The Quick Read: It is accurate at the market level and much less reliable for a single address, so treat any property-level projection as a screening estimate and not a forecast. Operators scaling up hold something a first-time buyer lacks: their own trailing actuals, which can calibrate every new projection. Lenders may use a projection on a purchase, but the number that qualifies a deal is not the cash flow you will bank.
- Market-level data suits market selection, seasonality, and comps.
- Address-level projections can miss in either direction, and misses grow in thin markets.
- Your own trailing-12 statements are the best calibration set you have.
- On STR purchases, select programs in the network go to 75% LTV with a 1.00 coverage floor, subject to lender guidelines.
- Clearing 1.00 on a projection is not the same as positive cash flow.
How Does a Projection Get Built?
A projection is a model, not a measurement. The tool pulls nearby comparable listings and builds a relevance-weighted index of how they perform. It layers on seasonality plus demand and supply, then outputs annual revenue, ADR (average daily rate), and occupancy, per RentalRecon’s guide to the Rentalizer.
The tool also shows a confidence score that reflects how many strong comps exist. That guide treats 70%+ as reliable and anything below 50% as a wide-variance rough range.
Two inputs matter most. Listing counts are observed directly. Occupancy and ADR are modeled, since the tool cannot see what a host actually banked. A methodology explainer from Rapid Eye Inspections describes property-level projections as estimates layered on estimates.
What Does AirDNA Say About Its Own Accuracy?
AirDNA’s published figures are strong, but they describe the aggregate. RentalRecon’s review reports that AirDNA claims roughly 95–99% aggregate accuracy on Airbnb revenue, including benchmarks of 94.9% for Airbnb and 98.7% for Vrbo against platform-reported data. The same review adds the key point: aggregate accuracy is not the same as accuracy on one property’s forecast.
Those figures are also the company’s own benchmark. The Rapid Eye explainer notes there is no published independent audit behind the roughly 97% number.
So a 97% headline does not mean your property’s number is 97% right. It means the model tracks markets well in total.
Where the Projection Is Strong and Weak
| Use | Reliability | Cross-check with |
|---|---|---|
| Market screening | Strong | Local supply trends |
| Seasonality shape | Strong | Your own calendar |
| Address-level revenue | Moderate to weak | Local manager estimate |
| Occupancy and rate mix | Weak | Trailing actuals |
| Net profit | Not covered | Your expense history |
| New-launch ramp-up | Weak | Your past launches |
How Far Off Can a Single Address Be?
Far enough to matter, and in both directions. One STR advisory blog, VaultSTR, says address-level estimates run 15–30% off, with misses clustering high. It suggests underwriting year one at 60–75% of the projected figure. That is that firm’s working rule, not an industry standard.
Other operators saw the opposite. One operator on BiggerPockets reported 67% projected occupancy against 76% actual, and a $244 ADR against $259 actual. Revenue came in about 20% above the projection.
An operator review from 10xBNB reports a $54,000 estimate against $37,000 actual on one property, and $42,000 against $61,000 on another. Same tool, opposite misses.
So “it always overestimates” is a myth. The honest read is that error is real, sizable, and not reliably signed.
Why Do Projections Miss?
Several mechanisms are reported across these sources:
- Survivorship in the comp pool. Failed listings drop out of the data, so comps skew toward winners.
- Inferred bookings. Occupancy comes from a classifier that separates booked from blocked calendar days. It reads calendar changes, not transaction records.
- Fees in the total. Revenue figures can include cleaning fees, which inflates what you may think is nightly income.
- Missing property traits. Finishes, views, and amenities are not in the comp numbers.
- Thin markets. Drift grows where listings are few. Sources point to under 100 listings, with confidence scores of 40–50% in markets of 20–30 listings.
- Unique homes. A Tampa-area manager on BiggerPockets calls the tool very accurate for uniform properties, like a 3/2 pool home, and likely useless for one-of-a-kind ones.
Launch-year ramp-up is another gap. Projections describe established comps, not a new listing building its reviews.
What Changes When You Are Scaling?
Scaling changes the stakes, not the tool. Here is where a portfolio operator should think differently.
