If you’ve felt like Airbnb data has been harder to interpret over the past year, you’re not wrong.
Between shifting fee structures, inconsistent reporting, and now a partial rollback of those changes, we’re in a weird in-between phase where a lot of the numbers investors rely on just… don’t line up anymore.
Let’s break down what actually happened, what’s happening now, and why it matters if you’re buying, selling, or underwriting a short-term rental.
What Changed (And Who It Affected)
For years, Airbnb operated on a split-fee model:
- Hosts paid roughly 3%
- Guests paid ~14–16% in service fees
Then Airbnb rolled out a major shift—primarily impacting hosts using property management software (PMS), but eventually influencing the broader ecosystem:
👉 They moved to a host-only fee model, where hosts pay around 15.5%, and guests see no service fee.
From a user experience standpoint, this made pricing simpler. Guests now see a clean, all-in number instead of getting hit with fees at checkout.
But from a host and investor standpoint, it fundamentally changed how revenue is displayed, tracked, and interpreted.
What’s Happening Now (The Quiet Reversal)
Here’s where things get interesting—and honestly, where a lot of confusion is coming from.
Airbnb is now quietly allowing some hosts to switch back to the split-fee model.
Not everyone. Not consistently. And not with a lot of transparency.
But as shown in the Instagram video I shared, some hosts are seeing the option reappear in their settings.
So now we’re in a hybrid environment where:
- Some listings are using host-only fees
- Others are back on split fees
- And many investors have no idea which is which when looking at data
👉 That inconsistency is the real problem.
Why This Is Messing With Your Revenue Numbers
At a high level, the economics didn’t change as much as it seems—they just got redistributed.
But the way revenue is reported absolutely did.
Under the old model, guest fees were added on top of your nightly rate. Under the new model, those fees are baked into your pricing.
So depending on how a host adjusted (or didn’t adjust) their pricing:
- Revenue can look artificially lower
- Or artificially higher
- Or just inconsistent across similar properties
Two identical listings could perform the exact same operationally and show completely different “revenue” depending on their fee structure.
That’s a problem.
Why Year-Over-Year Data Is Now Unreliable
This is where I see investors make bad calls.
You cannot cleanly compare:
- 2023 (mostly split fees)
- 2024–2025 (mixed or host-only fees)
Because you’re not comparing the same thing anymore.
It’s like trying to compare:
- A price before tax
- To a price after tax
Without adjusting for the difference.
So when people say:
👉 “Revenue is down year over year”
My first question is always:
👉 “Based on what structure?”
Because without normalizing for fees, that conclusion might be completely wrong.
Why This Breaks Proformas (And Investor Expectations)
If you’re using historical data or tools like AirDNA to project revenue, you need to be a little more skeptical right now.
Here’s why:
- Some data reflects split-fee structures
- Some reflects host-only structures
- Some reflects hosts who adjusted pricing correctly
- Others didn’t
And now Airbnb is introducing a third variable by letting some hosts switch back.
So your comp set isn’t clean.
Your assumptions aren’t consistent.
And your proforma—if you’re not adjusting for this—is likely off.
Not wildly wrong, but wrong enough to matter.
What You Should Actually Be Paying Attention To
Right now, the only number that really matters is:
👉 What does the host actually take home?
Not:
- What Airbnb says the listing earns
- Not ADR
- Not gross booking value
Focus on net revenue after fees and normalize your assumptions across every deal you’re analyzing.
Because until Airbnb fully commits to one model again (or at least becomes transparent about it), we’re going to keep dealing with messy data.
Final Thought
Airbnb didn’t just change their fee structure.
They changed how we have to interpret performance.
And now that they’re slowly rolling parts of it back, the data is even messier than before.
If you’re not adjusting for that, you’re not actually analyzing deals—you’re just reacting to numbers that don’t tell the full story.
