San Francisco has world-class food, a temperate climate, and some of the highest salaries in the country. It scores 38.2 out of 100.
Liberal, Kansas — population under 19,000 — scores 79.3. Neither number is a bug, and neither is a hot take. Both come out of the exact same formula, applied consistently. Here's what's actually happening underneath, and what it does and doesn't tell you about where to live.
Why San Francisco scores 38.2 despite being world-class
WhereAtHome's default score weights Housing Affordability at 20% and Income vs Housing at another 20% — 40% of the total, combined, before anything else is counted.
San Francisco's median home value is about $1.4M against a household income around $137K. Run those through the actual formulas and Housing Affordability lands around 15 out of 100; Income vs Housing lands around 27 out of 100. That's 40% of the score sitting in the teens and twenties before Walkability, Crime, Commute, or Purchasing Power get a vote.
Purchasing Power — not "economic stability," which is what this component used to be called before a correctness pass renamed it — measures income against the national benchmark, adjusted for cost of living when it's available. It does not measure job diversity, industry mix, or how strong the local economy actually is.
San Francisco's very high cost-of-living index (195, against a national baseline of 100) pulls that component down too, even though its raw income is well above average. Add up all six components at their fixed weights and 38.2 is what falls out. No amount of shuffling the remaining 60% of the weight rescues a home-price ratio that far underwater — the math simply doesn't have enough room left.
Why Liberal, Kansas scores 79.3
Liberal's median home value is about $147K against a median income of about $56K. That combination maxes out both housing components near 100 — the same two components worth 40% of the score, at the ceiling instead of the floor. Its average commute is 13.9 minutes, which scores 91 out of 100 against the national commute benchmark.
Here's the part that matters for reading any ranking correctly: Liberal has no recorded Walk Score.
Rather than defaulting a missing input to zero — which would look catastrophic — or to some assumed "average," WhereAtHome's scoring engine excludes the missing component entirely and renormalizes the remaining weights so they still sum to 100%. Walkability's 15% doesn't vanish and doesn't get punished; it gets redistributed proportionally across the components that do have data. That's a deliberate design rule, applied the same way to every city with a coverage gap, not a special case built for this example.
San Francisco and Liberal, component by component
Seeing the six numbers side by side makes the gap concrete instead of abstract:
| Component | Weight | San Francisco | Liberal, KS |
|---|---|---|---|
| Housing Affordability | 20% | ~15 | 100 |
| Income vs Housing | 20% | ~27 | 100 |
| Commute | 15% | ~33 | 91 |
| Crime | 20% | 35 (Moderate-High) | 50 (Moderate) |
| Walkability | 15% | 89 | excluded — no data |
| Purchasing Power | 10% | ~47 | ~37 |
San Francisco wins Walkability outright and edges ahead on Purchasing Power. Liberal wins everything else — including the two components worth the most combined weight, by nearly the maximum possible margin. That's the entire 38.2-versus-79.3 gap, in six rows.
The same six components, wildly different outcomes
Both cities run through identical math: Housing Affordability and Income vs Housing at 20% each, Crime at 20%, Commute at 15%, Walkability at 15%, Purchasing Power at 10%. This is a known challenge with any composite score, not just this one — the OECD's handbook on constructing composite indicators treats indicator selection, weighting, and aggregation as design choices, not neutral math.
Crime itself runs on seven tiers now, not six: Very Low, Low, Moderate-Low, Moderate, Moderate-High, High, and Very High, each mapped to a fixed safety score from 95 down to 10. The FBI itself cautions against using reported crime data to rank jurisdictions without accounting for local reporting practices and demographics — a caution that applies just as much to a single crime tier feeding into a bigger score.
Same formula, same weights, same tier definitions for a 19,000-person Kansas town and a global tech capital. The score doesn't know San Francisco is famous. It only knows what the six numbers say.
Some of those six components also lean on the same underlying numbers rather than six fully independent signals. Purchasing Power and Income vs Housing both start from median household income — one compares it to a national benchmark, the other compares it to the local home value. That's not double-counting exactly, since the two questions are genuinely different, but it does mean a single unusually high or low income figure has more influence over the final score than a glance at six separate percentages would suggest.
Why the ranking flips when you change what matters
Fixed weights are one model, not a law of nature. Give Walkability more influence and San Francisco's 89 becomes a much bigger asset relative to a place like Liberal with no recorded score at all. Give Housing Affordability more influence instead and the gap between these two cities only widens.
The city comparison guide works through exactly this: the same five cities produce three different leaders depending on whether you sort by the blended score, by crime, or by walkability alone.
Missing data doesn't mean zero — but it isn't invisible either
Coverage gaps are common enough to check for on every city you're seriously considering, not just the small ones. Before trusting a score, look at which components actually had data. A city missing its walkability score and one other input is being ranked on four components carrying the full weight of six — which can flatter or hurt a city relative to one with complete coverage.
The fix isn't to distrust every score; it's to check the component breakdown before comparing two finalists as if every input were equally complete.
That check is easier than it sounds. Every metric on a WhereAtHome city page lists its source and year right next to the number — a Census Bureau figure from 2023, a Zillow index updated this year, or a placeholder still waiting on real data. If a metric's source looks thin or its year is old, treat that component's contribution to the total with proportionally less confidence, the same way you would discount a single stale price in an otherwise current spreadsheet.
How to actually use a ranking
Treat any ranking, including this one, as a shortlist tool, not an answer key:
- Use it to cut a long list down, not to make the final call.
- Check the component breakdown, not just the headline number — a close overall score can hide one large gap and several near-ties.
- Sort by what matters to you on the rankings page instead of trusting the default weights, which are one reasonable choice among many.
- Confirm coverage — know which inputs a given city actually has data for before you treat its score as complete.
Full detail on every formula lives on the methodology page.
San Francisco and Liberal aren't really being compared against each other by most people reading this — almost nobody is actually choosing between a global tech hub and a 19,000-person Kansas town. That's exactly why they're useful together: at the extremes, it's obvious that the formula measures what it measures and nothing more. The same caveats apply, just less visibly, to any two cities that are actually on your list.
