How to read AirDNA and KeyBid Puls data before buying a short-term rental

A table of occupancy, average daily rate and revenue per listing looks like a solid argument for an offer. Yet each figure answers a specific question, and two serious sources can give different results for the same city. This guide explains what AirDNA's metrics measure, how to compare sources, which traps appear in the 2026 data and how to check the area around an apartment in KeyBid Puls before you negotiate the price.

What each metric measures

AirDNA analyses listings on Airbnb, Vrbo and Booking.com every day and publishes metrics calculated over the last 12 months. Before you compare cities or apartments, check what goes into each figure.

Two examples from AirDNA's documentation show why definitions matter. A listing booked for 5 nights at $250 a night, with a $200 cleaning fee, has an ADR of $290, not $250. A listing available for 150 days of the last 12 months, with 50% occupancy, had 75 booked nights, far fewer than the percentage alone suggests.

Where the data comes from and what it cannot see

AirDNA estimates bookings from what is publicly visible on the platforms and checks them against real data from some hosts. It is worth knowing how the mechanism works:

• Daily collection. The main source is the public listing pages on Airbnb, Vrbo and Booking.com, analysed every day: prices, availability calendars, host information and reviews. Data for Airbnb and Vrbo has been collected since 2015 and for Booking.com since 2019; the latter was added to the platform in June 2026.
• Booking or block. A night that is unavailable in the calendar is not automatically a booking. AirDNA uses artificial intelligence models and 16 booking signals, such as length of stay, the time between booking and arrival, and reviews, to tell booked nights from blocked ones. The result is an estimate, not a statement from the host's account.
• Quality control. Managers and hosts with more than 100,000 properties worldwide share their real booking data with AirDNA, which uses it to check and fine-tune its estimates.
• No duplicates. An algorithm compares 14 elements, including location, title, description, number of bedrooms and calendar, so that an apartment listed on several platforms is counted once.
• A one-month lag. The data is updated monthly: in April, the most recent month you see is March. The public market pages updated on 6 October 2026 contain data up to September 2026.
• Direct bookings. A booking made outside the platforms shows up in the calendar only as an unavailable night, so it can be mistaken for a block.

Other sources have limits too. The tourism statistics of INS, Romania's statistics institute, only cover units with more than 10 bed places, so an ordinary apartment does not appear in them. The Crosspoint Research report is published once a year. The imobiliare.ro and VDI.ro indices show asking prices in listings, not transaction prices: in September 2026 the imobiliare.ro index for Bucharest stood at €2,312/m², and the final negotiated price may be lower.

Same market, two sources: how to compare the figures

Crosspoint Research, a Savills associate, analysed the market in 2025. AirDNA's market pages, updated on 6 October 2026, cover the last 12 months up to September 2026. As AirDNA reports in dollars, we converted its figures at the ECB rate of 9 October 2026, $1.1206 to €1, as a guide.

AirDNA's figures are 1–22% higher, but the ranking of the cities is almost the same: only Constanța and Cluj-Napoca swap places. The differences have simple explanations: different periods and currencies and, very likely, different rules on which listings are counted and what goes into revenue. On occupancy the sources are close: Crosspoint reports 62% in Bucharest and 47.4% in Iași, against 60% and 44% at AirDNA.

The practical conclusion is to work with a range. For Brașov, for example, start from €10,500–11,700 a year rather than a single figure, and test your calculation at the low end.

When the figures jump: the 2026 example

For Bucharest, AirDNA's October 2026 pages show year-on-year changes that look spectacular at first sight:

If you read only the revenue row, you might think a Bucharest apartment earns twice as much as a year ago. The rest of the table says otherwise: the rate stayed almost the same, and revenue per available night rose by a fifth. The difference comes from the listings being counted, which are a third fewer and available for more nights a year. A rough calculation, annual revenue divided by RevPAR, gives about 350 available nights per listing now, against about 210 a year earlier.

The same pattern appears in all six cities we checked: active listings are down 29–39% and revenue per listing is up 74–159%. Six simultaneous local increases of this size are unlikely. A common cause, linked to methodology or to the listings included, is more likely. In June 2026 AirDNA launched its 5.0 data model and added Booking.com data, and for listings exclusive to Booking.com it only has data from March 2025. AirDNA warns that including this channel may affect year-on-year comparisons.

The rule that follows: to track demand, follow RevPAR and occupancy, not revenue per listing. When you see an unusually large annual change, first check whether the methodology has changed, and only then whether the market has.

From the city average to your area

A city average hides large differences. AirDNA itself recommends narrowing down to the neighbourhood, filtering by number of bedrooms and estimating the specific address. Four things are worth checking:

• Seasonality. On AirDNA's scale, Constanța has a seasonality score of 49 out of 100 and Bucharest 94: on the coast, the gap between good and weak months is much wider. The same rough calculation gives about 240 available nights per listing in Constanța, against 335–350 in the other five cities, a sign that many coastal apartments are not offered all year round.
• The peak month. According to Crosspoint, August brings in around €1,500 per property in Constanța. Multiplied by 12, that gives €18,000, double the annual average of €8,980.
• Market Score. AirDNA rates markets from 40 to 100: Bucharest scores 98, Timișoara 97, Cluj-Napoca 92, Brașov 86, Iași 67 and Constanța 41. AirDNA points out that the score is a starting point, not an investment decision. A low score can hide a good area, and a high one a street with no demand.
• Nearby competition. How many similar listings are within a few minutes' walk, at what price and with what reviews. This check is done on the map, not in the city average.

