BatchData vs DataTree is most useful as a comparison of discovery, derived signals and property research access. Homebase ultimately judged the evaluated providers equivalent at the source-data layer for our local house-buying purposes. We therefore chose to build sifting, or derivatives, and stacking in-house around the buyer's strategy. Our October studies supported basic-fact corroboration, without making every label or product interchangeable.
If your question is which homes match a repeatable search, test search behavior and the underlying facts. If your question is why an owner or financing record looks unusual, test the relevant reports and documents. Neither job is settled by a national accuracy slogan.
TL;DR
- Homebase's source-data conclusion led us to build local derivatives and list construction in-house; it did not establish identical estimates, flags or coverage.
- Two separate October studies showed close agreement in matched building facts, but their samples and denominators must stay separate.
- Vacancy and financing labels diverged even where basic facts corroborated; a flag and a report-derived proxy are not equivalent truth tests.
- Evaluate search, API, report and document access in the package you would buy, because trial restrictions can obscure capability.
Decide whether you need discovery or an explanation
BatchData advertises property searches using owner, equity and sale-related criteria, with API and bulk delivery. That makes search behavior a concrete purchase-test question. BatchData search.
DataTree provides access to recorded instruments, including deeds and financing documents, subject to coverage. First American also offers API delivery. A DataTree comparison should therefore include both research access and integration requirements. Recorded documents, First American API.
| Job | What to compare | What would leave it unresolved? |
|---|---|---|
| Find homes in your intended audience | Same geography, property definitions and owner categories | Different eligibility rules mistaken for missing coverage |
| Confirm basic property facts | Matched properties and field-specific denominators | A large export count without confirmed matches |
| Understand a vacancy label | Definition, date and underlying evidence | Treating a flag as a confirmed empty house |
| Investigate ownership or financing | Relevant reports and recorded instruments | A summary that does not explain the exception |
| Run the workflow repeatedly | Refresh process, allowances, rights and total effort | Trial behavior assumed to represent a production contract |
Study A: broad corroboration, not independent accuracy
On October 8–9, 2026, one study exported 18,981 DataTree records across 12 requested neighborhoods. One neighborhood was blank, leaving 11 usable neighborhoods in Arizona and Illinois. The matched comparison covered 8,843 BatchData properties.
For floor area, 99.97% of the 8,830 populated comparisons agreed within 5%. Tax-bill amounts agreed on all 8,804 populated comparisons. These are agreement findings: two records can corroborate one another without either being independently verified against the property or a current account.
The field denominators differ because populated values differed. The 18,981 exported records were not all matched properties, and neither count is the denominator for every field. This trial also involved 25 TotalView reports and five detail reports, with access restrictions limiting what could be investigated.
Study B: aligned building facts did not align every signal
A second October 8–9 study compared 5,859 saved BatchData candidates with a broader DataTree result set. It matched 5,857 within the neighborhood searches; one additional property was found by separate address lookup and one remained unresolved. Within the 5,857 matched set, floor area agreed on 5,856 properties.
Vacancy flags illustrated a different problem. BatchData flagged 60 matched properties and DataTree flagged 20, with only six shared. That is disagreement between signals, not independent confirmation of which houses were vacant.
Financing comparisons also required care: a BatchData free-and-clear flag was compared with a zero-open-liens proxy derived from DataTree. Those definitions were not equivalent. A disagreement could not establish which property had no enforceable debt.
Forty search sets completed, but a planned history review of 245 properties produced no additional usable TotalView reports because the trial refused them. These figures summarize saved historical reports, not an independent re-audit of every original response.
Why we kept local strategy in-house
The evidence moved our attention from finding a supposedly exclusive basic fact to deciding how facts should support a local buyer. Equal source data for our use did not mean equal transformations or identical record coverage.
Our Source, Sift, Stack, Skip framework starts with the neighborhood, then the audience: owner-occupants or landlords. Each motivation gets its own list. Combine lists, remove duplicates and do-not-mail records, and refresh before each consistent send. Requiring every motivation at once can exclude owners who belong on a single relevant list.
Local city records and obituary research can add coverage absent from a national feed. That is separate from asking a vendor to calculate another label from county facts.
For your buying test, request identical local properties and a few known exceptions from both providers. Record field dates, search exclusions and unavailable reports separately. Choose the access that resolves the work you actually have, while keeping your mailing strategy explicit and your unresolved ownership or debt questions open.
Direct mail to the neighborhoods where you want to buy
Homebase handles direct mail to property owners in the neighborhoods where you want to buy. Enter one address to find your neighborhood and get your first neighborhood analysis free. No card required.
Analyze a neighborhood free