Errors can compound instead of averaging out. Ten units in one market share one comp pool. If the model runs high for that pool, it runs high on all ten. Units in different markets are more likely to wash out.
Your own additions change the comps. Your fifth unit in a submarket becomes part of the data your sixth unit is judged against.
You hold a calibration set. Pull the projection you would have had at acquisition for each current property. Set it beside that property’s trailing 12-month statements. The gap, by market, is your variance factor. Apply it to the next deal.
Run the backtest this way: three properties in one market, each projected above actuals by a similar share. That pattern says the market’s comp pool runs hot, and you haircut the next deal there. If the misses scatter both ways, the tool is noisy but not biased, and you widen your range instead.
A non-QM practitioner on BiggerPockets, cited above, compares 12 months of actuals with the projection when underwriting. He found it at times scarily accurate, with the caveat that much depends on the host. A proven operator may beat the model. You can only know by checking.
Turning Revenue Into a Coverage Number
Projected revenue is gross. DSCR, the debt service coverage ratio, divides rent by the full monthly housing payment: principal, interest, taxes, insurance, and any HOA dues (PITIA). Repairs, vacancy, management, utilities, and capex sit outside it.
That gap is the trap. A deal can clear 1.00 on a projection and still lose money after real operating costs. One tool vendor, Chalet, warns that many STR deals lose money after debt service even when they look profitable before it. Treat that as a vendor claim, but the logic holds.
Underwrite in three bands instead of one number:
- Downside: your calibrated haircut applied, with lower occupancy.
- Base: the projection adjusted by your variance factor.
- Upside: the raw projection.
If the downside band still covers the payment with room to spare, the deal survives a miss. If only the upside band works, walk away.
How Do Lenders Treat a Projection?
Across our wholesale network, STR income usually comes from one of a few documents: an appraiser’s short-term rental analysis, a 12-month management statement, or 12 months of platform or bank statements. On purchases, a market-data report such as AirDNA can stand in where there is no operating history. Most programs lean on trailing 12-month statements when they exist. The complete DSCR loans guide walks through the documents.
Form 1007 is a long-term rent schedule. An appraisal firm, Class Valuation, argues it cannot support STR income and that appraisers should use a labeled short-term rental income addendum instead.
A lender accepting a projection does not prove the projection is accurate. It is a documentation convention. Lenders may also apply their own adjustment to projected revenue, and that varies by program.
Here is what STR files in the network typically look like, subject to lender guidelines:
- STR purchases go up to 75% LTV on the strongest files.
- STR refinances and cash-out sit around 70% LTV.
- A 640+ credit score is the usual minimum, and about 12 months of hosting history is expected.
- Coverage starts at 1.00 on both purchases and refinances.
- Loans run up to $3,000,000 on standard programs. Terms vary by lender guidelines, property type, leverage, credit profile, and full file review.
Reserves vary by lender, leverage, loan size, and transaction type. About 6 months of PITIA is common. Larger down payments can lift coverage, but they never erase credit floors, reserve rules, or property eligibility.
For operators whose projection leaves a deal thin, lenders review coverage below 1.00 through select lenders in the network, with leverage and terms adjusted. For the contrast between lender types, see Lendmire’s piece AirDNA DSCR Lenders vs LTR Lenders for an STR Portfolio Operator.
Short-term rental rules can vary by city, county, HOA, and property type, so investors should confirm local rules before relying on projected rental income.
Trust, Verify, or Walk
A simple checklist for each new acquisition:
1. Trust the projection for screening when the confidence score is high, the home is uniform, and the market is deep.
2. Verify when comps are thin, the property is unusual, or you have no actuals in that market. Get a local manager’s estimate and a second data source.
3. Walk when the deal only works at the raw projection, or when your own backtest shows the market running well below it.
Refresh the numbers when your own units in the market change the picture, or when trailing actuals drift from your assumption. A projection is a snapshot. Your statements are the record.
Key Terms Defined
ADR: Average daily rate, the average nightly price a booked night earns.
Occupancy: The share of available nights that are booked.
DSCR: Debt service coverage ratio, annual or monthly rent divided by the full housing payment.
