Property data collection: the lender's guide (2026)

The short version.

• What it replaces: value acceptance + property data replaces the appraisal. A trained collector visits the property instead of an appraiser, and nobody signs an opinion of value.
• What comes back: roughly 120 standardized attributes, 40 to 60 photographs and an ANSI compliant floor plan, built to the Uniform Property Dataset and submitted through an API before the note date.
• Who carries the risk: you do. The lender represents that the property has no safety, soundness or structural integrity issues, and without an appraisal the only way to confirm that is to read the collector's descriptions and look at the photos.
• Who this is for: valuation operations, appraisal desks, chief appraisers and vendor management at lenders, credit unions and AMCs. If you are a homeowner or an agent looking for collection work, this one is written past you.

The DU message comes back with the offer attached, so nobody orders an appraisal. You order a property data collection instead, the file goes quiet for nine days, and when it lands in the review queue the laundry room is missing, the rear elevation is one blurred photo taken into the sun, and the floor plan does not reconcile with the room count the collector typed in.

Now you have a choice nobody budgeted for. Send someone back, or push a file to a reviewer that you would not want to see in a post-closing sample.

If you run valuation operations, that gap is the part of appraisal modernization you actually own. The policy is public and settled. What is not settled is how a guided photo and measurement assignment gets done right the first time, by someone who is not an appraiser, in a house they have never seen.

We build Venta Capture, a product of VentaVid, for guided visual capture: the part where someone on site has to record exactly the right images and someone at a desk has to trust them. If your collections keep coming back a few rooms short, book a 15-minute demo and we will walk it through on a real property file.

What a property data collection actually is

Start with what it is not, because the argument about it is louder than the definition. Freddie Mac puts it flatly in its property data collection overview: a collection "is not an appraisal or appraisal report and doesn't involve the development of an opinion of value."

A property data collection is a factual observation exercise with the judgment deliberately removed, which is exactly why it can be done by someone who is not an appraiser. Fannie Mae's own description of the work excludes every subjective element an appraiser would normally supply: no comparables, no opinion of value, no condition or quality ratings, per its perspective on advancing collateral valuation.

What remains is a visit and a camera. Selling Guide B4-1.4-11 is specific about the visit: "the property data collector is the individual who personally visits the subject property," and what they produce is "a visual observation of the interior and exterior areas of the subject property."

The collector works from a handheld app or a web interface built to the Uniform Property Dataset, the shared schema both GSEs now require and the format Freddie Mac has mandated for applications dated on or after 1 April 2024. So the assignment is not "go and photograph the house." It is "fill this dataset, in this order, with images that satisfy these conditions."

A judgment task you solve by hiring a qualified person. A completeness task you solve with a workflow. A remote property inspection has the same shape, and so does a hybrid appraisal, where the same data goes to an appraiser instead of standing in for one.

The three paths a collection can feed

Three products get called "appraisal modernization" in the same meeting, and they are not the same thing.

Value acceptance + property data, under B4-1.4-11, replaces the appraisal entirely. No appraiser, no opinion of value, no report. The collection confirms eligibility, and the lender carries what the appraisal would otherwise have carried.

Freddie Mac's equivalent is ACE+ PDR, which the program page describes as "a collateral valuation offering that allows lenders to deliver eligible loans without an appraisal report." Same idea, different plumbing: the property data report must use the Uniform Property Dataset and go in through the Beyond ACE (bACE) API.

A hybrid appraisal is the one people blur. The collector still visits, but the appraiser stays in the file and still signs the opinion of value, working from someone else's observation. That is a different risk posture and a different conversation with your appraisal panel, which is why it gets its own guide for lenders.

The commercial case for the first two is published, not inferred. Freddie Mac's own page puts ACE+ PDR at an average of about $200 against about $600 for appraisal-based loans, and reports that as of the second quarter of 2025 it saved roughly 12 days on purchases and 10 on refinances.

Read that 12 days carefully, because it is the number the rest of this post is about. It is the saving when the collection comes back complete and usable the first time. A second trip does not shave it, it eats it.

What is actually in the deliverable

The dataset is more specific than most lenders expect before they run their first few hundred files. Fannie Mae describes a typical collection as about 120 standardized property attributes, 40 to 60 photographs, and an ANSI compliant floor plan, with the data and images embedded with geocoding technology, per its Property Data Collection 101 materials.

Freddie Mac's description of the content runs along the same lines, and adds the parts people forget to brief a collector on:

• Interior and exterior information, both observed on the same visit, which is why the occupant has to be home and cooperative.
• Measurements of the dwelling and other structures, used to produce the building sketch and the floor plan. Outbuildings count.
• Design features that impair use and cannot easily be changed, recorded factually rather than rated.
• Nearby factors that are potentially adverse or beneficial to the location.
• A comprehensive set of photographs, which in practice means every room and every elevation, not a representative sample.

The thing to internalize is that the deliverable is scored on completeness against a named dataset, so a collection with four rooms missing is not 90 percent finished, it is returned. There is no partial credit in a schema.

