Nixeny Dijital - Mersin / Review Economy 2026
The 19,615 businesses in Mersin included in this study have collected 1,290,379 Google reviews. Half of those reviews sit with 338 businesses. One question: in Mersin, who does digital attention go to?
What these numbers measure and what they do not
This is not a count but the sample of a defined, repeatable sweep: 72 business types × 13 districts, every query subject to Google's page limit. The accurate phrasing is not “all businesses in Mersin” but “the Mersin businesses included in this study”.
The tail index is 1.394 — below 2, so the theoretical variance is infinite. The median is 11 and the mean is 80.03; a single business can create the gap between them. Neither this report nor anything quoted from it uses a mean review count.
Google omits the field entirely when the review count is zero; there is not one row in the data set that says 0. Even so, the empty value was not quietly turned into a zero: the concentration measures are given twice, once with those 3,491 businesses excluded and once with them counted as zero.
This study is one moment in time. It describes the shape of the distribution today; it does not measure a rate of growth. The claim “the rich are getting richer” cannot be established without comparing the review counts of the same businesses on two different dates; section 08 exists for that.
Glossary · every term explained once
- Review economy
- The subject of this report: the total of Google reviews in a place and how they are distributed among businesses. A review count is a measure of attention; it is not a customer count.
- Base
- This report uses four separate bases: 19,615 active businesses, 16,124 rated businesses, 1,290,379 reviews, and each sector's or district's own business count. Every rate names which one it is over.
- Median
- The value exactly in the middle of the ranking. In this distribution the median is used rather than the mean: across 16,124 businesses the median review count is 11 and the mean is 80.03 — the difference is made by a handful at the very top.
- Gini coefficient
- Summarises how unequally a resource is distributed, between 0 and 1. At 0 everyone holds the same share; at 1 one party holds everything. The resource measured here is Google reviews.
- Lorenz curve
- The picture of the Gini. The horizontal axis is the cumulative share of businesses, the vertical the cumulative share of reviews. The diagonal is perfect equality; the further the curve sits from it, the more unequal the distribution.
- Top 1% share
- The share of reviews held by the top one percent of businesses ranked by review count. Under perfect equality this value would be 1%.
- Tail index (Hill)
- Measures how heavy the very top of the distribution is. A value below 2 means the theoretical variance is infinite: the mean does not represent a typical business and is never taken into a headline.
- Rank-size (Zipf) fit
- How closely the relationship between a business's rank and its review count sits on a straight line in log-log scale. It is a description of the distribution's shape; it is not evidence of a mechanism.
- Anchor venue
- The 381 venues — shopping centres, terminals, airports, parks, marinas, museums and chain supermarkets — that are not part of a sector contest but measure public or retail footfall. Every sector table is given twice, with them and without them.
- Competitive sector
- The share of businesses in a sector with 100 or more reviews. Within-sector Gini is not used for this: a dead sector where everybody has two reviews also produces a low Gini.
- Digital presence category
- The classification of the website field on a business's Google profile: its own domain, no link at all, or a third-party platform. The definitions come from the Mersin / State of Digital 2026 study.
- Spearman rank correlation
- Summarises whether two rankings move together, between −1 and +1. A value near zero means the two measures do not predict one another.
Half the reviews sit with 338 businesses
Of the 1,290,379 reviews collected by the 19,615 businesses in Mersin included in this study, half sit with only 338 of them — 2.1%. The Gini coefficient of the review count is 0.843: one of the most extreme values you can measure for the distribution of a resource.
This does not mean the businesses at the top are doing something wrong. What is measured is the shape of attention: the habit of leaving a review has settled into a structure that makes what is already visible more visible still. The curve below is a picture of that structure.
The horizontal axis is the cumulative share of businesses, the vertical the cumulative share of reviews. The dashed diagonal is perfect equality: the further the curve is from it, the more unequal the distribution. The shaded area is the Gini coefficient itself.
Each row is the cumulative share of all reviews held by the top N businesses ranked by review volume. The red row is where the share reaches exactly half.
The red end is the share that slice would hold under perfect equality — 1% for the top 1%. The navy end is the share it actually holds. The line between them is the concentration itself.
One business in six has not a single review
At the other end of the concentration there are 3,491 businesses: 17.8% of those included in this study. This is not a gap in the measurement. Google omits the field entirely when the review count is zero; this group means, definitively, “no reviews at all”.
One step above sit another 7,436 businesses: those with between 1 and 9 reviews. Together the two make up more than half the sample, and the reviews they hold amount to 2.0% of the economy.
