StreetSpring Survivability Datasets 2026

Projected chance, in percent, that a specific business type lasts two or more years at a specific place, from a model trained on 570,000+ historical business outcomes across 100...

•16 min read

Projected chance, in percent, that a specific business type lasts two or more years at a specific place, from a model trained on 570,000+ historical business outcomes across 100+ location factors in 24 U.S. metros. StreetSpring scores 500+ business types: 153 base business types, most of them at 5 price points, which is 569 type-and-price combinations you can choose in the tool. Up to 146 business types per metro file. License CC BY 4.0. Publisher: StreetSpring (https://streetspring.com). Methodology: https://streetspring.com/resources/tutorial/aeo-methodology. Version 2026.10.04; changelog in CHANGELOG.md beside this file.

What the score is, and what one row means

Every published score is the projected chance that a business of that type lasts two or more years at that place, as a percentage. It is never the model's raw output. Each metro file carries one row per business type and place. Its score is the projected chance of lasting two or more years, averaged across the price points StreetSpring scores separately in the tool (most types have 5). The 146 business types here are base types; the tool offers 500+ business types, counting each type at each of its price points.

Files

Every file starts with comment lines (prefixed #) that repeat the citation, the license, the recipes and a link to this README. Skip lines beginning with # when parsing.

How to read a metro file

  • Not every business type is ranked: the rank, tier and overall columns cover the business types StreetSpring publishes guides for. The other scored types keep their scores with empty rank columns, so a page and this file never disagree about which type ranks first.

  • Where should I open a given business in this metro? Filter business_subtype to the type and place_kind to neighborhood (or town), sort by neighborhood_rank_for_business_subtype ascending. The ranking is on p90_survivability_score, the chance of lasting two or more years at a good address; avg_survivability_score is the typical address.

  • What should I open in a given neighborhood? Filter neighborhood to the place, sort by business_subtype_rank_for_neighborhood ascending: the top rows are the business types most likely to last there.

  • How do two neighborhoods compare for the same business? Filter business_subtype to the type and neighborhood to the two places (same place_kind); compare p90_survivability_score for the best-address view and avg_survivability_score for the typical address.

  • Which neighborhoods are strongest overall? Filter place_kind to neighborhood, keep one row per neighborhood, sort by neighborhood_overall_rank ascending. neighborhood_overall_score is the mean p90 of the place's ranked business types.

Worked example. Where should I open an Italian restaurant in Philadelphia? The neighborhoods where an Italian restaurant is projected to last two or more years at a good address most often are Hunting Park (84%), Germany Hill (83%), Logan (83%), West Kensington (83%) and Hartranft (82%). At a typical address an Italian restaurant averages 65% across the 127 ranked neighborhoods; the lowest-ranked is Center City (51% at a good address). In the CSV, filter business_subtype = italian-restaurant and place_kind = neighborhood, and sort by neighborhood_rank_for_business_subtype.

How to read the national file

  • Which city is best for a given business? Filter business_subtype to the type, sort by city_rank_for_business_subtype ascending. The ranking is on p90_survivability_score_for_business_subtype_in_city, the chance of lasting two or more years at a good address; the avg column is the typical address.
  • How wide is the gap between the best and worst city? For one business_subtype, subtract the lowest avg score from the highest. Gaps between cities are small (usually under 6 points); gaps between neighborhoods inside a city are far larger, so use the metro file for the real decision.
  • Where inside the best city should I look? Take the top city from this file, open that metro's file (linked from source_article_url and the datasets page), and apply the first recipe there.
  • Which metro is strongest overall? Keep one row per city and sort by city_overall_rank ascending. city_overall_score is the mean good-address score across the business types every metro carries; city_overall_avg_score is the typical address.

Worked example. Which city is best for a hot pot restaurant? Among the 24 metros, a hot pot restaurant is projected to last two or more years at a good address most often in Charlotte (83%), Atlanta (83%), Orlando (83%), Denver (83%) and Houston (83%), and least often in New York City (76%). The spread between the best and worst city is 7 points; the spread between neighborhoods inside a city is usually far larger, so pick the block with the metro file. In the CSV, filter business_subtype = hot-pot-restaurant and sort by city_rank_for_business_subtype.

