Domino's Reviews: You Hate It! or You Love It!

4.2 stars. 42% one-or-two-star reviews. Both true. Same location.

That number, 42%, is not rounding error. It is not a bad week. At the single highest-volume Domino's Pizza location in a 6,199-review sample of Google Maps reviews of southern New Hampshire restaurants, 42% of 1,297 customer reviews were one or two stars. Another 46% were five stars. The middle, the three-and-four-star band you'd expect from a location that averages in the low fours, held about 12%.

Most of that location's customers were either very happy or very unhappy, with almost nothing in between. Two customer groups. And the official Google rating that a district manager checks as she builds her priority list for the week? It sits in the low fours. The split inside those reviews, the thing that would tell her whether this location needs attention, is not in that number.

We're Coherany. We took 6,199 Google Maps reviews from southern New Hampshire, stripped the restaurant names and star ratings, and asked our analysis to group them by language alone, nothing but the words customers wrote. What we can show you is that the gap between a rounded rating and the reality underneath it can be very large.

Three Rounded Numbers Don't Tell You Anything About the Split Underneath

Three rounded numbers don't carry information about what's underneath them. They just look like ratings.

When you check your Google Business Profile portfolio, you see averages. 4.2. 3.9. 4.6. Maybe you sort by the lowest ones, flag a few, move on. The display rounds, 5-star reviews outnumber 1-star reviews by enough to tip the calculation above 4.0. That arithmetic is not wrong. The problem is that averaging removes the one piece of information that actually tells you whether a location needs attention.

A location with forty 4-star reviews and zero 1-star reviews averages 4.0. A location with sixty 5-star reviews and forty 1-star reviews also averages close to 4.0. The published number looks identical. The operational situation is completely different.

The average rounds. The distribution doesn't.

Review-monitoring dashboards do something similar. They put another rounded number on top of the rounded number you already had: a sentiment score, an average rating trend, a response-rate percentage. A Google Business Profile trend line showing your rating rising from 4.2 to 4.3 does not show you the 42/12/46 split underneath it. None of those dashboards show you that the distribution underneath is split in two. The tools report what you already see. They do not show you the shape.

We Read 6,199 Reviews. We Didn't Tell the Analysis Which Restaurant They Came From.

We took 6,199 unique Google Maps reviews and ran them through our grouping system, no restaurant name, no star rating, no cuisine label. The analysis only saw the words customers wrote. Nobody told it which restaurants were good or bad. Nobody handed it a dictionary of "positive" versus "negative" words. We read the reviews cold.

Reviews that used similar language ended up together, regardless of where those customers ate. A customer who writes "took 90 minutes, pizza was cold" reaches for the same words as another customer who wrote "90-minute wait, box was cold" at a different location. Our analysis notices that. The star rating doesn't.

The full run produced 36 groups from 6,199 reviews across 72 restaurants and 37 cuisine types. Running time: roughly 12 minutes. That matters not because speed is the point, but because this wasn't a one-restaurant deep dive, it ran across all 72 locations at once, with no prior knowledge of which ones had problems.

One group dominated.

The Group Average Is 3.21. The Google Display Is 4.2.

We named it Group A. It held 1,261 reviews, and 1,243 of them came from the single highest-volume Domino's location in the dataset.

Inside that group, the average rating was 3.21 stars. Read that again: when you only look at the customers who described a similar experience, the average is 3.21, not 4.2.

The group average is 3.21. The Google display is 4.2. One of those numbers is doing work.

That is not what a 4.2-star location is supposed to look like. Or to be precise: a 4.2-star location can look exactly like that, and the rating will not tell you.

The 3.21 is more informative than the 4.2 because it describes the experience of customers who expressed something similar. The 4.2 describes the arithmetic result of all reviews combined, including the 46% who were delighted. When your two customer groups are that far apart, combining them into one average does not clarify the situation. It buries it.

The split inside Group A's reviews, lots of 1-star reviews, lots of 5-star reviews, almost nothing in the middle, is the entire diagnosis. A district manager sorting her portfolio by published average would see a 4.2 and assign this location low priority. Nothing in the Google Business Profile dashboard flags that 42% of the customer reviews at that high-volume location describe a bad experience.

