Inspection Scores vs. Online Reviews: What Each One Actually Tells You

You are standing on the sidewalk with your phone out. The restaurant has 4.7 stars and nine hundred reviews. Taped to the window is a letter grade, or a placard, or a two-digit number, and you have no idea whether it is good.

Most people resolve this by trusting whichever signal is friendlier. That is a mistake — not because one source is honest and the other is lying, but because the two are answering completely different questions. Once you know which question each one answers, the conflict mostly dissolves.

Two sources, two completely different questions

An online review answers: did a customer enjoy being here?

An inspection score answers: on one specific day, did this kitchen follow the food code?

Those questions barely overlap. A reviewer sees the dining room, the plate, and the person who took the order. An inspector sees the walk-in cooler, the sanitizer bucket, the dish machine's final rinse temperature, the hand sink under the prep table, and the shelf where raw chicken is stored relative to the lettuce. There is almost no part of an inspection that a guest is in a position to evaluate, and almost no part of a review that an inspector is asked to score.

When people say "the reviews are great but the grade is bad," they usually mean the two sources contradict each other. They don't. They describe two different rooms.

What an inspection score actually measures

An inspection is a scheduled or unannounced audit against a written food code — in most of the United States, a local adaptation of the FDA Food Code. The inspector walks the operation and records observed violations, and most systems weight those violations by risk.

The heavy-weighted items are the ones epidemiologists associate with actual foodborne illness: temperature control on potentially hazardous foods, cross-contamination between raw and ready-to-eat items, employee handwashing and hygiene, sick-worker policy, approved sourcing, and adequate cooking or cooling. The light-weighted items are structural and housekeeping — a cracked floor tile, an unlabeled squeeze bottle, a missing thermometer in a dry-storage area, worn caulk around a sink.

Three properties of that score matter and are routinely misunderstood:

• It is a snapshot, not a track record. The score describes a few hours on one date. A kitchen that is meticulous fifty-one weeks a year can be caught on a bad Tuesday.
• The scale is local. A 92 in one county is not comparable to a 92 in another, and a letter grade in one city can be earned under thresholds a neighboring city does not use. Comparing scores across jurisdictions is generally meaningless.
• The narrative beats the number. Two restaurants can land on the same score with completely different violation profiles — one with a dozen minor maintenance notes, one with a single serious temperature failure. The number flattens that difference. The report does not.

That last point is the single most useful thing to know. If you only ever read the score, you are reading a summary of a summary.

What online reviews actually measure

A star rating is an average of subjective experiences, and the things that move it most are the things customers can perceive: whether the food tasted good, whether service was fast and pleasant, whether the price felt fair, whether the room was pleasant to sit in, and whether the visit met expectations set by the photos.

Reviews also carry structural quirks worth naming:

• They skew to the extremes. People with a great or terrible experience write; people with an unremarkable one usually don't.
• They lag reality. A rating built over five years describes five years of ownership, staff, and standards. A restaurant that changed hands eight months ago may be carrying a stranger's average.
• Illness reports in reviews are unverified. A person who says they got sick may be right, may be recalling the wrong meal, or may be describing something unrelated. Incubation periods for common pathogens range from hours to days, so the last thing eaten is frequently not the source.
• Volume is not accuracy. Nine hundred reviews tells you a place is popular, not that it is careful.

None of that makes reviews worthless. It makes them a good instrument for one thing and a poor instrument for another.

Where the two disagree — and who is usually right

Four mismatch patterns cover almost every case.

High stars, low score. The most common conflict and the most instructive. It usually means front of house is strong and back of house has a process gap — often cooling, holding temperatures, or storage organization. Customers cannot see any of that. Read the actual report: if the violations are structural and repeat, that is a maintenance problem. If they are temperature and cross-contamination, that is a practice problem, and practice problems are the ones tied to risk.

Low stars, high score. Usually a clean, well-run kitchen attached to slow service, an awkward room, or food that simply isn't to local taste. This is the combination people most often misread as a warning. From a food-safety standpoint it is close to the best case.

Both low. Not much interpretation required.

Both high. Also straightforward — but check the inspection date before you relax. A spotless report from twenty months ago tells you about a kitchen that may no longer exist in the same form.

The general rule: when the two sources conflict, the inspection is more credible about the kitchen and the reviews are more credible about the meal. Neither one overrides the other, because neither one is trying to answer the other's question.

What neither source will tell you

Both instruments share blind spots, and it is worth being honest about them.

• What happens between visits. Inspections are periodic. Reviews are episodic. Neither is continuous monitoring, and no public system pretends otherwise.
• Whether a problem was fixed. Many violations are corrected on the spot while the inspector is standing there, and reports vary in how clearly they mark that. A violation noted and immediately corrected is a different situation from one carried into a follow-up visit.
• Supply chain. Inspectors verify that food came from an approved source; they are not auditing the farm, the processor, or the distributor.
• Staff turnover. The people who earned last year's score may all be gone. Food safety is a habit maintained by individuals, and habits leave when the individuals do.
• Allergen handling. Some jurisdictions check allergen awareness and labeling; many do not score it in a way that shows up usefully in a public record. If you have a serious allergy, no public score is a substitute for a direct conversation with the kitchen.

A practical way to read them together

Two minutes, in this order:

1. Pull the inspection record and look at the date first. Recency changes how much weight the score deserves. Anything older than a year is history, not status.
2. Read the violation list, not just the score. Sort mentally into practice problems (temperature, handwashing, cross-contamination, sick-worker policy) and structure problems (tiles, caulk, lighting, labeling). Practice problems carry the risk.
3. Check whether anything repeats. One temperature violation is a bad day. The same temperature violation across three consecutive inspections is a system that has not been fixed.
4. Now open the reviews — and filter to recent. Read the last few months, not the lifetime average. You are checking for whether the place you researched still exists in current form.
5. Scan for clustered illness reports. One is noise. Several independent accounts in a short window describing similar symptoms is a genuine signal, and worth reporting to your local health department regardless of what you decide to do about dinner.
6. Trust your eyes on arrival. Restrooms, staff hygiene, cold cases, and how the raw-protein station is organized are all visible without asking anyone a question.

Do that and the sidewalk standoff resolves itself. The reviews tell you whether the meal is likely to be good. The inspection record tells you whether the kitchen behind it is being run carefully. You need both answers, and you were never going to get them from one source.

About the author — Rick Jenkins is the founder of AI Revenue Forge. ARF builds vertical-specific AI virtual receptionists for service businesses in HVAC, dental, medspa, real estate, home health, credit repair, and pawn shops. Headquartered in Charlotte, NC. Part of Jenkins Worldwide Enterprises.