Reading the New Appraisal Dataset: What UAD 3.6 Makes Visible

Every appraisal your operation has ever ordered contained far more information than you could use. The comps, the adjustments, the condition observations — all of it arrived flattened into a form-shaped PDF, readable one report at a time and analyzable, practically speaking, never. That's the quiet significance of the redesign that becomes mandatory on November 2: UAD 3.6 doesn't just replace the familiar forms with a single dynamic report. It turns the appraisal into a dataset.

Most of the industry's preparation has focused on getting reports in — software versions, vendor readiness, submission plumbing. Fair enough; that's the deadline's demand. But the durable change is on the reading side, and lending teams that plan for it now will be asking better questions in December than their peers ask all next year.

What Does the New Appraisal Dataset Actually Contain?

UAD 3.6 replaces free-form text and catch-all fields with discrete, machine-readable data points: standardized condition and quality ratings applied consistently across property types, explicit market-conditions and time-adjustment fields, structured sales-comparison data where every adjustment is a value in a defined field, and property characteristics captured once, in one specified format. The same appraisal dataset standard now spans what used to be six-plus different forms — the redesigned URAR flexes to the property instead (Freddie Mac's timeline fact sheet summarizes the consolidation). In plain terms: what an appraiser once wrote, a system can now read.

Condition Granularity: The End of "C4, Probably Fine"

Under the old regime, condition lived in a coarse rating and whatever the narrative happened to mention. The new structure captures condition and quality with far more granularity — at the level of individual systems and updates, with dates and definitions attached. For credit teams, that converts a judgment call into a queryable field: show me our originations where the roof observation is past its stated life is now a database question, not a file-review project. For QC, it means condition inconsistencies between reports on similar properties surface mechanically instead of by a reviewer's memory.

Time Adjustments: The Market, Stated Plainly

Perhaps the most underrated change: market conditions and time adjustments become explicit, structured fields rather than narrative asides. When an appraiser adjusts a comp for a moving market, the direction and magnitude now sit in data. Aggregated across even a mid-sized lender's monthly volume, that's a live read on how appraisers — the people literally closest to the sales evidence — see each local market trending. Portfolio teams have paid vendors real money for weaker signals.

What this Means for Review and Credit Workflows

The first-order wins are mundane and immediate: automated completeness checks catch defects at submission instead of at underwriting, and revision cycles drop because "field missing" stops being a human discovery. (Both GSEs' submission platforms have been accepting 3.6 reports in broad production since January 26, so the plumbing is proven.) The second-order wins are where strategy lives. Panel management sharpens when appraiser-level patterns — adjustment behavior, condition-rating drift, comp selection radius — are queryable across every report rather than sampled in review. The quality metrics we outlined in Beyond Turn Time stop being a spreadsheet discipline and become a dashboard. And exception-based review finally gets a real foundation: route human reviewers to the reports whose data flags them, not to a random ten percent.

Who Owns the Dataset?

A question worth settling before November: where does your 3.6 data live, and who controls it? The structured appraisal data flowing through your AMC and platform vendors is generated on your loans, and your contracts should say plainly that you can export it — complete, machine-readable, and without a per-report toll. Vendors differ more here than their marketing suggests: some pass the full dataset through to the lender's systems, some render it and retain it, and some treat aggregated appraisal data as their own asset. None of those positions is illegitimate, but only one of them leaves you able to run the analysis described above without asking permission.

Add a retention rule and a named internal owner. The dataset only compounds in value if someone keeps it, keeps it clean, and answers for it.

A Caution and a To-do List

The caution: a structured field is still an opinion. Standardized data makes appraiser judgment more visible — it doesn't make it more correct, and false precision is the new failure mode. The dataset tells you what the appraiser concluded with new clarity; whether the conclusion holds still takes a reviewer's eye. Treat the new fields as better evidence, not as truth.

The to-do list for the next three weeks is short but real. Ask your appraisal-management platform what 3.6 data it will surface, store, and export — a vendor that renders the new report but discards the dataset is giving you a prettier PDF. Decide where the structured data lands in your systems and who owns querying it. And put one person on a thirty-day assignment for December: report back on what the new fields reveal about our pipeline that we couldn't see in October. That answer becomes your 2027 QC plan. Small lenders aren't exempt from the opportunity: even a hundred loans a year is an appraisal dataset worth querying.

The mandate date is an ending — of forms, of the transition period, of the compatibility era. But for lending operations, November 2 is better understood as the day the appraisal dataset arrives in earnest. The lenders who treat it as a compliance finish line will file the new reports the way they filed the old ones. The ones who treat it as an instrument will spend 2027 reading a market their competitors are still squinting at.