The average is a fiction: one county, 288 neighborhoods, a 13x income spread

Travis County, Texas — Austin and its inner suburbs — has a median household income around $99,000 at the neighborhood level. Quote that number in a site package and it sounds like one place: prosperous, tech-fueled, comfortably affluent.

Now look at the county the way the Census Bureau actually measures it: 288 census tracts. The lowest-income tract sits at $18,286. The highest reports $250,001 — and that's a ceiling, because the ACS stops counting there. A thirteen-fold spread, inside one county, inside one brand of "Austin."

Nobody's customer base is the county. A store's trade area is a handful of tracts, and whether those specific tracts are the $18k kind or the $250k kind is the entire ballgame. The county average describes neither — it describes a blended fiction that no customer actually lives in.

The resolution ladder

Demographic data comes at several resolutions, and each step down the ladder averages away more of the truth:

• Metro / county — fine for market-entry questions ("should we be in Austin at all?"), useless for a site. The 13x spread above is the proof.
• ZIP code — the most commonly quoted and the most misleading, because ZIPs were drawn for mail routes, not neighborhoods. A single ZIP routinely staples together an apartment district and an estate section; their average describes neither. Every flaw in the county number applies, just less visibly.
• Census tract — roughly 4,000 people, drawn to follow actual neighborhood boundaries. This is the resolution where "who lives near this address" becomes a real answer.
• Block group — a tract subdivision, 600–3,000 people. The finest public resolution; the right level for tight-radius categories.

The pattern in listing packages is predictable: the number quoted comes from whatever resolution flatters the site. A mediocre corner in a strong county gets the county figure. We've never once seen a broker one-pager quote the block group.

The two-question test

When a demographic claim shows up in a deal — "median income $95k," "growing area," "young professionals" — two questions expose it: at what resolution? and would the number survive one step down the ladder? The New Braunfels teardown we ran in June was exactly this test in action: a booming two-county market containing tracts from $39,755 to $173,969, ages 29 to 58. Every marketing sentence about that market was true somewhere in it and false somewhere else.

Tract- and block-group-level data is public, current, and free. The only reason area-level averages persist in deal packages is that pulling the fine-grained version used to take real work. It doesn't anymore — a demographic report resolves any address to its actual neighborhood in about 30 seconds, which is a fast way to find out which of the 288 versions of the county you're actually buying.