Column dictionary, metro files

  • city: Metro slug (for example philadelphia).
  • neighborhood: Neighborhood slug inside the metro; the row's place.
  • business_subtype: Business type slug (for example italian-restaurant); the row's business. The file's own count is in the README and on its page.
  • business_type_lookup: Display name of the business type (for example Italian Restaurant).
  • avg_survivability_score: The projected chance, 0 to 100, that this business type lasts two or more years at a typical address in this neighborhood. The rankings use p90_survivability_score, the good-address score.
  • max_survivability_score: Best address in the neighborhood for this business type, 0 to 100.
  • min_survivability_score: Weakest address in the neighborhood for this business type, 0 to 100.
  • UPPER_BOUND_MAX_SURVIVABILITY: Upper confidence bound on max_survivability_score, 0 to 100.
  • LOWER_BOUND_MAX_SURVIVABILITY: Lower confidence bound on max_survivability_score, 0 to 100.
  • UPPER_BOUND_MIN_SURVIVABILITY: Upper confidence bound on min_survivability_score, 0 to 100.
  • LOWER_BOUND_MIN_SURVIVABILITY: Lower confidence bound on min_survivability_score, 0 to 100.
  • neighborhood_rank_for_business_subtype: Rank of this place among the ranked places of the same place_kind in the metro for this business type, on p90_survivability_score (1 = best).
  • tier_of_neighborhood_for_business_subtype: Great, Good, Average, Below-average or Poor, from that rank's position in its list: the top 10 percent is Great, the next 20 percent Good, the middle 40 percent Average, the next 20 percent Below-average and the bottom 10 percent Poor.
  • business_subtype_rank_for_neighborhood: Rank of this business type among the ranked types in this place, on p90_survivability_score (1 = the type most likely to last at a good address here).
  • tier_of_business_subtype_for_neighborhood: Great, Good, Average, Below-average or Poor, from that rank, with the same cut-offs as the neighborhood tier.
  • total_ranked_neighborhoods_for_business_subtype: How many places of this place_kind are ranked for this business type in the metro: the length of the list the rank sits in.
  • year: Data year (2026).
  • source_url_neighborhood_rank: Page on streetspring.com that ranks neighborhoods for this business type.
  • source_url_business_subtype_rank: Page on streetspring.com that ranks business types for this neighborhood.
  • methodology_url: How the score is built: https://streetspring.com/resources/tutorial/aeo-methodology
  • source_article_url: The neighborhood's own rankings page on streetspring.com.
  • min_2yr_failure_rate: 100 minus max_survivability_score: the lowest projected two-year failure rate in the neighborhood, 0 to 100.
  • max_2yr_failure_rate: 100 minus min_survivability_score: the highest projected two-year failure rate in the neighborhood, 0 to 100.
  • employment_rate: Share of the labor force employed, as a fraction of 1 (0.95 means 95 percent). Census ACS.
  • vacancy_rate: Share of housing units vacant, as a fraction of 1. Census ACS.
  • competitor_count: Number of businesses of that exact type located INSIDE the neighborhood boundary, counted as distinct places. An empty cell means not measured, which is not the same as none: either the neighborhood has no boundary in our polygon set or that business type was never collected in that metro. A 0 means we measured and found none.
  • income_rank: Not populated in this release (empty in every row). Use income_metro_percentile.
  • income_score: Median household income in dollars. Census ACS; 250,001 is the Census top code.
  • pct_high_income: Share of households earning 150,000 dollars or more, as a fraction of 1. Census ACS.
