At Raddod we ran our widest behavioural scan yet. We read every written Google review in our database — 360,098 reviews across 2,219 businesses in four Saudi cities — end to end, with no sampling and no row cap. We asked a single, simple question: when a customer writes a review, do they name a specific person who served them? And when they do, does the rating that comes with it look any different?
The answer turned out to be one of the most stable patterns we have measured. About one review in seven names an individual employee, and those reviews average 4.69 stars against 4.06 for reviews that name nobody — a gap of 0.64 of a star. It holds in all four cities, in all four sectors, and — the part that matters most — it survives when we control for how long the review is. This article lays out the full numbers, and then spends real effort on what the finding does not mean, because that is exactly where most "do this and your rating will climb" advice quietly goes wrong.
Key Findings
- Across 360,098 reviews, about one in seven — roughly 14% — names a specific employee.
- Reviews that name a person average 4.69 stars; reviews that name no one average 4.06. The gap is +0.64 of a star.
- The gap holds in all four cities and all four sectors, and in every one of the 16 city-by-sector cells (from +0.39 to +0.88).
- It is not a by-product of longer reviews. Within short, medium and long reviews alike, the named ones still score higher (+0.62, +0.89, +0.70).
- This is a correlation, not a lever. The likely direction is that a great experience produces the named review — not the other way round.
The Numbers: One in Seven, and a Two-Thirds-Star Gap
Out of 360,098 reviews, our detector flagged roughly one in seven as naming a specific person — a waiter, a receptionist, a barista, a coach. The headline is not really how often it happens; it is what rides alongside it. A review that names someone averages 4.69 stars. A review that names no one averages 4.06. That 0.64 gap is large in a world where whole businesses are separated by a tenth of a star.
The first place to test a pattern is whether it breaks apart when you slice it. It does not. Here it is by sector:
| Sector | Share of reviews naming a person | Rating gap (named minus not named) |
|---|---|---|
| Gyms | 19.4% | +0.73 |
| Hotels | 16.3% | +0.70 |
| Cafés | 13.9% | +0.46 |
| Restaurants | 11.4% | +0.65 |
Gyms name a person most often — nearly one review in five — which fits the relationship a member has with a named coach over months. Hotels come next: a stay is long and multi-touch, from the front desk to housekeeping to breakfast, so there are more moments a single person can own. Cafés name least, and show the smallest gap, because the visit is quick — you order, you leave. But notice there is no sector where the gap disappears. Even the café figure, the weakest, is a clear +0.46.
And by city:
| City | Share naming a person | Rating gap |
|---|---|---|
| Jeddah | 15.3% | +0.58 |
| Mecca | 14.7% | +0.66 |
| Riyadh | 14.2% | +0.64 |
| Dammam | 11.8% | +0.68 |
Four cities, four positive gaps in a tight band. When we cross the two together into all 16 city-by-sector cells, every single one is positive, ranging from +0.39 to +0.88. A pattern that refuses to break under slicing is a pattern you can believe. One more note on the texture of it: this is overwhelmingly an Arabic-language phenomenon. Of the reviews that named someone, 47,571 did so in Arabic and 3,177 in English — naming the person who served you is woven into how Saudis write in their own language.
The Length Control: Why This Is a Finding and Not a Coincidence
Here is the section that turns an interesting correlation into something sturdier. We already know, from our earlier study of why longer Saudi reviews are angrier, that review length runs against rating: the more a customer writes, the lower they tend to score. In this corpus the gradient is clear — short reviews average 4.30 stars, medium ones 3.88, and long ones 3.48.
That raises an obvious objection. Maybe naming a person is just something people do in longer reviews, and longer reviews are a mixed bag. If that were the whole story, the naming gap should shrink or vanish once we hold length constant. It does the opposite. We split every review into short, medium and long and recomputed the gap inside each band:
| Review length | Average when named | Average when not named | Gap |
|---|---|---|---|
| Short (under 120 characters) | 4.86 | 4.24 | +0.62 |
| Medium (120 to 349) | 4.61 | 3.71 | +0.89 |
| Long (350 and over) | 4.03 | 3.33 | +0.70 |
The gap is not only intact inside every band — in medium-length reviews it is wider than the headline figure, at +0.89. So naming someone runs against the length gradient: longer reviews pull ratings down, yet named reviews pull them up at every length. Length cannot explain the naming effect, because the two forces point in opposite directions. That is the difference between a coincidence and a finding.
What This Does — and Does Not — Mean
This is the most important section in the article, so read it before you act on anything above.
It is a correlation, and the likely direction is the reverse of what you might hope. The honest reading is not "name-dropping causes five stars." It is that a genuinely outstanding experience is what makes a customer remember the person — and a happy customer who remembers a name is already in a five-star mood when they sit down to write. The naming is a symptom of a great visit, not a switch that produces one.
So this finding does not say you can raise your rating by getting your staff named. We want to be blunt about that, because it is the tempting misreading. Name badges, a server who says "I'm Fahad, I'll be looking after you," a culture where staff introduce themselves — these are good service in their own right, and they may well make named reviews more likely. But our data cannot promise they move your score, and if anything the causation runs the other way: the score comes first, the remembered name follows. Treat a named review as a thermometer, not a thermostat. It reads the temperature of the experience; it does not set it.
