Reference

Updates

When each source was last collected, and what has changed on the site itself. Both are kept here so that anyone relying on a figure can see how recently it was looked at and how the way it is shown has changed.

Data collections

The most recent completed collections, one line per source per day, newest first. “Records” is how many rows the source file or service returned; “rejected” is how many of those failed validation and were not published. A collection that failed before loading anything leaves the previous data in place, so it is not listed. Each source’s own update schedule, terms and latest period are on the data sources page.

Changes to the site

Changes a reader could notice, dated by the day they reached the live site, newest first. Errors that readers reported, and what was done about each, are in the corrections log; how figures are calculated is in the methodology.

• September 26, 2026

  On a phone the header now tucks away as you scroll down and returns when you scroll up, giving back the fifth of the screen it held still (it stays put while you are using it). Similar places are now of comparable size (within five times), so a city is no longer listed beside towns a fifteenth of its size, and the comparisons a profile suggests are ones that make sense.

• September 26, 2026

  Added an occupation lookup to every profile (search the forty largest occupations in the metro and see what each pays) and an occupation comparison on the comparison page: choose a job and see its median pay in both metros next to the difference in prices, so “will my pay go further there” has an answer. Search now suggests the closest names when a place is misspelled (“raliegh”, “pheonix”), the Finder says which limit is in the way when few places fit and what loosening each would give, and the shortlist adds a gross rent yield row and the hazards each place is rated elevated or high for. The shortlist also stops shading the Schools score across states and says when a link named more than four places.

• September 26, 2026

  The Place Finder can now limit everyday prices (the Bureau of Economic Analysis regional price level, US = 100) and population growth over four years, alongside the existing rent, home value, property tax, migration, schools, climate, crime, airport and hazard limits. A place with no published figure is counted and left out, as with every other limit, never estimated.

• September 26, 2026

  Added saved searches to the Place Finder (name a search and reopen it later, in this browser only), a “Moving here from somewhere else?” search on every city profile that opens the comparison with the move worked out in your direction, and print or save-as-PDF buttons with a proper print stylesheet on profiles, comparisons and the shortlist.

• September 26, 2026

  Fixed layout and accessibility problems found by testing on phones and with a keyboard: the home page, state pages and methodology no longer scroll sideways; comparison tables, charts and ranking tables fit a phone screen (a ranking’s supporting columns move to a wider screen and the CSV); tap targets and input sizes are phone-sized; the menu closes when keyboard focus leaves it; tied ranks are read aloud; the Finder announces its new order and flags an amount that is not a number. Links inside long lists no longer prefetch, which cut about 40% of the bytes a profile page loads.

• September 26, 2026

  Added “If you are moving” to every comparison: choose where you live now and where you are going and it works out the difference in everyday prices (the Bureau of Economic Analysis regional price parity), the income that would keep the same purchasing power, the change in a new two-bedroom lease against your income, the median home against your income, and the swap in winters and summers. Nothing is estimated where a figure is missing, and the page says what it leaves out (taxes, commuting, schools, the job you would get). The comparison table gained HUD’s market rent and the price index.

• September 26, 2026

  Made the shortlist and comparison winners stricter. Nothing is shaded where two survey figures fall within their margins of error, proficiency is not shaded across states that give different tests, and a “more favorable” mark no longer appears between scores that print the same or on permit counts (which only measure size) or percentage changes on fewer than 300 units. Rankings show signals to one decimal so neighbors no longer look tied at different ranks, say when several places tie for first, and describe their pools as “the top N of M that meet these criteria”. The hazard ranking that was titled “Safest US Cities” is now “US Cities With the Lowest Natural-Hazard Risk”, because it measures floods, storms and fire and not crime. Several research reports now count their findings across every eligible city rather than from the rows their tables happen to show, and their printed differences match the scores beside them.

• September 26, 2026

  Corrected the school proficiency year. Figures were labeled 2019–20 but came from the 2020–21 pandemic-year file, which held a partial sample (about a third as many California schools as in 2018) and nothing at all for Washington, the District of Columbia or Puerto Rico. They now come from 2018–19, the last full year of state testing, which covers every state and territory and reaches high schools, so more of the largest schools in a county show a figure. School scores, the best-schools ranking and the reports that use them were recalculated.

• September 26, 2026

  New York City’s death rates are now the five boroughs combined, where they had been Manhattan’s alone (2.3 homicide deaths per 100,000 against 3.3 for the five together, with the Bronx at 7.1). A metro or micro area made of a single county, which repeated that county’s page with fewer figures, is no longer offered to search engines or carries advertising; it stays published and linked. Metro pages no longer list a missing county, county pages name the safety section only where it exists, and a “more favorable” or “best” mark no longer appears between two figures that print the same.