Each square stands for one percent of the businesses included in this study. A hundred squares in all.
The red end is the bucket's share of the business count. The navy end is the same bucket's share of the review volume. The length of the line is the gap between that bucket's weight and its crowd.
| Bucket | Businesses | Business share | Reviews | Review share |
|---|---|---|---|---|
| No reviews at all | 3,491 | 17.8% | 0 | 0.0% |
| 1-9 reviews | 7,436 | 37.9% | 26,321 | 2.0% |
| 10-49 reviews | 5,272 | 26.9% | 122,467 | 9.5% |
| 50-99 reviews | 1,404 | 7.2% | 97,628 | 7.6% |
| 100-499 reviews | 1,539 | 7.9% | 321,249 | 24.9% |
| 500-999 reviews | 271 | 1.4% | 184,997 | 14.3% |
| 1,000+ reviews | 202 | 1.0% | 537,717 | 41.7% |
Source: Google Places API (New) business profiles · Business shares are over 19,615 businesses, review shares over 1,290,379 reviews.
The top of the ranking is not a sector contest
381 venues — shopping centres, terminals, airports, parks, marinas, museums and chain supermarkets, 1.9% of the sample — hold 16.1% of all the reviews. 22 of the 100 most-reviewed businesses come from this group.
These are not part of a sector contest; they measure public and retail footfall. Left in, they make small businesses’ real share look smaller than it is. That is why every sector table in this report was produced twice: with all businesses, and with these venues excluded.
| # | Business | Reviews | Rating | Sector | District | Cumulative |
|---|---|---|---|---|---|---|
| 1 | Forum Mersin AVManchor | 33,105 | 4.4 | Shopping centre | Yenişehir | 2.6% |
| 2 | Mersin Marinaanchor | 24,856 | 4.4 | Marina / harbour | Yenişehir | 4.5% |
| 3 | Palmcity AVManchor | 10,718 | 4.3 | Shopping centre | Yenişehir | 5.3% |
| 4 | Mersin Şehirlerarası Otobüs Terminalianchor | 9,954 | 3.7 | Terminal / transport | Toroslar | 6.1% |
| 5 | Sayapark AVManchor | 9,808 | 4.4 | Shopping centre | Yenişehir | 6.8% |
| 6 | Tarsus Doğa Parkı ve Hayvanat Bahçesianchor | 8,742 | 4.4 | Park / zoo | Tarsus | 7.5% |
| 7 | Liparis Resort Hotel & Spa | 8,308 | 4.5 | Hotel | Erdemli | 8.2% |
| 8 | Tarsu Alışveriş Merkezianchor | 7,361 | 4.1 | Shopping centre | Tarsus | 8.8% |
| 9 | GEZİPOL TURİZM SEYAHAT ACENTASI | 7,283 | 4.9 | Tour agency | Yenişehir | 9.3% |
| 10 | Mersin Hatay restaurant | 7,201 | 4.3 | Restaurant | Yenişehir | 9.9% |
| 11 | Doramar Resort & Aqua Tatil Köyü | 7,076 | 4.2 | Guest house | Erdemli | 10.4% |
| 12 | Mersin Gezi Otobüsü / JOYBUS TRAVEL | 6,901 | 5.0 | Tour agency | Yenişehir | 10.9% |
| 13 | Pelit Taşpınar Tesislerianchor | 6,900 | 3.8 | Roadside rest complex | Tarsus | 11.5% |
| 14 | 5M Migrosanchor | 6,790 | 4.1 | Supermarket / chain | Yenişehir | 12.0% |
| 15 | Loba Terrace | 6,683 | 4.2 | Restaurant | Yenişehir | 12.5% |
| 16 | Mersin Büyükşehir Belediyesi Kültür Parkıanchor | 5,968 | 4.4 | Park / zoo | Yenişehir | 13.0% |
| 17 | Borsa Tantuni | 5,788 | 4.4 | Restaurant | Yenişehir | 13.4% |
| 18 | Zirve Turistik Dinlenme Tesisi Ve Apart Otelanchor | 5,700 | 2.7 | Roadside rest complex | Erdemli | 13.9% |
| 19 | Gilindire Mağarasıanchor | 5,601 | 4.8 | Natural / tourist site | Aydıncık | 14.3% |
| 20 | LC WAIKIKI | 5,233 | 4.4 | Clothing shop | Mezitli | 14.7% |
Source: Google Places API (New) business profiles, 2026·08·25 · Business names and review counts are public on Google; every row can be verified by its Place ID and those records are shared on request · “Cumulative” is the share of total reviews held by the businesses down to that rank · This is not a quality ranking: a review count measures visibility, not service.
