Airbnb Lead Sources Compared: Scraped Listings, Data Brokers and Permit Records

Four kinds of company will sell you "Airbnb data," and they are selling four different things. Most disappointment in this market comes from buying one and expecting another.

We sell one of these, so read the last section with appropriate scepticism. The comparison itself is straightforward enough that it does not need slanting.

The four categories

Listing scrapers

Scraped listing data is genuinely useful for the job it does. If you are building a pricing tool, benchmarking amenities, or trying to understand what inventory looks like in a market, listing data is the primary source and there is no substitute.

What it cannot do is tell you who to call. A listing gives you a first name, a profile photo and a location circle a few hundred metres wide. The platform deliberately withholds identity, and every technique for recovering it — photo matching, review mining, parcel triangulation — is slow, error-prone and expensive relative to the alternatives.

There is also a durability problem. Scraping at volume runs against platform terms and anti-bot measures that change without notice. A pipeline that worked last quarter may not work this quarter.

Buy it if you need to understand inventory and pricing. Don't buy it as a contact list, whatever the sales page implies.

Market analytics

AirDNA is the best-known name here and the product is good. It answers questions like: what does a two-bedroom in this zip code earn, how seasonal is this market, is occupancy trending up or down.

These tools are built for investors deciding whether to buy, and they are honest about it. The confusion is entirely on the buyer's side — people hear "short-term rental data" and assume contact records, then find a market dashboard.

The estimates are also estimates. They are modelled from listing and booking signals, and they are reasonable at market level and progressively less reliable as you drill down to individual properties. That is fine for their purpose and worth knowing.

Buy it if you make investment or pricing decisions. Don't buy it for a sales team.

General data brokers

This is the category to be careful in, because the products are opaque by design.

The pattern is a consumer marketing database — assembled from purchase histories, surveys, public records and other broker files — matched against some short-term-rental signal, then packaged as "Airbnb host leads." Sometimes there is a real signal underneath. Often the match is loose, the file is years old, and the same rows have been sold to everyone in your industry already.

Two warning signs. First, unwillingness to name a source: a vendor who cannot tell you which register or filing a record came from is telling you it did not come from one. Second, contact rates that are impossibly high — a "97% phone match" on a dataset whose underlying public records mostly do not publish phone numbers means an append vendor guessed, and appended phone numbers in this category are frequently the wrong person entirely.

A wrong number is not neutral. It costs a call, it annoys a stranger, and under the wrong circumstances it creates a compliance problem you did not budget for.

Buy it if you have tested a sample and it holds up. Don't buy it on the strength of a coverage claim alone.

Permit records

Which brings us to what we do.

Permit data has one structural advantage over everything above: the operator identified themselves to a government agency, under their own name, at a specific address, on a specific date. Nothing is inferred. There is no matching step where errors creep in.

It also has four real limitations, and any vendor who does not volunteer them is not being straight with you.

Coverage follows administration, not market size. Permitting is local, so data quality tracks the publishing decisions of individual records offices. Florida licenses vacation rentals at the state level under Florida Statute 509.241, so its coverage is statewide. California has no state program, so its file is assembled city by city and county by county. Some large markets publish nothing machine-readable at all.

Contactability varies enormously. This is the big one. Of 302,337 records in our database, 104,962 carry a phone or email — and that third is not evenly spread. Louisiana and Wisconsin publish contact details on most records. Texas publishes almost none. Both are useful; they support completely different outreach.

Unpermitted operators are invisible. People who rent without registering do not appear in a register. In markets with hard caps and long waitlists, that population is not trivial.

No performance data. A permit record cannot tell you whether a property books solidly or sat empty. If you need that, you need the analytics category, and the two are complements rather than substitutes.

And two states have nothing at all: Idaho and Iowa have both preempted local short-term-rental permitting, so no registers exist. Detail here.

How to choose

The question is not which category is best. It is which one answers your question.

"Should I buy a property in this market?" — Market analytics.

"What are competitors charging and what amenities do they list?" — Listing data.

"Who runs short-term rentals here and how do I reach them?" — Permit records.

"I need 50,000 rows and I don't much care about accuracy." — A broker will sell you that, and you should test a sample before you scale.

Plenty of companies use two of these together. Analytics tells you which markets are growing; permit data tells you who to call in them. That combination is more sensible than trying to make either one do the other's job.

Questions worth asking any vendor

• Where does each record come from? A specific answer — a named register, agency or portal — or the deal is off.
• When was it last refreshed, and how? "Continuously from the source" and "we bought a file in 2023" are different products.
• What percentage has a phone or email, in my states? National averages hide everything that matters.
• Are contact details from the source or appended? Appended data in this category has a poor accuracy record.
• What are you missing? An honest vendor has a ready answer. A vendor claiming complete national coverage of local permit data is describing something that does not exist.

Ours, for the record: 46 states rather than 50, roughly a third of records contactable with wide state-by-state variation, nothing on unpermitted operators, and no revenue estimates. The per-state numbers are all published.