When a Florida family placed their mother in a facility that carried a 4-star CMS rating, a reviewer described walking in to find urine smell, clueless staff, an administrator who used profanity in meetings, and cockroaches running up the walls. Her review opens with a single sentence: "According to CMS, this is a facility with an overall 4-star rating. How?"
She is not alone in asking. An analysis of 10,328 Google reviews from 268 nursing and care facilities across all 50 states found 88 reviews in which families gave a facility 1 or 2 stars on Google while explicitly referencing a positive official or advertised star rating — describing the gap between what the government measured and what they personally experienced. Across those 88 reviews, the average Google rating was 1.11 stars against official ratings that families described as four or five stars. That is not a rounding error. That is a measurement failure.
The CMS Care Compare star rating system is the most widely used tool for evaluating nursing homes in the United States. It is built on three components: health inspection results, staffing data, and quality measures. Each is meaningful. None of them captures what 25.5% of reviewers in our dataset — 2,633 people across 268 facilities in all 50 states — experienced badly enough to give a 1-star rating. A Michigan reviewer put it plainly: "It was only during a State inspection that the residents received actual plates, cups, glasses, and silverware. Inspections should be unannounced for an authentic experience rather than a cover-up leading to a high rating." A Colorado resident — writing as a current resident, not a family member — described the five-star reviews on his own facility as "laughable," documenting unresponsive management, medication delays, subpar food, and what he called "downright hostile CNAs." Understanding what the star system measures — and what it systematically cannot — is the most important research skill a family can develop before placement.
01 — Understand what the three CMS star rating components actually measure — and when they were measured.
The CMS Care Compare rating combines health inspections (weighted most heavily), staffing levels, and clinical quality measures. Health inspections are conducted by state agencies on a roughly annual cycle and are supposed to include unannounced visits, but the timing and depth of those inspections varies considerably by state. Staffing data is reported by the facilities themselves and verified against payroll records quarterly. Quality measures are calculated from clinical data submitted for Medicare billing. All three reflect documented, formal data collected at specific moments in time. None of them reflects what happened on a Tuesday night last month when the agency staffing ratio dropped and call lights went unanswered for an hour. The star rating is a historical document. A recent review is a current report.
02 — Know that inspection scoring rewards the absence of documented violations, not the presence of good care.
Health inspection scores are built on deficiency citations — things that went wrong badly enough to be documented by a state surveyor during a scheduled visit. A facility with no cited deficiencies gets a high inspection score. But the absence of documented violations is not the same as the presence of safe, compassionate, responsive care. Staff rudeness — the single largest complaint category in our dataset, appearing in 502 negative reviews — is almost never a citable deficiency unless it crosses into documented abuse. Cold food, unanswered call lights, inadequate family communication, dismissive management, medication delays that don't result in hospitalizations — none of these reliably produce inspection citations. They produce Google reviews. One-star reviews, at that.
03 — Understand that inspections can be — and are — gamed.
The Michigan reviewer quoted above described a facility where residents received actual plates and silverware only during state inspections. This is not an isolated observation. It reflects a structural vulnerability in any inspection system that operates on a known or semi-predictable cycle: facilities have time to prepare. Staff get briefed. Common areas get cleaned. Paperwork gets organized. The facility that passes an annual inspection with flying colors may look nothing like the facility that operates the other 364 days of the year. Families who tour during business hours experience a version of the same dynamic. Unannounced visits — which families can request and conduct themselves — are the only reliable antidote. Reviews are the next best thing: they are written by people who showed up without a scheduled appointment and saw what was actually there.
04 — Staffing data reflects averages, not the night your parent needs care.
Staffing ratings are calculated from reported staffing levels averaged across the reporting period. A facility can maintain a strong average staffing score while routinely running thin on nights, weekends, and holidays — the exact times when negative reviews cluster most severely. In our dataset, 342 negative reviews specifically referenced nights or weekends, averaging 1.13 stars. Staffing data submitted to CMS doesn't show you the Saturday night when one CNA covered a 30-bed floor, or the Sunday morning when no nurse was present on a hall from 7am to 12:30pm — a scenario documented in a Tennessee one-star review. The average is real. It is also insufficient. Ask any facility what their staffing looks like specifically on nights and weekends, and ask to see the numbers. A facility confident in its overnight coverage will answer directly. One that deflects to overall averages is telling you something.
05 — Quality measures reflect clinical outcomes for documented Medicare patients — not experience.