• September 26, 2026

  Shortened the notes under the school, housing, weather, safety and coverage sections of every profile. The long explanations of how to read those figures (why a proficiency column can show a dash, what a county death rate does and does not say, how air quality is measured) were printed in full on every page; they now appear once, in the Reading the numbers guide, and each profile keeps a sentence beside the figure and a link. The data-sources table on a profile no longer repeats each source’s update schedule, which the data sources page already lists. Also fixed the breadcrumb markup on research reports, which Search Console had flagged.

• September 25, 2026

  Added property taxes: the median real estate tax bill owners pay and the rate it works out to against the median home value, from the American Community Survey. The Finder can now limit the property tax rate. Property taxes are shown, not scored: the Area Score still has no tax component.

• September 25, 2026

  Added where people are moving, from IRS tax-return migration data: net domestic migration for every county, the share of arrivals who came from another state, and the average income of households moving in against those moving out. A city carries its county’s figure and says so. A research report looks at the counties where arrivals out-earn those leaving.

• September 25, 2026

  Added a research report on what each Place Finder limit does to a search: how many cities survive a rent ceiling, a winter floor or a hazard on its own and in combination, counted with the Finder’s own rules. Profiles now state the household income at which a new two-bedroom lease stays within 30% of income. The compare page suggests pairs of cities of comparable size in different metros, state pages open the Finder for that state, and the home page lists the tools.

• September 25, 2026

  Added home price change: how far prices have moved over one, five and ten years, from the Federal Housing Finance Agency’s repeat-sales index, with two new rankings (where prices have risen most, and where they fell over the past year). The index is published for metropolitan areas and states, so a city shows its metro’s figure and says so; the largest metros, which the agency publishes by division, show their division’s. Each ranking and each research table can now be downloaded as a CSV.

• September 25, 2026

  Corrected how the site describes itself. The methodology and editorial policy said a profile needs at least twelve measured values across four categories; the check has required forty across seven for some time, and those pages now read their numbers from the same place the check does. Also corrected the FAQ’s update schedule, the number of ZIP profiles allowed per city, which school grades the proficiency figures cover, and where advertising can appear.

• September 25, 2026

  Added a shortlist. Save up to four places from any profile or from the Finder and compare them side by side, with the best figure in each row marked and any county or metro figure labeled as such. The list is kept in your browser only, and a link carries it to someone else.

• September 25, 2026

  The Best Places Finder now takes must-haves as well as priorities: a rent ceiling from your income or a cap, a winter floor, limits on summer heat, unemployment, crime, school math and airport distance, and hazards to rule out. A place with no published figure for a limit is left out and counted, never estimated, and one switch keeps those places. You can sort on the figures themselves, see a table, download a CSV and share the search as a link.

• September 25, 2026

  Fixed site search for ZIP codes. Only about a tenth of ZIP pages were findable, so many ZIPs, including well-known ones, returned nothing even though their page exists; every ZIP page is now found. A metro area is also found from its second state (Chicago from Indiana), and search no longer reads far more of the database than it needs.

• September 25, 2026

  Added two research reports: how much of a city income ranking is noise once the survey’s margin of error is counted, and how often a city’s labor data is really its county’s. The Reading the numbers guide now shows when it was published and last updated.

• September 25, 2026

  ZIP profiles are no longer submitted to search engines. They remain published, are linked from their city page and are found by ZIP search, and they carry no advertising. Most of what a ZIP page shows is inherited from its city or county, so the city page is the one offered as the result.

• September 22, 2026

  The unemployment and population charts now say in their text description whether the series is rising, falling or holding steady. The sub-sections of a profile are marked up as headings, so a screen reader can move between them.

• September 19, 2026

  Where a place is unusually near to or far from a major airport, the Quick Read can now say so, with how the place ranks nationally. The cost-of-living sentence now says that housing drives most of the difference between places in the national index, which describes the index rather than the individual place.

• September 19, 2026

  Site search now finds Puerto Rico. State pages show access, climate, development and school figures. A profile with no housing data now explains why instead of leaving the section out, and air quality appears on pages that have no nearby weather station.

• September 19, 2026

  Where a table figure belongs to the county or metro area rather than the place itself, the row now says whose figure it is. Rankings built on older source data now state that data’s vintage. Added the Safest US Cities ranking.

• September 14, 2026

  Fixed broken comparison links between places that share a name. Fixed a trend calculation that labeled a rise from a zero baseline as flat. Fixed several mobile touch-target and overflow problems.