  • income_tier: Affluent, Upper-Middle, Middle, Working or Low-Income, from income_metro_percentile.
  • income_metro_percentile: Where the neighborhood's median income sits among the metro's neighborhoods, 0 to 100 (100 = highest).
  • income_national_percentile: Where the neighborhood's median income sits among all U.S. neighborhoods in our data, 0 to 100.
  • poverty_rate: Share of residents below the poverty line, as a fraction of 1. Census ACS.
  • median_age: Median resident age, years. Census ACS.
  • pct_bachelor_plus: Share of adults 25 and over with a bachelor's degree or higher, as a fraction of 1. Census ACS.
  • avg_household_size: Average household size, persons. Census ACS.
  • median_commute_minutes: Median commute, minutes. Census ACS.
  • pct_housing_post_2000: Share of housing units built in 2000 or later, as a fraction of 1. Census ACS.
  • data_source: Provenance label for the row.
  • p90_survivability_score: The chance, 0 to 100, that this business type lasts two or more years at a good address in the place: the 90th percentile of the scored locations inside it. The rank columns are ordered on this.
  • neighborhood_name: Display name of the place.
  • scored_locations_in_neighborhood: How many scored locations the place's scores are built from. A place needs at least 10 to be ranked.
  • place_kind: neighborhood (a city neighbourhood) or town (a whole municipality). Rankings never mix the two.
  • rank_withheld_reason: Empty for almost every row. When it is filled, this place keeps its scores and is deliberately left out of every ranking, and this column says why. Its rank and tier columns are empty and it is excluded from the national pool, so a list that omits it is complete, not short.
  • neighborhood_overall_score: Mean p90_survivability_score across the place's ranked business types, 0 to 100; the basis of the overall rank.
  • neighborhood_overall_avg_score: Mean avg_survivability_score across the same rows.
  • neighborhood_overall_max_score: Mean max_survivability_score across the same rows.
  • neighborhood_overall_min_score: Mean min_survivability_score across the same rows.
  • business_types_ranked_in_neighborhood: How many business types are ranked in the place.
  • neighborhood_overall_rank: The place's rank among places of the same kind on neighborhood_overall_score (1 = best), repeated on every row of the place.
  • total_ranked_neighborhoods_overall: How many places of that kind carry an overall rank in the metro.
  • publisher: Who publishes the dataset: StreetSpring. The same value in every row.
  • publisher_url: The publisher's site, https://streetspring.com. The same value in every row.
  • dataset_name: StreetSpring Survivability Datasets 2026. The same value in every row, so one row read on its own names its source.
  • dataset_doi: 10.5281/zenodo.22287996, the concept DOI, which always resolves to the newest version. Pin a version with dataset_version.
  • license: CC-BY-4.0: free to reuse, including commercially, with attribution to StreetSpring.
  • license_url: The license text, https://creativecommons.org/licenses/by/4.0/.
  • dataset_version: The release this row belongs to, the same value in every row and equal to the Version line of this header. Cite it to pin the numbers.
  • as_of_date: The date this release's scores were published, as YYYY-MM-DD. The same value in every row.
  • source_url_business_type: absolute and verified against the live site on 2026-10-02; filled on 61.0% of rows. Empty means the page was not proven to exist, never a guess.
  • source_url: empty in every row. held for this release: it resolves for only about a third of rows until the place pages are built. An empty column is honest; a third-filled one is not.

