Google Places vs Google Maps Results: Which Endpoint Do You Need?
Michael Carter · · 11 min read

Key takeaways
- With a city in the query, 103 of 130 /maps listings (79%) were also among the 20 /places listings for the same query, but the two endpoints ranked the same business first in only 2 of 13 queries.
- Without a city in the query, /places and /maps shared none of their 40 listings: each request returned businesses in a different US city, for example New York on /places and Miami on /maps for "coffee shop".
- /places returns 20 listings per credit with price ranges and review snippets; /maps returns 10 per credit with coordinates, a ChIJ place ID, the CID and exact review counts on every listing.
- The /g/ ID inside every /maps link equals the /places placeId, which joined all 103 shared businesses; joining on the name alone paired 4 listings with the wrong branch of a chain.
- /places gave a street address for only 182 of 360 listings and for none on 7 of 18 queries, while /maps had one on all 170 listings that were not hotels.
Google Places vs Google Maps results come down to fields: use /maps when you need coordinates, a place ID, the CID and street addresses, and /places for 20 listings per credit with price ranges and review snippets. With a city in the query, 79% of /maps businesses also appeared on /places, ranked differently.
This comparison is for developers and data teams choosing between the /places and /maps endpoints of BestSERP (bestserp.com), a Google SERP API. It builds on the field counts in our local search data guide with new measurements from 20 queries run on both endpoints on November 18, 2026: overlap, ranking, the effect of the city, joining the two, and the cost of 1,000 businesses.
Google Places vs Google Maps at a glance
Both endpoints take the same core parameters (q, gl, hl, pages, noCache) and answer with a places array, but they fill it from different Google surfaces. Counts are from 360 /places and 180 /maps listings.
/places |
/maps |
|
|---|---|---|
| Source | Places tab of Google Search | Google Maps search |
| Listings per credit | 20 (a hotel query returned 0) | 10 |
| Coordinates and CID | Never (0 of 360) | Always (180 of 180) |
| Place ID form | /g/... (358) or /m/... (2) |
ChIJ... (180 of 180) |
| Street address | 182 of 360 (51%) | 170 of 180 (all but hotels) |
| Price range | 168 of 180 food and drink listings | Never |
| Review snippet | 354 of 360 | Never |
| Review count | Rounded from 1,000 up | Exact |
| Median response time | 3.6 s | 4.6 s |
| Phone number, website | No | No |
The parameter lists are in the /places reference and the /maps reference.
Where each endpoint's results come from
/places reads the Places tab of Google Search: a longer list of businesses than the three in a local pack on the main results page. /maps reads the result list of a Google Maps search, where every business is pinned to a point. Both are Google's own ranking for the same query on two surfaces, so they share most businesses but not their order.
Neither is Google's Places API, the official Google Maps Platform service billed per request; what the Google Places API costs compares the two at real volumes.
Do they return the same businesses?
Mostly, when the query names a city. Across 13 such queries, 103 of the 130 /maps listings (79%) were also among the 20 /places listings for the same query. Only 76 of them (58%) were in the /places top 10, so the /maps list is not simply the first half of the /places list.
Shared businesses per query, out of 10: pizza in Glasgow 10; bakery in Boston, bookstore in Minneapolis, gym in Seattle, hair salon in Atlanta, thai restaurant in San Diego and veterinarian in Portland, Oregon 9; coffee shop in Leeds and plumber in Phoenix 8; coffee shop in Chicago 7; dentist in Houston 6; auto repair in Dallas and pizza in Denver 5.

Calling both adds few new businesses: on the 13 queries, /maps (13 more credits) brought 27 that the 260 /places listings lacked. What it adds is location data for the 103 shared ones.
Does the ranking agree?
Not closely. The two endpoints put the same business first in only 2 of 13 queries ("pizza in Glasgow" and "thai restaurant in San Diego"). Their top 3s shared no business in 5 queries and one in 6.
For the businesses on both lists, the position moved by 3.8 places on average. The rank correlation (Spearman) per query had a median of 0.53, from -0.14 for "plumber in Phoenix" to 0.90 for "auto repair in Dallas". If you track local rankings, pick the surface your customers use and keep its positions in its own history, as you would in keyword rank tracking; never mix /places and /maps positions in one series.
Put the city in the query, and check ipCountry
Without a city in q, the two endpoints shared none of their 40 listings on four US queries, because each request returned businesses in a different city: New York on /places and Miami on /maps for "coffee shop", Denver and San Diego for "dentist", Orlando and Los Angeles for "pizza", Miami and New York for "plumber".