PITIA: Principal, interest, taxes, insurance, and any HOA dues.
LTV: Loan-to-value, the loan balance as a share of the property’s value.
Trailing 12: The last twelve months of actual platform or bank statements.
Frequently Asked Questions
How accurate is an AirDNA projection at the property level?
Expect a meaningful miss. Reported ranges run 15–30% off at the address level, and the direction is not consistent. Market-level data is far more dependable than a single-address figure, so use it for screening and not as a forecast.
Does accuracy improve as I add units?
Only partly. Your own actuals improve your calibration, but units sharing one market also share one comp pool and one possible bias. Spreading across markets is more likely to average errors out. Backtest by market.
Should I haircut the projection?
Usually, yes, until your own data says otherwise. One advisory firm suggests underwriting year one at 60–75% of the projected figure. That is their rule, not a standard. Your backtest gives you a better factor than any rule of thumb.
Can I use a projection to finance a purchase?
Often, on purchases without operating history, subject to lender guidelines. Operating properties usually rely on trailing 12-month statements.
Does clearing 1.00 mean the deal makes money?
No. Coverage compares rent to the housing payment only. Repairs, vacancy, management, utilities, and capex sit outside it, so a deal can qualify and still underperform.
If you are buying or refinancing a rental property and want to see how the numbers work, Lendmire can help you compare DSCR loan options based on the property income, credit profile, leverage, and investor goals. Tax treatment varies; consult a qualified professional.
Investors focused on short-term rentals can review DSCR loans for Airbnb and short-term rentals.
About Lendmire
A non-QM mortgage broker (NMLS# 2371349), Lendmire arranges DSCR financing for real estate investors in 41 markets — 40 states plus Washington, D.C. Because deals are underwritten primarily on property cash flow rather than personal income documentation, the structure suits self-employed buyers and entity-owned portfolios. Lendmire places loans through wholesale investor lenders; it is not a direct lender. Scotsman Guide named Lendmire a Top Mortgage Workplace in both 2025 and 2026.
Get Started
Ready to find the right loan for you?
In about 30 seconds you can review financing options available for your investment property. No commitment required.
Informational only. Not a Loan Estimate, approval, or commitment to lend. Program availability and eligibility are subject to lender guidelines, credit approval, property review, and underwriting.
References
1. RentalRecon – Rentalizer guide
2. Rapid Eye Inspections – Where AirDNA gets its data
3. RentalRecon – AirDNA Review
4. VaultSTR – Rentalizer accuracy
5. BiggerPockets – How accurate is AirDNA for STR revenue projection
7. BiggerPockets – How accurate are AirDNA estimates and calculators
8. Chalet – Airbnb calculators
9. Class Valuation – Form 1007 and STRs
This article is part of Lendmire’s short-term rental loan program — full qualification details, guidelines, and scenarios live on the program page.
Related reading: Does a Portfolio STR Operator Show Reserves on Every DSCR Loan? · How to Qualify for an STR DSCR Portfolio Loan Past Conventional Loan Limits · How to Qualify for a Blanket DSCR Portfolio Loan at the Five-Million Mark
Brandon Miller
Founder & CEO, Mortgage Loan Originator, Lendmire LLC
- Mortgage Loan Originator · NMLS# 1129696 · Verify on NMLS Consumer Access
- North Carolina Real Estate Broker · License# 343312 · Verify on NCREC
- North Carolina Insurance Producer · License# 19053198 · Property, Casualty, Life, Health · Verify on NAIC SBS
- Lendmire LLC · Firm NMLS# 2371349 · Verify firm licensure
Required disclosures. Lendmire (NMLS# 2371349) operates as a licensed mortgage broker, not a direct lender or depository. The discussion in this article is general in nature and should not be relied upon as financial, legal, or tax advice — every investment scenario is unique and should be reviewed by a qualified professional. Any loan inquiry is subject to lender underwriting, and this article is not a commitment to lend or a guarantee of approval. Mortgage rates, loan terms, and program guidelines vary by borrower, property, and state, and may change without notice. Equal Housing Opportunity. Verify licensure at NMLS Consumer Access.