Grade Your Patient Booking Flow

What a Scheduling Decision Tree Is

When a patient opens your MyChart portal to book an appointment, they don’t land directly on a calendar. They work through a decision tree: a sequence of yes/no questions and routing logic that determines which provider, visit type, and time slot they’re eligible to see.

The decision tree might ask whether the patient is new or returning, which symptom or condition brings them in, whether they need an in-person visit or can be seen via telehealth, and which location is convenient. Each answer routes them down a different branch until they either reach a bookable slot — or hit a dead end.

For most health systems, this tree was built organically: specialty by specialty, department by department, as MyChart configurations accreted over years. No one designed it end-to-end as a patient experience. The result is a flow that works fine for straightforward cases and falls apart for anyone slightly outside the expected path.

The Three Things We Score

Luma evaluates each specialty’s scheduling decision tree across three independent dimensions. Together they capture the full patient experience of the booking flow.

AI Score (0–100): Clarity & Structure

An AI model reads the full decision tree for a given specialty and assigns a score from 0 to 100 based on overall quality. It evaluates whether questions are clearly worded and unambiguous, whether the tree’s branching logic is coherent, whether the flow feels intuitive to a first-time patient, and whether the structure reflects what the specialty actually needs to route appropriately. A score near 100 means the tree is clean, purposeful, and easy to navigate. A score near 0 means a patient reading the questions would routinely be confused about what to answer.

Dead-End Count: Paths That Refuse Online Booking

A dead end is any path through the decision tree that ends with a message along the lines of “we can’t schedule you online — please call the office.” Dead ends are not always wrong: some clinical situations genuinely require a phone call to triage appropriately. But they are a hard stop for online conversion, and each one represents a patient who tried to self-schedule and could not.

Luma counts the total number of dead ends in each specialty’s tree. A tree with many dead ends is one that a significant fraction of patients will abandon in favor of calling, waiting on hold, or simply giving up.

Friction Count: Obstacles That Slow or Confuse

Friction is a broader category than dead ends. It captures ambiguous or poorly worded questions that force patients to guess, excessive depth (trees requiring more than four levels of answers before reaching a bookable slot), and branch imbalance (some paths requiring far more questions than others, without clinical justification). High friction doesn’t necessarily prevent booking — it just makes it harder than it needs to be, increasing the chance that a patient gives up or selects the wrong path.

One Grade Per Specialty

The three raw scores roll up into a single composite grade for each specialty, expressed as a letter grade from A through F. The weighting reflects how much each dimension actually affects a patient’s ability to complete an online booking:

AI quality carries the most weight because it reflects the holistic patient experience of the tree. Dead-ends receive a heavy penalty because each one is a guaranteed conversion loss. Friction is real but recoverable — it slows patients down without always stopping them.

Benchmarked Against Peers

A grade only tells you so much on its own. A B might be excellent if every competitor in your market is at C or D — or it might mean you’re falling behind if your top peer just hit an A. Luma puts each specialty grade in context by comparing it to your cohort.

For every specialty, the platform calculates your cohort rank: where you sit relative to peers operating in the same market. It also surfaces a top-peer delta — the precise gap between your composite score and the highest-scoring competitor on that specialty. If you’re 5 points behind your top peer on Primary Care, you know exactly how far the leader is and which specialty to prioritize.

The benchmarking extends to dead-end ratios as well. Your dead-end count is compared against the cohort median so you can see whether your number of blocked paths is typical for your market or an outlier in either direction. A dead-end ratio well above the median flags a tree that is pushing more patients to the phone than your peers require.

These comparisons turn a grade from an internal report into a competitive signal — actionable intelligence about whether your booking experience is winning or losing patients at the first touchpoint.

Why It Matters

Friction and dead-ends are silent conversion killers. A patient who encounters an ambiguous question might answer incorrectly and land in the wrong branch. A patient who hits a dead-end might call — or they might open a competitor’s portal instead. Neither event leaves a trace in your analytics. You never see the abandonment. You only see that the appointment didn’t happen.

This is why the scheduling experience has a direct line to campaign performance. Every dollar spent bringing a patient to your scheduling page is wasted if the page itself turns them away. A campaign that drives high-intent traffic into a D-grade decision tree is not performing as badly as it looks — it is performing exactly as badly as the tree allows.

Fixing a dead-end or restructuring a confusing branch lifts the ceiling on every downstream investment. It improves organic conversion, paid search conversion, and the return on any referral or recall program built on the same scheduling flow. The decision tree is not a technology configuration issue — it is a revenue issue, and treating it that way changes what gets prioritized.