How to Analyze Google Reviews Across 50 or More Locations
A step-by-step method for reviews across many locations, with worked numbers from 102 hotels, gyms and restaurants: tagging, comparing like with like, recent ratings and review surges.
To analyze Google reviews across 50 or more locations, pull every review for every location through the Google Business Profile API (or an export from the listings tool you already use), tag each review with every problem it names, compare each location only with locations of the same type, and read recent reviews rather than the all-time rating, after checking for review-campaign surges that inflate it. The rest of this post shows each step with numbers from three of our published analyses: 102 locations and 21,248 public reviews across hotels, gyms and restaurants.
Step 1: collect every review, with its location and date
For locations you own or manage, Google's Business Profile API returns every review: the text, the star rating, the reviewer, the time and your reply. It works only for verified locations and needs OAuth access, and a batch method returns reviews for several locations in one request. The Places API is not a substitute for a full analysis, because it returns at most five reviews per place.
Official sources
Where the full set of reviews can come from
Google's documented options, and what each one returns.
| Source | What you get | Limits Google states |
|---|---|---|
| Business Profile API, reviews.list | Every review for one location: text, star rating, reviewer, time, your reply; plus the location's average rating and total count | Only for verified locations you manage; up to 50 reviews per page; OAuth access required |
| Business Profile API, batchGetReviews | Reviews from several locations in one request | Same access as above |
| Places API, place details | Rating, total review count and a few reviews for any place | A maximum of 5 reviews per place |
Whatever the source, keep a location ID, the location's type, the review date and the star rating on every row. Pull the full history once, then only new reviews each week, so the dataset stays current without starting over.
Step 2: tag every review with every problem it names
A review often names more than one thing: a slow line and a cold order, or a dirty room and a surprise fee. Tag every problem in it, not just the main one, from a fixed list with definitions. With 50 locations the volume is usually too large to read by hand, so a language model does the tagging and a hand-read sample checks it.
Then rank problems by what they do to the rating, not by how often they come up. In the Orlando hotel reviews, front desk and check-in was the most-mentioned topic, in 27% of comments, and its measured effect on a one- or two-star review was about two points and not significant. A dirty room raised the chance of a one- or two-star review by 34 points. Our post on which complaints cause bad reviews covers that method.
Step 3: compare each location with locations of the same type
A single league table of 50 locations mixes where a location is with how it is run. Group them first: urban and suburban, airport and resort, mall and street, or by brand tier. Then rank each location against its own group.
In both the hotel and restaurant data, the spread inside a group was wider than the spread between groups. Among the 45 Marriott-family hotels, the tier averages for one- and two-star reviews ran from about 16% to 26%, while inside the full-service tier alone the best hotel was at 15.3% and the worst at 47.4%. The brief puts 84% of the gap between hotels within tiers and 16% between them.
Dimension Labs analysis
Inside one hotel tier, best and worst are up to 33.7 points apart; the tier averages differ by 10.1
Share of each hotel's rated reviews that are one or two stars, 45 Marriott-family hotels in Orlando. Dot: the tier average. Line: from the best hotel in the tier to the worst.
Dimension Labs analysis
The same holds for restaurants: stores in one ring are further apart than the rings are
Share of each store's reviews that are one or two stars, 24 Chick-fil-A restaurants in Atlanta grouped by distance from downtown. Dot: the ring average. Line: best store to worst.
Small locations need care. A hotel with 64 reviews can look best in its group by chance. The hotel brief adjusts each rate toward the average in proportion to how few reviews it has (empirical-Bayes shrinkage): one 64-review hotel goes from 9.4% raw to 12.7% adjusted, while a 643-review hotel barely moves, from 20.1% to 20.2%. Rank on the adjusted rate, or at least set a minimum count.
From the hotel brief
A 64-review hotel moves 3.3 points when adjusted; a 643-review hotel barely moves
One- and two-star share before and after the brief's empirical-Bayes adjustment toward the average.
RawAdjusted
Step 4: use recent reviews, not the all-time rating
The rating on a Google profile averages a location's reviews over its whole life, so it moves slowly and can hide a recent decline. In the Chicago gym data, the 33 clubs had 36,279 reviews in total, and the last 18 months were about a quarter of them. At the weakest urban club Google showed 4.1 stars, while reviews from those 18 months averaged 2.8.
From the report: Chicago gym report
Use a fixed recent window, such as the last 12 or 18 months, for every location, and report the review count next to each rate.
Step 5: check for review-campaign surges
A push to collect reviews, such as a QR code at the desk or a follow-up email, can lift five-star volume without any change in the experience. In the gym data, about 4,903 five-star reviews arrived above the normal rate between February and October 2025, while one- and two-star reviews barely moved. Total monthly volume tracked five-star volume almost exactly (a correlation of 0.999) and one- and two-star volume much less (0.48).
From the report: Chicago gym report
Surges are rarely even across locations, so they distort comparisons. Outer-suburban clubs received most of the surge reviews; outside the surge months their share of one- and two-star reviews was 7.8%, against 3.1% over the whole period. To check your own data, chart monthly reviews by star band for each location and look for months where five-star volume jumps and the rest does not. Compare locations on months outside the surge, or on the share of one- and two-star reviews, which a campaign moves less.
From our published analyses
Three numbers that change how you read a location's reviews
Public Google reviews of hotels and gyms, January 2025 to July 2026.
Step 6: join reviews to what you already track
Reviews show what guests or members say. To see what it costs, join them by location and month to sales, traffic, cancellations, staffing and ticket data. A location whose reviews keep naming slow service and whose sales are falling in the same months is the first one to visit. For a worked restaurant example, see how to compare restaurant locations using guest reviews; for gyms, why gym members say they will cancel.
FAQ
Can I get all Google reviews for locations I do not manage?
Not through Google's APIs. The Business Profile API covers verified locations you manage, and the Places API returns a maximum of five reviews per place.
How many reviews does a location need before I compare it?
There is no fixed cut-off, but rates from a few dozen reviews swing a lot. Adjust small locations toward their group average, as the hotel brief did, or set a minimum such as 60 reviews in the window. Every store in our restaurant comparison had at least 64.
Should I compare star averages or the share of bad reviews?
The share of one- and two-star reviews is easier to explain and moves less when a review campaign adds five-star reviews. Use the average as a second view.
Dimension Labs reads every review for every location, tags each one with your business's own Dimensions, ranks locations against comparable locations, flags review surges, and joins it all to sales, traffic and staffing data. The agent sends the reports and alerts on a schedule. See multi-location intelligence and review intelligence, or start with one file of reviews. Plans start at $299 a month, with no demo or sales call needed.