BestSERP targets by country (gl) and language (hl) only, with no city parameter, so when q names no place the businesses follow wherever Google thinks the searcher is. Google's page on how it estimates your location names the IP address and, in some cases, areas searched for in the past (checked November 18, 2026).
The country code does not pin a place either. Three /maps requests for "coffee shop" with gl: "gb" that ran from US IPs returned coffee shops in Houston; Bristol, England; and Portland, Oregon. With the city named, "dentist in Bristol" from US IPs still returned dentists in Bristol, England, on both endpoints.
Each search runs from an IP in the gl country when one is free, otherwise from another country, and ipCountry says which: 33 of 33 US requests ran from US IPs, but only 5 of 16 UK requests from UK IPs. When the country matters, check it and rerun mismatches, as in checking Google rankings in another country.
Same business, different fields
For the 103 businesses on both endpoints, the name and star rating were identical every time. Other fields differed:
- Review counts. Identical for all 66 businesses with fewer than 1,000 reviews. For all 37 with 1,000 or more, /places showed a rounded figure: 2,273 became 2,300 and 33,715 became 34,000.
- Addresses. /places gave a street address for 182 of 360 listings, and none on 7 of 18 queries: only the city ("Atlanta, GA") or another line of the listing ("10+ years in business"). /maps had one on all 170 non-hotel listings.
- Categories. Same text for 77 of 103. /places often uses a shorter label: "Pizza" and "Thai" where /maps says "Pizza restaurant" and "Thai restaurant".
- Price range. Only on /places, and only for food and drink: 168 of 180 coffee, bakery, pizza and Thai listings, 0 of 180 for the services and shops.
- Opening status. Both give the status at request time ("Closed · Opens 7 AM Wed"), not a timetable; it agreed for 95 of 103. On 118 /places and 50 /maps listings,
hoursalso carried the rest of the listing line: keep the status, drop the rest. - Hotels. "hotel in Miami" returned 0 listings on /places (still 1 credit) and 10 hotels on /maps, with "4-star hotel" as category and no address or hours.
How to join /places and /maps results
Join on the /g/ ID. /places returns it as placeId; /maps hides it in every link, next to the CID and the coordinates. Here is the /maps link for Sawada Coffee, split at its parts:
https://www.google.com/maps/place/Sawada+Coffee/data=!4m7!3m6
!1s0x880e2cdada6e91e9:0x85e817197841a73e 0x85e817197841a73e = 9648987599800477502 = cid
!8m2!3d41.8836251!4d-87.6487119 latitude, longitude
!16s%2Fg%2F11ckwy0qw6 /g/11ckwy0qw6 = placeId on /places
!19sChIJ6ZFu2tosDogRPqdBeBkX6IU placeId on /maps
All 180 /maps links held a /g/ ID, and in all 180 the hex number after the colon equaled the cid. The alternatives to the ID join did worse:
| Join | Right | Wrong | Missed |
|---|---|---|---|
/g/ ID from the /maps link |
103 | 0 | 0 |
| Normalized name | 100 | 4 | 3 |
| Name plus street number | 101 | 0 | 2 |
Names fail on chains ("Flour Bakery + Cafe" had three Boston branches), and street numbers help only when /places has a street. This function does the ID join; we ran it on the two saved Chicago responses on November 18, 2026 with Python 3.12.10 (fetch responses as in the quickstart):
import re
from urllib.parse import unquote
KG_IN_LINK = re.compile(r"!16s(%2F[gm]%2F[^!?&]+)") # the /g/ or /m/ ID inside a /maps link
def kg_id(maps_listing):
m = KG_IN_LINK.search(maps_listing.get("link", ""))
return unquote(m.group(1)) if m else None
def join(places_resp, maps_resp):
"""One row per /maps listing, with the /places price range and snippet when the business is in both."""
