Review Queue Staffing: How Many Reviewers You Need

Review queue staffing comes down to one calculation: fields routed to review each day, times the seconds a reviewer spends on each field, divided by the productive review time one person has in a day. Plan that number for your busy days rather than your average day, then check it against a turnaround target and a few accuracy metrics.

How many reviewers does a review queue need?

A review queue needs enough reviewer hours to clear every routed field within its turnaround target. Routed fields are the extracted values that scored below your confidence threshold and wait for a person to approve or correct them. Headcount follows from four inputs: documents per day, fields per document, the share of fields routed, and seconds per field.

The arithmetic is short. Multiply documents by fields to get fields read per day. Multiply that by the routed share to get fields sent to review. Multiply by seconds per field and divide by 3,600 to get reviewer hours. Divide those hours by the productive review hours one person gives the queue each day, then round up.

A worked staffing calculation for an example clinic

Every number below is fictional and chosen to show the method. None of them is a benchmark, and your own figures will differ.

The rounding step matters. Two reviewers would give the queue 12 hours, half an hour short every day, and that shortfall piles up. The third person also covers a sick day or a vacation without the queue stalling.

The routed share is the input teams most often guess. It depends on your threshold, your documents and your scan quality, so measure it on a sample of your own pages. How a threshold turns into routed volume is worked through in the guide on how much review volume a 90% extraction threshold creates.

Seconds per field is the second input to measure, not assume. Time reviewers on real queue items for a few days, and keep separate figures for short fields such as dates and long ones such as a service line with several amounts. A trainee will be slower than an experienced reviewer, so recheck the figure after the team settles in.

Productive review hours deserve the same honesty. Few reviewers spend a full shift inside the queue. Phone calls, team meetings, payer portals and training all take time, and a plan that counts eight queue hours per person will be short every day. Ask reviewers to log their queue time for two weeks, then use the median day, not the best one.

Fields per document also varies more than teams expect. A one-page intake form carries far fewer fields than a multi-page EOB where every service line adds its own date, code and amounts. Calculate each document type separately and add the hours together, instead of using one blended average that hides a heavy type behind several light ones.

What turnaround targets should a review queue meet?

Turnaround targets for a review queue set how long a routed field may wait before a reviewer clears it. A target is an internal promise, set by the team that depends on the data. Billing work and clinical work have different deadlines, so most teams set one target per document type instead of one for the whole queue.

Start from what happens downstream. An EOB field waits on payment posting, so a target such as "cleared the same business day" keeps deposits and postings in step. An intake form waits on a visit, so the target might be "cleared before the patient's appointment". A lab report re-keyed into an EHR waits on a clinician, so ask the clinical team what delay they will accept and set a shorter target there. These are example targets, not rules.

Outside deadlines sit far beyond any sensible queue target, but they set the hard limit. For Medicare, 42 CFR 424.44 on time limits for filing claims says a claim for services furnished on or after January 1, 2010 must be filed no later than one calendar year after the date of service. Other payers set their own limits in their contracts. A review queue that is measured in days has failed long before it threatens a filing limit.

Measure turnaround as the share of fields cleared within target, plus the age of the oldest field in the queue. The share tells you how the week went. The oldest item tells you, right now, whether something is stuck, such as a document type no one on the current shift is trained to review.

Order the queue to match the targets. When reviewers pick the next item, the field closest to missing its target should come first, not the oldest document overall or the easiest field. A queue sorted only by arrival time lets a same-day EOB wait behind a batch of scanned records that had a week to spare.

Reviewer productivity metrics that do not reward speed over care

Reviewer productivity metrics should show whether the queue is cleared on time and cleared correctly. Track throughput together with accuracy, and never rank people by speed alone. A reviewer who approves every proposed value without looking will top any speed chart while letting misread fields into the billing system.

A short set of metrics covers most needs:

• Fields cleared per productive hour, per reviewer and per document type, so a slow week on handwritten intake forms is not read as a slow reviewer.
• Error rate found in a quality sample of that reviewer's approvals and corrections, checked by a second person.
• Share of low-confidence fields approved without any change, watched as a trend. A sudden rise can mean rubber-stamping.
• Share of fields cleared within the turnaround target, for the team as a whole.

Metrics to avoid are the ones that reward haste. Fields per hour on its own pushes people to skim. Correction count as a score punishes the reviewer who catches the most mistakes. A systematic review of automation bias in JAMIA lists workload and time constraint among the factors that push people to over-rely on automated output, and lists training and an emphasis on user accountability among the mitigators.

The quality sample and the correction log carry most of this work. The guide to the audit trail for reviewer corrections covers what each entry should hold. The plan for reviewer training for document review covers how new reviewers learn to check a value instead of approving it.

Staffing a review queue for month-end and payer spikes

Staffing a review queue for spikes means planning reviewer hours for the heaviest days you can predict. Month-end closes, payer remittance runs, a backlog of scanned charts or a new clinic going live all raise volume. A spike also tends to raise the routed share, because rushed scans and unfamiliar layouts score lower.

The spike-week version of the example

The same fictional clinic, in a month-end week, might see these figures. They are examples only.

A spike day needs five people where a normal day needs three. Hiring for the peak leaves two people short of work on every normal day. The practical answers are to cross-train billing staff who review part-time during known spikes, to let lower-priority document types wait under a longer target, or to accept a planned backlog that clears in the quieter days after the spike.

Part-time reviewers need the same training as full-time ones. A billing specialist who helps only at month-end still has to know each document type's field rules and still has their corrections logged and sampled. Train them before the first spike, not during it, and give them the document types they already know from their main job.

Changing the confidence threshold during a spike is the wrong lever. A higher threshold routes more fields, not fewer. Lowering it to cut volume lets more uncertain values through unreviewed, which trades a staffing problem for a data quality problem.

Plan reviewer seats against the spike as well. On the current Schema Care plans, Clinic includes three reviewers and Group includes unlimited reviewers, and the plan list on the pricing section shows each plan's monthly page allowance next to its reviewer count.

Staffing the review queue from average volume: the classic error

Staffing the review queue from average volume leaves it behind on every day that is busier than average. Fields that arrive on a heavy day wait for capacity, and that capacity only frees up on light days later in the week. The weekly total balances, yet many fields miss their turnaround target.

Here is the arithmetic with the example clinic staffed at exactly 12.5 reviewer hours a day, its average need. The daily volumes are fictional.

The week averages 600 documents a day, and the queue is empty by Friday. Yet fields from Monday sit in the queue until Wednesday or Thursday. Against a same-day target for EOB fields, most of Monday and Tuesday's overflow misses.

Two fixes work. Staff to the busy days, or move part-time help onto the queue on the days that are known to be heavy. Watch the oldest-item age each morning so a backlog shows up as a number before it becomes a missed posting. For how review staffing fits into the wider billing operation, the revenue cycle automation category covers the stages before and after the queue.

A last check on any staffing plan: revisit it when an input changes. A new document type, a threshold change, a new payer layout or a drop in scan quality each moves the routed share, and the headcount moves with it.