For the coast, with figures and scenarios, see how to value an apartment in Constanța and Mamaia, and for the big cities, our yield comparison for Bucharest, Cluj-Napoca and Brașov.

Estimates for an address: what they assume

Tools that estimate the revenue of an address start from comparable listings. Rentalizer, AirDNA's tool, looks for similar listings within a 15 km radius, chosen by distance and by number of bedrooms, bathrooms and guests, and takes a weighted average of their past results, factoring in seasonality, demand and revenue growth. Three details change the result a lot:

• 365 days of availability. The projection for the next 12 months assumes the apartment can be booked every day of the year. If you use it yourself, renovate or close in the off-season, actual revenue will be lower.
• What the figure contains. Projected revenue includes the cleaning fee but does not include the host fee, other fees or taxes. So it is not a net income: you still need to deduct operating costs, the tourist tax, income tax and the health contribution (CASS).
• Quality of the comparables. The Comp Set Strength indicator, with the values Low, Medium and High, shows how close the comparable listings' revenues are. If one earns $50,000 a year and another $200,000, confidence in the average is low.

The same principle applies to any estimate, including KeyBid's: they are assumptions to verify, not promises.

Example: from RevPAR to annual income

Suppose a Bucharest apartment performs exactly like the AirDNA average, with a RevPAR of about €38 ($43). Annual income depends on how many nights you actually offer it:

Between 250 and 350 nights, the difference is €3,800 a year. In practice, nights added in the off-season usually earn less than the average, so the real difference may be smaller. All of these are amounts before your expenses. For a sense of scale, our Bucharest analysis starts from the Crosspoint figure: after commission, cleaning, utilities, income tax and CASS, gross takings of €12,072 become €4,455 net, a yield of 3.5%, before the tourist tax and the accountant. If you start from AirDNA figures, do not deduct platform service fees a second time, since AirDNA already removes them. Compliance costs, line by line, are in our analysis of the real cost of compliance.

How to check the area in KeyBid Puls

KeyBid Puls helps at the stage when you have an address and want to understand the area. A simple workflow:

1. Choose the city in KeyBid Puls, for example Bucharest or Constanța.
2. Select the hexagon closest to the address. Cities are divided into hexagonal zones with a 750 m radius, so you work with the zone, not the city average.
3. Read the nightly price estimates for your property type, from studio to house or villa, and the pricing strategies suggested for the zone.
4. Check upcoming events and the 7-day forecast. Puls shows the estimated impact of the weather, weekends and local events on price.
5. Compare the estimates with real listings in the same hexagon: 5–10 similar apartments, with their prices for the same dates, their reviews and their calendars.
6. Run the figures through the net calculation: commission, cleaning, utilities, income tax, CASS and the local tourist tax.

For a specific listing on Storia, Imobiliare or OLX, the KeyBid ROI Calculator extension shows an estimated annual yield directly on the page, calculated from the listing's public data. We explain how to use it step by step in our KeyBid ROI Calculator extension guide. If you want checking the area to become a habit after you buy, see the 15-minute weekly routine in KeyBid Puls.

Common mistakes

• Relying on a single source. For the same city, Crosspoint and AirDNA differ by up to 22%. Work with the range between them.
• Reading growth in revenue per listing as growth in demand. In Bucharest, +102% in revenue per listing came with only +20.3% in RevPAR.
• Using the city average for a specific street. Narrow down to the area and the apartment type, and check the listings nearby.
• Forgetting available nights. 60% occupancy over 250 nights means 150 booked nights; over 350 nights, 210.
• Deducting the commission twice. According to AirDNA, revenue and ADR are calculated without platform service fees. For Airbnb listings where the host pays the whole fee, the rule in the European Economic Area with the single 15.5% fee, AirDNA takes it out of the calculation. Check what the figure you start from contains before you deduct the commission.
• Treating an estimate as a valuation. Neither Rentalizer nor the KeyBid PDF report is an ANEVAR valuation. For a mortgage, the bank requires a valuation by an authorised valuer.

Frequently asked questions

Which matters more: occupancy or average daily rate?

Neither on its own. High occupancy at a low rate and a high rate with low occupancy can bring in the same income. RevPAR, occupancy multiplied by the average daily rate, combines the two. For your own income, multiply the area's RevPAR by the nights you will actually offer, then deduct your costs.

Why do AirDNA and Crosspoint give different figures for the same city?

Because they cover different periods, 2025 and October 2025 – September 2026 respectively, report in different currencies and, very likely, count listings and revenue differently. The ranking of the cities is almost the same, so both sources are useful. Use them as the two ends of a range.

Can I use a revenue estimate in a mortgage application?

You can present it as your own analysis, but it does not replace the valuation the bank requires from an ANEVAR-authorised valuer. Raiffeisen notes that for property investment loans, potential rental income can be taken into account in some cases, so ask the bank early which documents it accepts. The details are in our guide to financing and due diligence.

How often is the data updated?

AirDNA updates its data monthly, with a one-month lag: the public pages of 6 October 2026 contain data up to September 2026. The Crosspoint report is annual. The weather forecast and events in KeyBid Puls look at the period ahead, so you keep checking them after you buy, every time you set your prices.