Then it has to land somewhere. On the Fannie Mae side, the collection must be "successfully submitted to Fannie Mae's Property Data API," it has to happen before the note date, it stays valid for 12 months from the collection date, and the loan goes out the door with Special Feature Code 774. On the Freddie Mac side the PDR goes through bACE in UPD format.

Both sides also expect the report to be read, not filed. Freddie Mac is explicit that property data reports "must be reviewed and underwritten the same way appraisal reports are" and must appear in the seller's quality control sampling. That is the same discipline behind any serious photo evidence standard for inspections.

Who is allowed to collect, and what you have to verify

This is where a modernization program turns into a vendor-management program.

B4-1.4-11 puts three checks on you before a collector ever drives out: the lender must verify the collector has an annual background check, professional training, and the knowledge to complete the collection competently. The collector must also have no interest in, or ties to, the loan, its participants or the property.

Those are not one-time checks. The guide requires the lender to review collector credentials annually and to monitor the work through quality control, prefunding and post-closing. Independence is its own rulebook: Fannie Mae's property data collector independence requirements have applied to applications dated on or after 1 November 2023, and the definition of an independent party reaches the collection company and related persons, not just the individual with the phone.

Freddie Mac loads the seller up in the same way. Sellers are accountable for the accuracy and reliability of PDRs, for overseeing and training collectors, for evaluating them continuously and documenting performance, for addressing recurring deficiencies, for continuing education after a UPD change, and for stopping the use of chronically underperforming collectors.

Sit with that last duty for a second. The policy assumes you will find collectors who underperform, and it tells you to measure them and drop them. It does not tell you how to spot it in a file.

The market has not quietly accepted this, either. The 2024 REALTORS® Appraisal Process Report, fielded in September 2024, found 65 percent of respondents concerned about the quality of the data collected, 80 percent perceiving it as lower quality than data collected by appraisers, and 73 percent concerned about adequate training. That is a survey of agents, not a lender audit, and agents have an interest in the answer. It is still the sentiment your borrower's agent brings to the appointment.

Policy and survey land in the same place. Nobody doubts that a trained person can record 120 attributes. What everybody doubts is whether the person who showed up on a Tuesday actually did.

Where the reps and warranties actually land

Everything above is process. This is the part that should set your effort level on the capture step.

When you exercise value acceptance + property data, you represent that the property does not have safety, soundness or structural integrity issues, that there are no significant incomplete construction or renovation items, and that the property meets eligibility requirements. Without an appraisal, the Selling Guide expects you to confirm all of that by reviewing the collector's descriptions and the photographs, which makes the photo set the basis of the decision rather than a record of it.

So image quality is not a service-level nicety. It is the evidence behind a representation you made to a secondary-market buyer.

The failure path is written into the guide too. The collector has to identify and communicate safety, soundness or structural integrity issues and significant incomplete work. When the data shows a failing item, the guide says the lender "may need to obtain a professionally prepared report from a qualified professional," and lists the usual suspects: well, septic, foundation, roof, electrical, mold.

From there the file gets longer. Repairs have to be completed, and you have to provide satisfactory evidence of completion before you can sell the loan, which is where Form 1004D and the completion alternatives come in. That step is a second visit with its own turn time and its own evidence problem, so it gets its own guide.

The path also closes off entirely in a long list of cases: two to four unit properties, co-ops, manufactured homes, proposed construction, construction-to-permanent loans, investment properties qualifying on rental income, HomeStyle Renovation and Refresh, Texas 50(a)(6), leasehold or resale-restricted properties, gifts of equity, manually underwritten loans, and transactions at $1,000,000 or more.

If you are standing up this workflow and you want to see what a guided capture looks like before you commit to a process, book a demo with us and bring a collection that came back wrong.

Where this breaks in practice

Four things go wrong, and all four are capture-step problems rather than policy problems.

• Incomplete photo sets. The dataset wants every room and every elevation. A collector working from a checklist in their head misses the utility space, the attic reached by permanent stairs, or the side elevation nobody walks past. The file comes back and the clock restarts.
• Collectors of uneven quality. Both GSEs tell you to measure performance and drop the weak ones, which only works if the work product is comparable between collectors. If each collector decides what a "rear elevation" photo looks like, you cannot compare anything.
• Re-trips. Every gap is a drive. The second visit also needs the occupant home again, which is the part that turns a two-day fix into a nine-day one.
• No provenance for the reviewer. Images have to be geocoded and the submission is logged, but a reviewer pulling the file three months later still sees a folder of photographs. Establishing that those images were recorded at that address on the day claimed is a separate question, and it is the one that gets asked when something goes wrong.

Walk the calendar yourself, with your own numbers. Freddie Mac measured about 12 days saved on a purchase. Say you run 200 eligible files a month and one in ten needs a gap-fill or a second visit, and say each of those loses five days of the saving. That is about 100 file-days a month handed back, on a program you adopted to get them.

Those inputs are illustrative, not a measured result. Swap in your own volume and your own return rate, because the ratio is the only part that matters: the entire time saving depends on first-time completeness.

