← All resources

From remit feed to 13-week cash forecast

Your payer mix and remit cadence are the strongest signal in any healthcare cash forecast. Here's how to turn an 835 feed into a 13-week projection your CFO can defend.

What an 835 feed actually tells you about cash

Your 835 feed, the same file at the center of reconciling 835s and paper EOBs into one pipeline (opens in a new tab), is the most predictive cash signal you have. It reveals which claims have been paid, adjusted, or denied, along with anticipated deposit timing in near real-time, provided the paper remits are digitized into the same feed by a lockbox that turns EOBs into structured data (opens in a new tab).

Most healthcare finance teams leverage the 835 feed for posting and reconciliation only, overlooking its forecasting potential despite containing the exact claim-to-deposit matching data (opens in a new tab) needed for cash projections.

The three components of a forecast from 835

  1. Open AR by payer, aged. Extract from your billing system as your inventory of expected cash.
  2. Per-payer payment velocity. Calculate from trailing 90 days of 835 history, expressed as "days from claim to remit" per payer, converting AR into expected timing.
  3. Per-payer denial and adjustment rate. Determine from the same 90-day window, converting gross AR into net expected cash.

Multiply open AR by payer × probability of payment × velocity by payer = expected cash by week. Sum across all payers to produce a 13-week forecast grounded in actual remit history rather than assumptions.

What beats spreadsheet-based forecasting

A spreadsheet forecast built from monthly aggregate AR is wrong by 10 to 20 percent on any given week. A remit-feed-based forecast is typically within 3 to 5 percent, the kind of accuracy gain we quantify across the four layers of billing reconciliation automation (opens in a new tab). This difference carries meaningful implications when managing reserves and timing transactions.

Treasury teams should execute this analysis weekly, not quarterly.