Two Practices, Same Denial Rate, Different Futures
Two practices run a 9% denial rate. On a spreadsheet they look identical, and a consultant comparing them would call it a tie. Walk into each one and they are opposite operations.
The first has a denials team. Three people, good at their jobs, working a queue of returned claims: reading remittance codes, pulling records, writing appeals, resubmitting corrections, sitting on hold with payers. When denials rise, the practice adds a fourth person. The team is busy, the work is real, and everyone can see it happening, which is why leadership feels productive about it.
The second practice has almost no one working denials, because there are far fewer to work. Not because their payers are kinder. Because most of the claims that would have bounced never got submitted with the defect in the first place. Eligibility got checked the day before the visit. The claim got scrubbed against the payer’s rules the moment the charge was entered. The authorization got flagged three weeks before it expired. The defect that would have become a denial got caught while it was still a two-minute fix, by a process, before the claim ever left the building.
Same denial rate today. But the first practice’s rate holds flat no matter how many people it hires, and the second practice’s rate keeps falling while its denials team keeps shrinking. The difference isn’t effort or talent. It’s when they touch the problem. One works denials after the claim comes back. The other prevents them before the claim goes out. That single choice, where in the timeline you put the work, decides almost everything about what denial management costs you and whether it ever gets better.
The Reactive Model Is the Default, and It’s Backwards
Almost every practice runs denials the same way, and it wasn’t a decision. It’s just where the work naturally piles up if nobody moves it. A claim goes out. Weeks later, some of them come back refused. Someone has to deal with the refused ones, so a person, then a team, forms around the returns. That team becomes the denial management function, and denial management comes to mean the work of handling denials that already happened.
Read that again, because the trap is hidden in the definition. The entire function is built around a claim that has already failed. Every hour it spends is spent after the money is already late, after the visit has gone cold, after the appeal clock has started ticking. It’s a cleanup crew standing downstream of a process that keeps breaking claims, and the crew’s size is set by how many broken ones arrive. Work harder and you clean faster. You never clean less, because you’re not touching the thing making the mess.
This is why the most common denial strategy, hire another biller, so reliably disappoints. Another biller makes you faster at cleanup. It does nothing to the rate of defects, so within a quarter the new person is as buried as the old ones, and the denial rate sits exactly where it was, now costing one more salary to maintain. You didn’t fix the leak. You bought a bigger bucket.
Shift Left: Move the Work to Where the Defect Is Born
Borrow an idea from how software gets built. Engineers learned decades ago that a bug caught the moment a developer types it costs almost nothing to fix, and the same bug caught after it ships costs a fortune, because now it’s tangled into everything downstream and customers are already hit. So they moved testing as early as possible, to the left end of the timeline, closer to where defects are born. They called it shifting left, and it quietly changed the economics of the entire industry.
A denied claim is a bug that shipped. It’s a defect that made it all the way to the payer before anyone caught it, and now it costs the full price: rework, appeal, delay, and the real risk it never gets paid at all. Shift-left denial management applies the same move to the revenue cycle. Stop waiting for the payer to find your defects and hand them back. Catch them yourself, at the moment they’re created, before the claim is ever submitted.
Concretely, that means the checks move from after submission to before it. Eligibility is verified on a schedule ahead of the visit, not discovered wrong when the claim bounces. The authorization is watched like a runway with days remaining, not found expired in a remittance. Coding is bounded by written rules at entry, not corrected after a payer flags the pattern. Every claim passes an automated scrub against known payer rules before it leaves, so the defect the payer would have caught gets caught by you first, while the visit is fresh and the fix is trivial. The denial that never gets manufactured takes zero hours to work, because it doesn’t exist.
The Economics Only Work in One Direction
Here is the arithmetic that makes this more than a philosophy. The cost of a defect is not fixed. It grows the longer the defect survives, and it grows steeply.
A coverage problem caught two days before the visit is a phone call to the patient. The same problem caught after submission is a denial, plus an appeal, plus a delay of weeks, plus a patient who now owes money they didn’t expect, plus the staff hours to manage all of it, plus a real chance the balance is never collected. Same defect. One version costs five minutes. The other costs a small pile of money and three people’s afternoons. Nothing about the defect changed except how long it was allowed to live before someone caught it.