Of the four, provenance has the longest tail. When a file is challenged the argument is rarely about whether a photo exists, but about what it shows, when it was taken and where. Anyone who has worked a disputed photo or video record knows the distance between holding an image and being able to stand behind it, and the same questions turn up when verifying any customer-submitted photo.

Fixing the capture step, not the policy

The honest boundary comes first, because this is a policy-sensitive workflow and vague product claims are worse than no claims.

Venta Capture is not an approved Fannie Mae or Freddie Mac property data collection vendor. It does not submit to the Property Data API or to bACE, it does not produce a property data report in UPD format, and it does not produce an appraisal or an opinion of value. If you need a UPD-compliant collection delivered into a GSE pipeline, you need an approved collection vendor and a panel you manage. Nothing below should be read as a claim to be that.

What it is: a way for a guided, verifiable photo and video set to reach a file. You send a secure link. The person opens it in the mobile browser on the phone they already have, with no app to install and no account to create, and a guided workflow walks them through exactly the images you need, slot by slot, with an example photo and a plain instruction at each step. A retake asks why, and stores the answer. Each submission arrives as a structured case with its own reference, sealed and externally verifiable.

The mechanism matters here precisely because a data collection is a guided assignment: when the organisation builds the steps, the order and the instructions, the right images come back the first time instead of the second. That is the same reason a remote collateral inspection works better as a built flow than as an email asking for photos.

Four places it earns its keep around a GSE collection rather than inside one:

• Triage before you order. A short guided capture from the borrower tells you whether the property has an obvious condition problem that will cost you the no-appraisal path, before you spend the order.
• Gap-fill instead of a re-trip. When a collection comes back missing a room or an elevation, a targeted capture from the occupant can close a documentation gap, for the purposes your own policy allows, without putting a second visit on the calendar.
• The condition-problem file. Once a failing item appears you are assembling your own evidence anyway, for a professionally prepared report or for repairs. Guided capture gives you a structured, dated, sealed record instead of a thread of forwarded attachments.
• Completion evidence and QC re-looks. The repair-completed step needs a record somebody will look at months later, and prefunding or post-closing sampling needs a comparable second record rather than a phone call.

Two limits we hold to, because they are what make the rest credible. Capture records observed signals, never verdicts: a submission carries what was observed, measured and established, and a person decides what it means. A receipt timestamp establishes when the submission arrived, not when the condition it shows came into being.

The same loop sits behind the lending work we have written up elsewhere, from collateral verification to construction draw inspections and loss draft inspections. The property-side pages live at Venta Capture for property, the mechanics at the product overview, and the wider tooling question in our guide to site inspection software for lenders.

Frequently asked questions

Is a property data collection an appraisal?

No. Freddie Mac states plainly that property data collection is not an appraisal or appraisal report and does not involve developing an opinion of value. It is a factual observation of the interior and exterior, recorded against a standard dataset.

Who can perform a property data collection?

A trained third-party collector who personally visits the property, or a licensed or certified appraiser or appraiser trainee. The lender has to verify the collector has an annual background check, professional training and the competence to do the work, and has to review those credentials annually.

How many photographs does a property data collection need?

Fannie Mae describes a typical collection as 40 to 60 photographs, alongside roughly 120 standardized property attributes and an ANSI compliant floor plan. The real requirement is coverage of the dataset, so the count follows the property rather than a quota.

What is the difference between value acceptance + property data and a hybrid appraisal?

Value acceptance + property data replaces the appraisal: there is no appraiser and no opinion of value, and the lender carries the representations. In a hybrid, the collector visits but an appraiser still develops and signs the opinion of value from the collected data.

How long is a property data collection valid?

Twelve months from the collection date, under Fannie Mae's Selling Guide, and the collection must be submitted to the Property Data API before the note date. The underlying value acceptance + property data offer also has to be no more than four months old at the note date.

What happens if the collection finds a property condition problem?

The collector must identify and communicate safety, soundness or structural integrity issues and significant incomplete work. Depending on what turns up, the lender may need a professionally prepared report from a qualified professional, and necessary repairs have to be completed with satisfactory evidence of completion before the loan can be sold.

Can the borrower collect the data themselves?

Not for a GSE property data collection. The collector has to personally visit the property and must have no interest in or ties to the loan, the participants or the property, which rules the borrower out. Borrower-captured images can still support your own underwriting, triage or condition review, just not the collection itself.

Talk to us

If your collections keep coming back a few rooms short, the fix is upstream of the review queue. The policy is not going to change, the collector is not going to get a longer briefing on the driveway, and the 12 days the program promised you only shows up when the images arrive right the first time.

What it takes to try it.

• A free plan, no credit card, and you can be live in about 10 minutes.
• Build a guided flow, send yourself a test link, and see what arrives in the case file.
• Stuck? Book a free setup call and we will build your first flow with you.

The VentaVid team builds Venta Capture for teams who have to see a property before they can act on it.

Start for free and build a property flow this afternoon, or book a demo and we will run it against a collection that came back wrong. More on the platform at Venta Capture.