Methodology

The atom of every calculation is the Census block. ZIP Code Tabulation Areas (ZCTAs) are mapped to districts, counties, cities, and media markets with crosswalks from the Missouri Census Data Center's Geocorr 2022 application, built on the 2020 Census.

When a ZCTA straddles a boundary, an allocation factor splits it. The factor is the share of the ZCTA's population that lives in blocks inside the boundary. A row that reads "62.4% of ZCTA" means 62.4% of that ZIP's residents live inside the selected geography. Allocated counts (population, VAP, CVAP, households) apply the same block-level split.

Congressional districts are the maps in effect for the 2026 elections. Ten states (AL, CA, FL, LA, MO, NC, OH, TN, TX, UT) adopted new maps after the 119th Congress was seated. Their districts use each state's enacted or court-ordered 2026 plan, sourced from official state redistricting files. All other states use 119th Congress boundaries. New-plan crosswalks were rebuilt from Census blocks and validated with block-level point-in-polygon checks at 99.9% agreement or better in every state.

State senate and state house districts use 2024 TIGER/Line boundaries. Counties use 2024 TIGER/Line. Media markets are the 210 Nielsen Designated Market Areas (DMAs), assembled from 2020 county definitions.

Total population, voting-age population (VAP, adults 18+), and household counts are exact block-level figures from the 2020 Census PL 94-171 redistricting file. They are counts, not estimates. Households are occupied housing units.

CVAP is an estimate from the Census Bureau's ACS 2020–2024 CVAP special tabulation. County figures are official published rows. Other geographies are allocated from block groups to blocks by VAP share.

CVAP splits by race and ethnicity (Hispanic or Latino, and non-Hispanic White, Black, and Asian alone) come from the same tabulation. Each group is allocated by its own block-level VAP. Because each estimate uses its own weights, group values in a single ZCTA row may not sum to that row's total CVAP. District and media-market totals are internally consistent. Other and multiracial categories are not shown, so the four groups do not sum to 100%.

Language surfaces come from the ACS 2020–2024 five-year release: table B16004 (adults by language and English ability), C16002 (household language), C16001 (12-group detail, population 5+), and B16008 (citizenship by language). ZCTA, state-legislative, and most congressional figures are the Census Bureau's own published rows. Congressional districts in the ten 2026-redistricted states, and media markets, are allocated from census tracts by VAP share.

"English-speaking" counts adults who speak only English or speak it "very well." "Spanish-dominant" counts Spanish speakers whose English is less than "very well." Surfaces overlap by design, so English plus Spanish can exceed total VAP. An English flight and a Spanish flight buy different inventory; grade each against its own audience. The List Grader's natural efficiency is always targets ÷ adults (VAP 18+) on the chosen surface. Targets are people, so the denominator is people, never households.

Non-citizen adult counts reflect citizenship only. They include lawful permanent residents, visa holders, and refugees. They are not an estimate of any narrower immigration status. The Census does not measure legal status beyond citizenship.

The CTV delivery universe is households with a broadband subscription of any type, minus cellular-data-only households (ACS table B28002). These are the homes a connected-TV impression can plausibly reach. This universe sizes delivery and reach. It is never a rating denominator.

Impressions = budget ÷ CPM × 1,000, spread across the ZIP list in proportion to each ZIP's CTV-targetable households. ZIPs without broadband data fall back to total households.

ZIP-targeted impressions land in the intended ZIP at an editable match rate (default 90%). IP-to-ZIP geolocation is not exact, and mismatches count as out-of-district. Run-of-network (DMA-wide) buys take no match-rate haircut.

Household GRPs are gross ratings: 100 × in-district impressions ÷ the geography's full household universe. Rating universes are ACS 2020–2024 households and adults 18+. Nielsen universe estimates count TV households instead, which typically run 2–4% lower, so Nielsen-computed GRPs read slightly higher on the same delivery. A toggle switches the planner from household GRPs to VAP TRPs. VAP TRPs apply an editable co-viewing factor (default 1.4 persons per impression) against adults 18+. With a pasted list, the grader also shows list target TRPs: matched impressions × co-viewing expected to land on the list's own targets, ÷ the list's targets, assuming viewing spreads evenly across each ZIP's adults. Targets in the out-of-district part of a straddling ZIP still count. Citizenship is never used in any rating calculation.

The weekly reach ceiling assumes ad-supported streaming reaches an editable share (default 48%) of all households in the bought geography. The default comes from matching CTV delivery back to offline household files, so broadband, device, and app-usage attrition is already inside it. Implied frequency = in-district impressions ÷ that ceiling. The frequency bands are planning guidance, not hard limits: 2× per household per week is a common floor, 5–7× the usual target, and 8× a practical ceiling for auction-based buying. Effective-frequency research spans a wide range, from a single well-timed exposure to weekly caps of 3–7×, with returns usually diminishing past 7×. Above 8× the planner flags the plan: auction supply gets harder to fill at that frequency, and direct or programmatic-guaranteed deals are usually the better route. Lighter schedules can still work. All planner outputs are planning estimates, not measurement.

District-to-DMA and city-to-DMA overlaps are computed at the Census block level (cities join blocks through the Census Bureau's 2020 Block Assignment Files), so household counts in each intersection are exact to the block. The premium column in the planner compares the cost per in-district point of a run-of-network DMA buy against the ZIP-targeted buy. An optional broadcast CPM prices the same budget as a market-wide broadcast buy: delivery spreads across the market's total households (no broadband haircut) and reads as household GRPs only, since broadcast is untargeted and the planner's demo TRPs assume an audience-targeted buy.

Census counts (population, VAP, households) are exact. Everything sourced from the ACS (CVAP, language, broadband) is a survey estimate and carries sampling error. Treat small differences as noise, and treat small-area language and CVAP splits as directional. If a number looks wrong, send us a note.