We are not claiming to be more accurate than Google. We read the same public, written reviews that anyone can see on a business's Google profile. We are describing a pattern inside that text — who gets named and how those reviews score — not grading businesses or correcting Google's numbers.
Our detector is a heuristic, and we will tell you its slack. We flagged a review as naming a person using Arabic honorifics followed by a name (الأخ، الأخت، الكابتن، الكوتش، الشيف، المدير، الأستاذ، الدكتور), explicit thanks constructions (شكراً لـ، أشكر), and the English equivalents — titles such as Mr, Chef and Coach, and phrases like "thanks to" and "ask for" followed by a capitalised name. This is not perfect. A loose pass that also caught role words like "the employee" and "special thanks" put the share at 16.3%; a stricter pass that stripped those out put it at 14.1%. That is why we say "about one in seven" rather than quoting a decimal — there is roughly two points of honest slack in the rate. The reassuring part is that the +0.64 gap barely moved between the two passes, which is exactly why we trust the gap even though the share is fuzzy.
A named mention is a pattern match, not proof. It does not guarantee the review praised the person — though, given the ratings, the vast majority do. And because first names like محمد and أحمد are so common, we cannot be certain the same individual is meant each time a name appears. We are measuring a writing behaviour across a market, not profiling any person.
The Named Individual as a Quiet Reputation Engine
Step back from the averages and a human picture appears: in many businesses, one person becomes the story customers tell. Without identifying anyone, the pattern is easy to see in the data. At one Jeddah hotel, a single first name recurs across 62 separate reviews. At another Jeddah hotel, one name appears in 53. At a Dammam seafood restaurant, 52. At a Riyadh hotel, 35. These are individual people who, review after review, became the reason a stranger remembered the place.
We are deliberately not attaching those names to those businesses — these are private people, not public figures, and it would not be fair to build a profile of them from a review database. What we can say at the aggregate level is which first names show up most across the whole corpus, simply as a reflection of common Saudi names: محمد, أحمد, خالد and عبدالله lead the list. That is a fact about naming, not about any one employee. And it is a reminder of the earlier caveat — when a name this common recurs 62 times at one address, some of those mentions may well be different people who happen to share it. We cannot tell them apart, and we are not trying to.
The useful takeaway is not a tactic; it is a lens. The people on your floor are not interchangeable from the customer's side of the counter. When an experience lands, it very often lands as a person — and that person quietly carries a slice of your reputation whether or not anyone planned it that way.
Two More Things the Scan Surfaced
Crowding is a sector story, not a city one. We had wondered whether one city reads as more crowded than the rest. The tidy version of that story did not survive. Mecca does mention crowding slightly more than average — 1.81% of its reviews, against 1.19% to 1.35% elsewhere — but the rating penalty on a crowding mention in Mecca (−0.66 stars) is almost exactly the corpus average (−0.67). There was no Mecca effect worth a headline. The real pattern is by sector: gyms mention crowding in 2.90% of reviews, more than double the overall rate, and a crowding mention costs a gym −1.10 stars, by far the steepest penalty. Hotels barely raise it at all (0.72%). The plain reading is that a crowded gym is simply a bad gym; crowding is a mild, universal irritant everywhere else. When crowding does come up, it travels with complaints about service (in 30% of such reviews), waiting (13%), cramped space (10%) and parking (10%).
Price is the single most rating-destructive theme we measured. Reviews that complain about price or value — غالي, مبالغ, "expensive", "overpriced", "not worth it" — make up 2.77% of the corpus, and they average just 2.77 stars against 4.19 for everything else. That −1.41 gap is more than double the naming gap in size, and it is remarkably uniform: restaurants −1.38, hotels −1.45, cafés −1.54, gyms −1.49. We flag this one honestly as partly tautological — a review that complains about price is, almost by definition, a negative review — so the interest is less "price lowers ratings" and more that, of all the themes we tracked, value is the one that travels with the lowest scores across every sector at once.
How We Did This
We read the entire table of written reviews using keyset pagination — ordering by a stable id and walking the full dataset a thousand rows at a time — so there was no sampling and no silent cap. Coverage was 100% of the 360,098 reviews attached to the 2,219 businesses in Riyadh (806), Jeddah (689), Dammam (439) and Mecca (285), across restaurants (965), hotels (585), cafés (360) and gyms (307).
Name detection was the heuristic described above: Arabic honorific-plus-name, Arabic explicit-thanks, and the English title and thanks constructions, with a stricter second pass that removes role words to give the honest 14% rate. We report both passes rather than hide the one we like. Everything in this article is a direct count from that scan; we did not quote or invent any individual review. And to say it once more, because it matters: this is a correlation measured across a market, not a causal claim about any business, and not a verdict that competes with Google's own rating. The full scoring methodology lives on our methodology page.
This is a more rigorous successor to an earlier, smaller look at staff names (since corrected) that covered two cities and originally overstated the conclusion as causal; the four-city scan confirms the direction and lets us be far more careful about what it does and does not prove. For the national picture of who ranks where, see our pillar ranking, the best restaurants in Saudi Arabia.
See What Your Own Reviews Are Saying
The pattern is market-wide, but the more useful question is about you: are your customers naming anyone, and who? Raddod is a Saudi AI platform that reads your written reviews and surfaces the people, moments and recurring themes inside them — the signal under the star count. Get your free report at raddod.com and see what your customers actually wrote.