81.8% for hotels, zero for accountants
“The most competitive sector” is not measured here with a within-sector Gini. A dead sector produces a low Gini too: if everybody has two reviews the distribution is equal. Instead a threshold was used that measures how many players are actually on the field — the share of businesses in a sector with 100 or more reviews.
The result is sharp: 81.8% of hotels pass that threshold, while in sectors like accountancy and interior architecture not a single business does. The distinction is about how a customer chooses a business: decisions about lodging and food are made with reviews, decisions about an accountant are not.
Each tile's area varies with that sector's review volume and the depth of its fill with the share of businesses holding 100+ reviews. The percentage inside a tile is the competitiveness rate, not the volume share.
The red line is the median of the 67 sectors in the comparison: 5.3%. Each row carries its own base size beside it.
Each tick is one sector, and the horizontal axis is that sector's share of businesses with 100+ reviews. The tick height carries no data; it only keeps overlapping values distinguishable. Only the 67 sectors with at least 50 businesses are drawn.
- Hotel81.8%n 99
- Restaurant53.3%n 616
- Guest house39.1%n 261
- Patisserie37.2%n 269
- Café30.2%n 414
- Interior architecture0.0%n 183
- Accountant0.0%n 248
- Cold storage0.0%n 55
- Customs brokerage0.0%n 111
- Construction company0.2%n 441
29.3% of the reviews are in one district
Yenişehir alone collects 29.3% of the reviews. But that is not an economic share: because the same query effort was spent in every district, this number measures a share of the reviews in the sample. What is comparable is reviews per business: 111.6 in Yenişehir, 22.9 in Mut.
The share of businesses with no reviews at all splits the same way. In Mut one business in three has not received a single review on Google; in Mezitli that rate is 11.8%. This is what the rural-urban divide looks like on Google.
The red line is where the 13 districts would sit under an equal split: 7.7%. Each district's own business count is written on its row.
Each row is divided into 25 squares; the filled squares show the share of businesses in that district with no reviews at all. Rows above 25% are red: past that threshold, “having no reviews” stops being the exception.
| District | Businesses | Reviews | Per business | Median | Within-district Gini |
|---|---|---|---|---|---|
| Yenişehir | 3,393 | 378,491 | 111.6 | 17.0 | 0.864 |
| Mezitli | 1,909 | 161,448 | 84.6 | 19.0 | 0.805 |
| Tarsus | 2,587 | 160,590 | 62.1 | 10.0 | 0.854 |
| Akdeniz | 2,739 | 134,431 | 49.1 | 9.0 | 0.833 |
| Silifke | 2,057 | 120,348 | 58.5 | 12.0 | 0.817 |
| Erdemli | 1,654 | 116,428 | 70.4 | 12.0 | 0.848 |
| Toroslar | 1,493 | 83,953 | 56.2 | 14.0 | 0.794 |
| Anamur | 1,322 | 52,184 | 39.5 | 7.0 | 0.822 |
| Mut | 961 | 22,053 | 22.9 | 6.0 | 0.819 |
| Bozyazı | 512 | 17,883 | 34.9 | 8.0 | 0.821 |
| Aydıncık | 249 | 17,769 | 71.4 | 10.0 | 0.858 |
| Gülnar | 370 | 11,732 | 31.7 | 4.5 | 0.891 |
| Çamlıyayla | 239 | 8,060 | 33.7 | 7.0 | 0.831 |
Source: Google Places API (New) business profiles · “Median” is the middle review count among the rated businesses in that district · The within-district Gini is above 0.79 everywhere: the concentration exists in every district, only its degree changes.
The 6,281 businesses rated a flat 5.0 have a median of 3 reviews
64.5% of the rated businesses hold a rating of 4.5 or above; the median rating is 4.8. The Google rating has stopped being a signal that separates one business from another.
Nor is the pile-up at the very top a quality ranking. The 6,281 businesses with a flat 5.0 have a median review count of 3; 73.0% of them are under 10 reviews, and together they account for only 8.3% of all reviews. A business can be a 5.0 on three reviews; that means three people were happy.
The column heights are scaled to the largest bucket; the number above a column is a business count. The last bucket is a single rating value: a flat 5.0.
The bucket at the top of the scale, with how many reviews it rests on. The rows on the right are the median review count at each rating step from 4.5 upwards.
The axis runs from 4.0 to 5.0 rather than from zero: a fall of 0.8 is invisible at full scale. The dot is the median rating, the thin line shows the distance from 4.0.