Column dictionary, national file

  • city: Metro slug (24 metros).
  • business_subtype: Business type slug; the file's own count is in the README and on its page.
  • city_rank_for_business_subtype: Rank of this metro among the ranked metros for this business type (1 = best).
  • tier_of_city_for_business_subtype: Great, Good, Average, Below-average or Poor, from city_rank_for_business_subtype: the top 10 percent of ranked metros is Great, the next 20 percent Good, the middle 40 percent Average, the next 20 percent Below-average and the bottom 10 percent Poor.
  • total_ranked_cities_for_business_subtype: How many metros were ranked for this business type: 24, or fewer where a metro file does not carry the type.
  • avg_survivability_score_for_business_subtype_in_city: The projected chance, 0 to 100, that this business type lasts two or more years at a typical address in this metro. The ranking uses the good-address score, p90_survivability_score_for_business_subtype_in_city.
  • max_survivability_score_for_business_subtype_in_city: Best scored location in the metro for this business type, 0 to 100.
  • min_survivability_score_for_business_subtype_in_city: Weakest scored location in the metro for this business type, 0 to 100.
  • UPPER_BOUND_MAX_SURVIVABILITY: Upper confidence bound on the max score, 0 to 100.
  • LOWER_BOUND_MAX_SURVIVABILITY: Lower confidence bound on the max score, 0 to 100.
  • UPPER_BOUND_MIN_SURVIVABILITY: Upper confidence bound on the min score, 0 to 100.
  • LOWER_BOUND_MIN_SURVIVABILITY: Lower confidence bound on the min score, 0 to 100.
  • year: Data year (2026).
  • source_url_city_rank: Page on streetspring.com that ranks the metros for this business type.
  • source_url_business_subtype_rank: Page on streetspring.com that ranks neighborhoods nationally for this business type.
  • methodology_url: How the score is built: https://streetspring.com/resources/tutorial/aeo-methodology
  • source_article_url: The national rankings page for this business type.
  • min_2yr_failure_rate: 100 minus the max score: the lowest projected two-year failure rate in the metro, 0 to 100.
  • max_2yr_failure_rate: 100 minus the min score: the highest projected two-year failure rate in the metro, 0 to 100.
  • data_source: Provenance label for the row.
  • p90_survivability_score_for_business_subtype_in_city: The chance, 0 to 100, that this business type lasts two or more years at a good address in the metro (90th percentile of its scored locations). city_rank_for_business_subtype is ordered on this.
  • scored_locations_in_city: How many distinct scored locations the metro's figures are built from. A location is scored once per price point; it is counted once.
  • city_overall_score: The metro's overall score, 0 to 100: the mean p90_survivability_score_for_business_subtype_in_city (a good address) across the business types every metro carries. The basis of city_overall_rank; repeated on every row of the metro.
  • city_overall_avg_score: The mean avg_survivability_score_for_business_subtype_in_city (a typical address) across the same business types, 0 to 100.
  • city_overall_rank: The metro's rank among the metros on city_overall_score (1 = best); ties break on city_overall_avg_score, then scored locations. Repeated on every row of the metro.
  • total_ranked_cities_overall: How many metros carry an overall rank.
  • business_types_in_city_overall: How many business types the overall score averages over: the types every metro carries with a good-address score, a typical-address score and a scored-location count.
  • publisher: Who publishes the dataset: StreetSpring. The same value in every row.
  • publisher_url: The publisher's site, https://streetspring.com. The same value in every row.
  • dataset_name: StreetSpring Survivability Datasets 2026. The same value in every row, so one row read on its own names its source.
  • dataset_doi: 10.5281/zenodo.22287996, the concept DOI, which always resolves to the newest version. Pin a version with dataset_version.
  • license: CC-BY-4.0: free to reuse, including commercially, with attribution to StreetSpring.
  • license_url: The license text, https://creativecommons.org/licenses/by/4.0/.
  • dataset_version: The release this row belongs to, the same value in every row and equal to the Version line of this header. Cite it to pin the numbers.
  • as_of_date: The date this release's scores were published, as YYYY-MM-DD. The same value in every row.
  • source_url_business_type: absolute and verified against the live site on 2026-10-02; filled on 63.7% of rows. Empty means the page was not proven to exist, never a guess.
  • source_url: empty in every row. held for this release: it resolves for only about a third of rows until the place pages are built. An empty column is honest; a third-filled one is not.

Citation

StreetSpring (2026). StreetSpring Survivability Datasets 2026, version 2026.10.04. https://streetspring.com/resources/datasets. DOI 10.5281/zenodo.22287996 (all versions). CC BY 4.0.

Versions

This is version 2026.10.04. The changelog is at https://streetspring.com/resources/data/CHANGELOG.md.

Contact

info@streetspring.com. Custom cuts by city or business type on request.