From tile maps to night lights to TV markets - and segments you can build yourself on a Mac.

Note: This is a public method demo with Bruce Boston and Tim Ellis. The shipping product is turbolapper - AFM (Apple Intelligence on-device; data stays on your Mac). See requirements and Technology.

How should you map the United States?

A Monday conversation that spilled into Team Data Crunch: Bruce and Tim were mid-thread on US demographic views - not a polished slide deck, just the question every operator eventually hits.

If you run an online store, a multi-location business, or a hotel that draws guests from across the country, you eventually ask: where is the activity coming from? The default answer is often a state-by-state tile map. It is useful. It is also misleading.

Summary

• Data: Public US geography and demographics - state layouts, night lights, Nielsen-style TV markets (DMAs), and segment-style ranks built from open sources.
• Method: Compare views of the same country: equal-state tiles → lights-at-night photo → TV market polygons → bubble maps that overindex personas by market.
• Result: California is not one block. Markets beat equal tiles for marketing reality. You can push work that used to need warehouse teams onto a Mac with turbolapper’s engine.
• You get: A sharper map of where people actually live and watch, plus DIY segments without a Claritas-scale invoice.
• Next: Watch the session · Join the waitlist

View 1 · Equal state tiles

The session starts with a familiar tile grid of the 50 states (the kind of layout you can build in Google Sheets or Excel from guides like geographyrealm.com).

Why people use it: simple, printable, easy to drop a metric per state.

What it gets wrong: Connecticut can look about the same size as Texas or California. Population, complexity, and commercial reality do not match the square. California is not a single story - and a one-cell state map pretends it is.

That is a fine starter view. It is a weak final answer for “where should I spend?”

View 2 · Lights at night

Bruce’s favorite population picture is not a chart at all: US lights at night (the classic satellite view of humanity as power).

What the photo carries that a state tile cannot:

• Where people actually live - intensity, corridors, emptiness
• Ports and industrial edges that light up beyond residential density
• Interstates as man-made spines of activity
• Geographic hard lines - that dim “you shall not pass” band through the interior before Salt Lake, Denver/Boulder, Vegas, and the California valleys come back on

Humanity becomes a proxy for power use (and vice versa). You will not map bat colonies this way - but you will see Austin - San Antonio corridors, Phoenix vs the dimmer interior, and migration hints if you compare decade stacks later (NASA keeps historical series; a future episode if the Slack crowd asks).

View 3 · Nielsen TV markets (DMAs)

To move from “pretty picture” toward decision grain, the session pulls Nielsen TV markets - the mid-century answer to “who shares the same local news?”

Born in the broadcast era (Wikipedia lore in-session: I Love Lucy / $64,000 Question period), markets group counties by who watched which local stations. Cable often followed the same county boundaries. Marketers still use them: when McDonald’s bought a local news spot, which counties were expected to show up the next day?

Why this beats a state tile:

• California is many markets - roughly a dozen in the session cut (Eureka, Chico, San Francisco, Sacramento … Yuma / El Centro spanning CA - AZ agriculture).
• Phoenix vs Tucson - Phoenix’s footprint shows how far people care about that market; Tucson’s coverage is smaller for a reason.
• Wichita - Hutchinson vs Topeka - borders that still matter for local media and cable news.

Classic pipeline Bruce describes: ZIP → county → DMA, then spend and response by market. That is the analytics motion retailers and media teams have run for decades - now explorably local on a Mac.

View 4 · DIY segments & overindexes

Commercial segmentations (e.g. Claritas-style socio-economic × life-stage groups) cost real money. The provocation in the session: can you approximate useful nuance with public data and your own engine?

Tim’s reminder that matters for every “college town” or “retirement state” cliché:

> No town is only one thing. Chico has students and retirees and families. Density is a mix, not a single label.

Bruce’s maps move from equal tiles to bubbles sized by population, then into persona / segment overindexes by market - names in the session include flavors like solo sunsetters and golden cruisers. Hover a market and you see which segments overindex; Florida is not only “retirement” - Jacksonville singles, Miami migration intensity, Panama City domestic migration at the top of the distribution in the cut they hovered.

Ideas parked for a future crunch (if Slack wants them): multi-segment pie bubbles for the top three overindexes per market.

What Local Analytics means here

This episode is less “one SQL hero query” and more view design:

1. Start with the decision - concentration of customers, ad response, migration, where to put the next store.
2. Change the map until the grain matches the decision - state tiles → lights → DMAs → segments.
3. Keep the corpus local while you explore - the engine Bruce used is the turbolapper stack on a Mac; the shipping product is turbolapper - AFM (Apple Intelligence; keeps data on your Mac).
4. Stack questions for free on the app side - the second and third cut is where tile-map lies get caught.

Bruce’s career framing: work that once needed BigQuery / Teradata and a team of five becomes “I wonder if I can do that locally now?” Tim’s product framing: they are building the free-to-start Mac app so you can run the same motion - engine first, App Store UI catching up in the session’s rough preview.

How to run a version of this yourself

1. Pick one business question (e.g. “where do my US customers concentrate?”).
2. Load state-level metrics - then admit where equal tiles mislead.
3. Overlay a richer geography (DMA / metro / county) from public sources.
4. Build 3–5 named segments from public demographics - not 60 commercial codes on day one.
5. Overindex segments by market; label confidence when sample is thin.
6. Keep proprietary CRM joins on-device when you move off open data.

FAQ

What does this demo show?

How Team Data Crunch explores US geography: equal-state tiles, lights-at-night, Nielsen TV markets (DMAs), and DIY segment overindexes - all as Local Analytics method on a Mac. Watch: YouTube · Community.

What is a DMA / TV market, and why use it instead of state?

A DMA (designated market area / TV market) groups counties that historically shared the same local broadcast news. Marketing still uses them for local spend and response. States over-smooth places like California; markets keep commercial grain.

Is this a Claritas or Nielsen replacement?

The session shows inspiration and DIY public-data segments you can inspect yourself. Use proprietary vendors when you need their taxonomy; use Local Analytics when you want private exploration and custom cuts on your Mac.

Can I do this in turbolapper - AFM?

Yes - that is the product direction. Confirm macOS 26 + Apple Intelligence requirements, then join the waitlist. Related method demos: federal data · retail prep · NYC taxi.

Watch and go further

• Watch: US markets & segments session
• Related: Federal data · Retail prep · NYC taxi · Community
• Product: Waitlist · Technology · Pricing