not a small efficiency difference. That’s a different sport entirely.
The Continuous Evolution Imperative
The organizations winning at retrospective risk adjustment treat their processes like software: constantly updating, regularly deploying improvements, always beta testing something new. They don’t have an annual process review. They have monthly optimization sprints.
Take encounter prioritization. The 2019 approach was simple: review high-dollar members first. Today’s leaders use predictive models that factor in documentation availability, provider patterns, condition complexity, audit risk, and seasonal timing. They’re not just prioritizing; they’re optimizing across six dimensions simultaneously.
Or consider validation methods. Five years ago, checking MEAT criteria meant a human reading through notes. Now, advanced operations pre-validate using AI, flag specific concerns for human review, and maintain audit trails that prove systematic compliance. The human role shifted from checking to confirming, from searching to strategizing.
The feedback loops have compressed from annual to weekly. Smart plans know within seven days if a process change improved outcomes. They test new approaches on small populations before full deployment. They fail fast on bad ideas and scale quickly on good ones. This isn’t reckless experimentation; it’s systematic evolution.
The Technology Multiplication Effect
Here’s what the 2019-minded organizations don’t grasp: modern technology doesn’t just automate old processes, it enables entirely new capabilities. It’s not about doing the same things faster. It’s about doing things that were previously impossible.
Real-time documentation quality scoring, for instance. As providers create notes, AI evaluates completeness for risk adjustment purposes and suggests improvements before the encounter closes. This isn’t faster retrospective review; it’s eliminating the need for retrospective correction entirely.
Cross-encounter pattern recognition represents another leap. AI can now identify diagnosis patterns across a member’s entire history, spotting conditions that appear intermittently but consistently. A human reviewing individual encounters would never catch these patterns. The technology isn’t replacing human judgment; it’s revealing information humans couldn’t see.
Predictive audit modeling has become table stakes. Before submitting codes, leading organizations run them through models trained on thousands of actual audit outcomes. They know statistical probability of validation before CMS does. This isn’t hoping documentation holds up; it’s knowing it will.
The Organizational Metabolism Shift
The biggest barrier to evolution isn’t technology or budget. It’s organizational metabolism: how quickly you can absorb and implement change. Most health plans still operate on annual planning cycles in a world that demands monthly adaptation.
Breaking free requires fundamental shifts. First, abandon the big bang transformation mindset. You don’t need a massive three-year initiative. You need continuous small improvements that compound. Think evolution, not revolution.
Second, push decision authority down. If every process change requires committee approval, you’ll never keep pace. Give front-line teams authority to test improvements within defined parameters. Let them iterate without permission as long as they measure and share results.
Third, celebrate intelligent failures. Not every improvement works. That’s fine. What’s not fine is not trying anything new for fear of failure. The teams that evolve fastest are those that view failed experiments as valuable data, not career risks.
Your Monday Morning Test
Here’s a simple test for Monday morning. Pull up your retrospective risk adjustment process documentation. Check the last modified date. If it’s more than 90 days old, you’re already behind. If it’s more than a year old, you’re in crisis whether you know it or not.
Next, list five things you’re doing exactly the same as in 2019. Now list five things your most successful competitor is probably doing that you’re not. The gap between those lists? That’s your competitive disadvantage, growing wider every day you don’t evolve.
The comfortable stability of proven processes is actually dangerous stagnation in disguise. Your 2019 playbook might feel safe, but it’s making you systematically vulnerable to organizations that treat retrospective risk adjustment as a dynamic capability requiring constant evolution. The question isn’t whether to evolve, it’s whether you’ll do it voluntarily now or be forced to later, after competitors have taken your market share.