The group does not tell you why. Our analysis shows the split exists and how large it is. Delivery time, staffing, location accessibility, ordering channel, something is driving the split. That read belongs to the store manager and district manager. Our job ends when our analysis puts the location on your priority list. Figuring out why still belongs to your team.

Sort by the Split, Not by the Average.

The question most district managers have not been able to answer with existing tooling is: which of my locations has a split customer base, lots of fans and lots of haters, even though the rounded rating looks fine?

That is the question this analysis was built to answer. Not "which location has the lowest rating?" That is visible in any dashboard. The question is which location is quietly splitting its customers in two while showing a 4.2 on the screen, at the highest volume.

Sorting your portfolio by rounded average puts locations with genuine operational problems at medium or low priority if enough delighted customers are padding the number. Sort by the split instead, specifically by the share of 1- or 2-star reviews inside each location's biggest customer group, and the locations that need attention surface differently.

For a district manager overseeing multiple units, this changes the Monday morning conversation. Instead of "location seven has a 3.6, let's talk about that," the conversation becomes "location seven shows a 4.2 but 42% of its highest-volume customer group is unhappy. Let's talk about that."

The question a district manager actually needs to ask is: which of my locations is splitting its customers in two while I'm not looking? That question runs against your own review data, not someone else's demo.

We ran this on southern New Hampshire. The same analysis runs on your portfolio. Coherany is the AI analyst you ask in plain English: which locations show this pattern? The engine returns a ranked list, sorted by the customer split underneath the average. Your data. The same approach that found Group A.

Sort by the split, not by the average.

Four Things This Data Cannot Tell You

These caveats are not fine print. They are the reason to trust the signal.

One: This is one location in one geography

The Group A pattern is documented at a single Domino's location in southern New Hampshire. Whether other Domino's locations show the same split, whether other franchise brands in other markets generate comparable distributions, is not established by this sample. The pattern here is striking. It is not a guarantee that your highest-volume location looks like this. That requires running your own data.

Two: The data shows the split exists. It cannot tell you why.

This is the most important caveat. The analysis tells you a location's customer base is splitting in two. It does not tell you whether the split comes from delivery time variability, staffing patterns, peak-hour capacity, order accuracy, or something else entirely. Our job ends when our analysis puts the location on your priority list. Figuring out why still belongs to your team. Anyone who tells you the software can diagnose the cause is selling you something it cannot do.

Three: Lower-volume locations may not generate a reliable signal.

Roughly 100 reviews per location is the practical floor, fewer reviews than that and a clear pattern usually does not surface. Newer units, lower-traffic stores, or recently opened locations may have too few reviews for the grouping to show anything clear.

Four: This is a 2025 snapshot.

The reviews were scraped in 2025. Operational changes since then, new management, staffing adjustments, a renovated kitchen, a new delivery contract, are not in the data. The finding is a snapshot of that location as of 2025. What has changed since is the store manager's domain, not the software's.

Three Questions for Monday Morning

When a district manager sorts her portfolio by average rating, she's making a call about which locations need attention this week. Three questions worth asking about that call:

One

Pull the distribution for your highest-volume location this quarter. Not the average, the actual breakdown: what percentage of reviews are 1 or 2 stars? What percentage are 5? You can get this from your GMB dashboard right now. If you have not looked at it this quarter, look before you read the next question.

Two

What is the highest-volume location in your portfolio? Not the lowest-rated. The highest-volume. That is the location where a split customer base matters most, where the most customer interactions are happening, where a hidden 42% low-star share is concentrated in the most reviews.

Three

The Group A location generated 1,297 reviews, printed a 4.2 on Google, and had 42% of its customers rating it one or two stars. Would your current workflow have flagged it? Or would it have put that location in the "fine, low priority" bucket and moved on?

The Google Business Profile dashboard showed a rounded number in the low fours and nothing else. The split was invisible until we looked at it.

Your highest-volume location is either fine or it isn't. The average will not tell you which one. 4.2 stars said nothing about 42%.