`Geo.metro`: Google metro codes not found in appendix

[JacksonKjar]

The Geo.metro attribute is supposed to use "Google metro codes":

Google metro code; similar to but not exactly Nielsen DMAs. See Appendix A for a link to the codes.

Appendix A directs us to https://developers.google.com/google-ads/api/data/geotargets, which provides a downloadable CSV of Google's geo targets.

However, the page makes no obvious mention of "metro codes", and inspecting the CSV suggests that none of the "Target Types" correspond to a metro code similar to Nielsen DMAs.

select "Target Type", count(*) as count
         from "geotargets-2026-07-16.csv"
         where "Country Code"  = 'US'
         group by "Target Type"
         order by count
         ;
┌────────────────────────┬───────┐
│      Target Type       │ count │
│        varchar         │ int64 │
├────────────────────────┼───────┤
│ Country                │     1 │
│ District               │     1 │
│ Region                 │     2 │
│ City Region            │     4 │
│ Borough                │    41 │
│ State                  │    51 │
│ National Park          │    57 │
│ Colloquial Area        │    97 │
│ University             │   169 │
│ Airport                │   174 │
│ Congressional District │   684 │
│ Municipality           │  1947 │
│ County                 │  3098 │
│ Neighborhood           │  4911 │
│ City                   │ 19676 │
│ Postal Code            │ 33371 │
└────────────────────────┴───────┘

It's unclear how this field should be used.

[JacksonKjar]

Additionally, Google's Authorized Buyers OpenRTB guide defines Geo.metro as:

Google metro code found in geo-table.csv.

Analyzing that CSV reveals:

SELECT Target_Type, count(*) as count
FROM read_csv('geo-table.csv', header=false, ignore_errors=true,
    columns={'Criteria_ID':'BIGINT','Name':'VARCHAR','Canonical_Name':'VARCHAR',
             'Parent_ID':'VARCHAR','Region_Code':'VARCHAR','Country_Code':'VARCHAR',
             'Target_Type':'VARCHAR'})
WHERE Country_Code = 'US'
GROUP BY Target_Type
ORDER BY count DESC;

┌────────────────────────┬───────┐
│      Target_Type       │ count │
│        varchar         │ int64 │
├────────────────────────┼───────┤
│ Postal Code            │ 24609 │
│ City                   │ 13632 │
│ Neighborhood           │  3712 │
│ County                 │  3087 │
│ Municipality           │   612 │
│ Congressional District │   255 │
│ DMA Region             │   210 │
│ State                  │    51 │
│ Borough                │    15 │
│ Colloquial Area        │     3 │
│ Region                 │     2 │
│ City Region            │     1 │
│ Country                │     1 │
└────────────────────────┴───────┘

There is no "Metro" target type in this file. There are 210 entries of type "DMA Region" (US-only), with Criteria IDs in range 200500–200881. These seem to be Nielsen DMA codes + 200000:

Criteria_ID	Nielsen DMA	Name
200501	501	New York, NY
200803	803	Los Angeles, CA
200602	602	Chicago, IL
200807	807	San Francisco-Oakland-San Jose, CA

Meanwhile, the link in Appendix A (developers.google.com/google-ads/api/data/geotargets) is a completely different CSV with different target types and no DMA/metro concept at all.

Summary of the ambiguity:

1. The IAB spec says "Google metro code; similar to but not exactly Nielsen DMAs"
2. Appendix A links to a Google Ads CSV that has no metro codes
3. Google's own RTB implementation references a different CSV (geo-table.csv), presumably referring to the "DMA Region" target type, which looks to just be Nielsen DMAs + 200000
4. Google's geotargeting page now marks geo_criteria_id as deprecated and recommends using lat/lon instead

Related: #153 (the original Appendix A link was dead; the replacement still doesn't resolve this)