Multiply that across every claim, every month, and the two models diverge hard. The reactive practice pays the full, grown-up cost of every defect, because it only ever catches them fully grown. The shift-left practice pays the tiny newborn cost, because it catches them at birth. This is why the reactive denial rate never really moves while the proactive one keeps falling. It isn’t that one team tries harder. It’s that catching defects late is structurally the most expensive way to run a revenue cycle, and no amount of skill at cleanup can beat not making the mess. The payer AI teardown covers a second reason late is now a losing game: the machine on the other side got faster at finding your defects, so the after-the-fact race is one you increasingly can’t win. That full picture is in the payer AI teardown.
What This Does to Your Best People
The part that doesn’t show up in the arithmetic, and matters just as much: what the reactive model does to the humans inside it.
In a reactive shop, your best billing people spend their days as a rework department. They are good, experienced, capable of real judgment, and you have them fixing problems that were baked in weeks earlier by someone upstream they can’t control and often can’t even identify. It’s repetitive, it’s thankless, and it’s demoralizing in a specific way: they’re always cleaning, never building, always a step behind a process that keeps handing them defects. Good people burn out on that, and when they leave they take years of payer-specific knowledge with them.
Shift-left changes what the job is. When automation catches the routine defects before submission, the fifteen percent of claims that reach a human are the ones that actually need a human: the genuinely complex appeal, the payer being wrong, the case worth fighting. Your experienced biller stops being a cleanup crew and becomes what you hired her to be, someone applying judgment to hard problems instead of drowning in easy ones a machine should have caught. The team gets smaller and the work gets better at the same time, which is the opposite of what usually happens when you make a department more efficient. That reframe, humans on judgment and machines on the repeatable core, is the same principle laid out in the 80% standard.
The Operating Model, in Four Layers
Shift-left is not a single tool you buy. It’s a stance, built out of four layers of work that already exist as their own disciplines, arranged so that the whole thing points forward instead of back. This piece is the stance. The four legs underneath it are where the building actually happens.
The first layer prevents the defect at the source: the five places denials are manufactured upstream, each one closable before a claim ever carries it. That’s the five factories, and it’s where most of the rate lives.
The second layer works the ones that still get through, because no prevention is perfect, and the ones that slip need to be handled by cause rather than by date, with named owners and a deliberate exit. That’s the owned list, and in a shift-left practice it’s a short list instead of a queue.
The third layer measures the whole thing honestly, because a practice that shifts left needs to see the rate actually falling, counted the same way every month with no flattering exclusions. That’s the real denial rate, and it’s how you prove the stance is working instead of just feeling productive.
The fourth layer is rhythm: the daily and weekly cadence that keeps every defect younger than a day and ships one upstream fix a week, so prevention compounds instead of decaying. That’s the clean claim routine, and it’s what makes the other three permanent instead of a one-time cleanup that slides back.
Prevention, handling, measurement, rhythm. The reactive practice has only the second layer, scaled up, and wonders why the rate won’t move. The shift-left practice runs all four, pointed forward, and watches the rate fall while the team that used to fight denials shrinks.
See Which Model You’re Running
You already know which practice you are, but two numbers make it concrete. First: of the hours your team spends on denials this month, what share is spent after the claim came back versus before it went out? In most practices the honest answer is nearly all of it after, which is the reactive model measured. Second: over the last year, did your denials team grow, and did your denial rate fall? If the team grew and the rate held flat, you have proof that scaling cleanup doesn’t fix the source, and you have it in your own numbers.
That gap, between the work you do after and the work you could do before, is the whole opportunity, and it’s the difference between a denials cost that grows with your practice and one that shrinks as you build. The four legs above are how the building gets done. This is the stance that points them the right way.
Where to Start
Pick the single biggest denial cause from last month and ask one question: where upstream could this have been caught before the claim went out? That question, asked once, is the whole shift in miniature, and it usually names your first build. Then grab 30 minutes with us. Prep nothing. We’ll show you what a shift-left denial operation looks like running in real operations, the checks firing before submission and the short list of what still gets through, and you’ll see the gap between preventing denials and cleaning them up priced against your own rate.