by_id = {p["placeId"]: p for p in places_resp.get("places", []) if p.get("placeId")}
rows = []
for m in maps_resp.get("places", []):
p = by_id.get(kg_id(m), {})
rows.append({
"title": m["title"], "address": m.get("address"),
"lat": m.get("latitude"), "lng": m.get("longitude"),
"placeId": m.get("placeId"), "cid": m.get("cid"), "kgId": kg_id(m),
"rating": m.get("rating"), "reviews": m.get("ratingCount"), # exact count on /maps
"hours": m.get("hours"),
"mapsRank": m["position"], "placesRank": p.get("position"),
"priceLevel": p.get("priceLevel"), "snippet": p.get("snippet"),
})
return rows
Printing the match count and the first five rows gave:
10 /maps listings, 7 also on /places
maps #1 places #11 Caffe Umbria /g/11bwh9wdq2 $1–10 41.8888,-87.6312
maps #2 places #1 Sawada Coffee /g/11ckwy0qw6 $10–20 41.8836,-87.6487
maps #3 places #4 Intelligentsia /g/1hc55s20n $1–10 41.8843,-87.6258
maps #4 places #9 The Wormhole Coffee /g/1tcysf87 $1–10 41.9084,-87.6746
maps #5 places #8 Blue Bottle Coffee /g/11r2t1qsjy $10–20 41.8878,-87.6365
Store the ChIJ place ID and the CID from /maps as your keys. Google's Place IDs documentation says place IDs "may change over time" and recommends refreshing those more than 12 months old (page last updated November 9, 2026).
Listings per request, per page and per credit
/places returns twice as many listings per credit: 20 in 18 of 19 requests (the hotel query gave 0), against 10 in 18 of 18 on /maps. With pages: 2, /places returned 40 listings for 2 credits in 8.1 seconds (39 different businesses, the first 20 the same as on one page); /maps gives 10 more per page.
At BestSERP's list price of $0.50 per 1,000 credits (pricing, after 3,000 free searches), the cost per 1,000 businesses from our runs:
| What you collect | Credits per query | Per 1,000 businesses |
|---|---|---|
| /places listings | 1 (20 listings) | $0.025 |
| /maps listings | 1 per 10 listings | $0.05 |
| Both, every distinct business | 2 (about 22 businesses) | $0.045 |
| Both, businesses with both field sets | 2 (about 8 joined) | $0.126 |
The last row buys complete records: /maps location data plus the /places price range and snippet. Duplicates across queries raise every figure, as the local business lead list tutorial shows.
Which endpoint do you need?
Choose by the field you cannot do without. This table sums up the Google Places vs Google Maps choice by use case:
| Use case | Endpoint | Why (measured November 18, 2026) |
|---|---|---|
| Map pins, distances, geofencing | /maps | Coordinates on 180 of 180 listings |
| Dedupe or refresh businesses later | /maps | ChIJ place ID and CID on every listing |
| Lists that need street addresses | /maps | Street on 170 of 170 non-hotel listings, /places 51% |
| Who appears on the Places tab of Google Search | /places | It reads that tab, 20 per credit |
| Restaurant and cafe price ranges, review snippets | /places | Only /places has them |
| The most names per credit | /places | $0.025 per 1,000 listings |
| Exact review counts | /maps | /places rounds from 1,000 up |
| Hotels | /maps | /places returned no listings |
| All of the above | Both, joined on the /g/ ID |
2 credits per query |
| Phone numbers or websites | Neither | Google's Places API Place Details |
The product pages, the Google Places API and the Google Maps API, list both endpoints' parameters and limits.
How we compared
- When and how: November 18, 2026, 04:28 to 04:33 UTC (late evening in the US, so most listings were closed);
hl: "en",noCache: true, both requests of a pair started together (except the reruns below); 49 requests, 50 credits. - Queries: the 13 with a city listed above, "hotel in Miami", "coffee shop", "dentist", "pizza" and "plumber" (
gl: "us"), and "dentist in Bristol" and "coffee shop" (gl: "gb"). - Exclusions: only responses whose
ipCountrymatchedglcount. We reran UK requests that ran from US IPs (the Leeds and Glasgow pairs then ran 1.6 minutes apart); "dentist in Bristol" and the UK "coffee shop" never got UK IPs on both endpoints and appear only in the location section, and "hotel in Miami" had no /places listings. That leaves 17 pairs: 13 with a city (11 US, 2 UK) and 4 without. - Matching and location: the
/g/ID in each /maps link against /placesplaceId; /maps centers looked up in the US Census Bureau's TIGERweb; /places cities read from addresses and names. Phone-like text was removed from the saved responses.
17 pairs on one night are a small sample: read the counts as a picture of the two endpoints, not as fixed rates.
Next steps
The Google Places vs Google Maps decision is about fields: /maps for location and identity, /places for volume and price ranges, both when you need complete records.
- The full picture of local data, including Shopping: the local search data guide.
- A complete /maps project with dedupe and location checks: build a local business lead list.
- Ready to compare on your own city: create a free account for 3,000 free searches.