The clinical quality measures in the CMS star rating include things like the percentage of residents with pressure ulcers, the percentage who received flu vaccines, and the percentage who experienced functional decline. These are meaningful clinical indicators. They are not indicators of whether residents are treated with dignity, whether families' calls are returned, whether food is warm, whether call lights are answered in a reasonable time, or whether staff speak to residents with basic human respect. Ninety-two negative reviews in our dataset specifically mentioned dignity, humiliation, disrespect, or degradation — averaging 1.08 stars. Not one of those experiences would show up in a quality measure unless it resulted in a documented clinical outcome. The CMS system was designed to measure clinical safety. It was not designed to measure whether your mother feels like a person.
06 — The gap between star ratings and reviews is widest at specific facilities — and that pattern is detectable.
Across our dataset, the CMS gap is not evenly distributed. Most facilities with strong CMS ratings also have strong review profiles — the systems align. The dangerous cases are the facilities where official ratings are high and review sentiment is sharply negative, which means real problems are accumulating faster than the inspection and reporting system can detect them. One FL reviewer noted a 4.8 Google rating that was "false," documented by a facility that was "filthy" with bedbound patients rotated improperly and roaches visible on walls. A NC reviewer described a facility carrying a 5-star rating that her family chose specifically because of that rating, only to document a stay defined by neglect, missed meals, and a social worker she described as unresponsive. The word "deceiving" appears in multiple reviews from families who made placement decisions based on official ratings they trusted. NursingHomeIQ is built specifically to surface facilities where the gap between official ratings and review sentiment is widest — because that gap is the most actionable signal available to families doing pre-placement research.
07 — The review volume and trajectory matter as much as the average.
A facility with a 4.2-star Google average means something very different if that average is built on 12 reviews over five years versus 200 reviews in the last 18 months. It means something different again if the most recent 20 reviews average 2.8 stars while the older ones averaged 4.8 — a trajectory pattern consistent with ownership change, management turnover, or gradual decline that inspection cycles haven't yet caught. Sixty-three reviews in our dataset specifically mentioned ownership or management changes; those associated with negative outcomes averaged 1.25 stars, while those associated with positive changes averaged 4.89 — meaning ownership change is either a major quality inflection point in both directions. Treat the star rating as a static snapshot. Treat the review trajectory — volume, recency, trend — as the dynamic signal.
08 — Use reviews to audit the star rating, not replace it.
This is the practical synthesis. The CMS Care Compare rating is a real data source built from real inspection findings, real staffing reports, and real clinical outcomes. It belongs in any serious facility evaluation. What it cannot do — structurally, by design — is tell you what 10,328 families told us across 268 facilities in every state: what it felt like to be there, what happened at 2am, what the food tasted like, whether someone answered the call light, whether staff spoke to residents with kindness or contempt. The star rating tells you what the government found when it looked. The reviews tell you what families found when they lived it. NursingHomeIQ puts both in the same place, weights them against each other, and flags the facilities where they diverge sharply — because that divergence is exactly what families don't have time to find on their own in the middle of a placement crisis.
Supporting Data and Insights
This article draws on an analysis of 10,328 verified Google reviews from 268 nursing and care facilities across all 50 US states.
Key findings from the dataset:
88 reviews gave a facility 1 or 2 stars on Google while explicitly referencing a positive official or advertised star rating; those reviews averaged 1.11 stars
25.5% of all reviews in the dataset — 2,633 reviews — were 1-star ratings; 61.3% were 5-star, reflecting the strong polarization characteristic of nursing home reviews
502 negative reviews documented staff rudeness, unprofessional behavior, or lack of empathy — the largest single complaint category in the dataset, and one that is rarely captured in CMS inspection citations
342 negative reviews referenced night or weekend care specifically, averaging 1.13 stars — a pattern structurally invisible to staffing averages
199 negative reviews documented call light response failures — a daily operational reality that inspection cycles don't reliably detect
198 negative reviews documented food quality failures — not a clinical measure
141 negative reviews documented family communication failures — not a clinical measure
92 negative reviews specifically referenced dignity violations, disrespect, or humiliation — averaging 1.08 stars, the lowest average of any thematic category in the dataset
52 negative reviews mentioned filing formal complaints, contacting the state, or involving an ombudsman — suggesting that even when families escalated formally, the experience remained severe enough to produce 1-star reviews
6 reviews specifically described facility behavior changing during or before state inspections — including one Michigan reviewer who documented that residents received proper dishware only during inspection visits
The structural gap: CMS Care Compare measures what state surveyors find during periodic inspections, what facilities report about their own staffing, and what Medicare billing data shows about clinical outcomes. It does not measure staff attitude, night-shift coverage quality, call light response time, food temperature, family communication responsiveness, or whether residents are treated with basic dignity. Google reviews measure almost exclusively the things CMS does not. Used together, they provide a more complete picture than either source provides alone. Used in isolation, the star rating gives families a misleadingly incomplete basis for one of the most consequential decisions they will ever make.