Half the reviews, with a quarter of the businesses
Businesses with their own independent website are 24.3% of the sample, but hold 52.1% of all the Google reviews. The 13,294 businesses with no web link at all on their Google profile are 67.8% of the sample and 40.0% of the reviews.
This is the finding that ties the review economy to the Mersin / State of Digital 2026 study: the attention that is earned collects in the minority that has built its own digital infrastructure. The 515,501 reviews the second group has earned have no address to go to.
A website does not bring reviews. Both may be the result of the same thing.
A large business is expected to have both a site and many reviews. This section shows only that two measurements move together; it does not show which causes the other, and this data cannot show it.
The red end is the category's share of the business count. The navy end is its share of the review volume. For businesses with their own site the navy end sits on the right; for those with no link it sits on the left.
The median is the middle review count among the rated businesses in that category. No mean is used: every category has a tail of its own.
A single snapshot cannot say “the rich get richer”
This section presents no finding; it refuses a claim. Every number above describes the shape of the distribution today. The sentence “reviews are collecting in ever fewer businesses” is a claim about change, and it cannot be built from a single cross-section. That requires the review counts of the same businesses on at least two different dates; the 2027 repeat of this study will produce that data.
What can be said is this: the tail of the distribution is heavier than pure multiplicative noise would produce. A log-normal process would give a Gini of 0.784; the observed value is 0.843. The 0.059 difference between them shows that the shape is consistent with cumulative advantage — not that it was caused by it.
Both axes are logarithmic. The straight line is the least-squares fit to the top 1,000 ranks; it is a description of the observed curve, not evidence of a mechanism. The fit overestimates the first rank, and the axis is kept high to show that.
| Zipf alpha (top 500) | 0.721 |
|---|---|
| Zipf R² (top 500) | 0.992 |
| Gini implied by log-normal | 0.784 |
| Observed Gini | 0.843 |
| Difference (excess tail) | 0.059 |
| Hill tail index (top 1%, k = 162) | 1.394 |
| Top 1% share, against perfect equality | 38.3× |
What this data supports
- That reviews are extremely concentrated, and that this concentration can be measured with the Gini, the Lorenz curve and the top-slice shares.
- That the tail of the distribution is heavier than a purely multiplicative (log-normal) process would produce.
- That the rank-size (Zipf) relationship shows a strong fit; the gap between the top ranks falls off by a regular law.
- That the tail index is below 2; the mean review count does not represent a typical business.
What this data does not support
- That reviews are collecting in ever fewer businesses over time — a single cross-section cannot measure growth.
- That the businesses ahead today will be ahead tomorrow — that needs panel data.
- That the cause of the concentration is Google's algorithm — this study tests no causation.
Limitations
- Google Places is not a complete business register.
- Not every registered business is on Google, and some of those that are may have closed. The universe is the local businesses visible as active on Google.
- The sweep is a sample.
- Text Search returns at most 60 results per query. In dense sector-district combinations coverage is cut at that limit; the businesses left out are most likely in the tail, that is, the ones with few reviews.
- A review count is not a customer count.
- Only some customers leave a review, and that rate varies by sector. A hotel's review count and a pharmacy's review count do not measure the same behaviour.
- The authenticity of the reviews was not audited.
- No fake-review detection was performed in this study. The numbers are the numbers Google shows.
- The sector assignment is a heuristic.
- Even though 418 rows were corrected against Google's primaryType field, another assignment could be defended for businesses with several lines of work. The sector rankings depend on this assignment rule.
- District shares are not economic shares.
- Because the same query effort was spent in every district, these numbers measure shares of the reviews in the sample, not the economic size of the districts.
- The website relationship is not causation.
- That businesses with their own site hold a high share of reviews does not show that the site brings the reviews. A large business is expected to have both a site and reviews.
- A single cross-section cannot measure growth.
- Preferential attachment, a trend, or the claim that “the rich get richer” cannot be tested with this data. The 2027 repeat of the study will produce the panel data that can test it.
Citation rules
We ask everyone who quotes the numbers in this report to keep the study’s own rules. All six exist to prevent one particular misreading.
- 01The mean review count is never taken into a headline; the tail index is below 2.
- 02Do not say “all businesses in Mersin”; say “the Mersin businesses included in this study”.
- 03An empty value is not merged with a zero; both Gini variants are published.
- 04Businesses rated a flat 5.0 are not called “the best”; the median review count is given beside them.
- 05District shares are not economic shares; they are shares of the reviews in the sample.
- 06A sector group's Gini is not the average of its member sectors; it is calculated from the pooled vector.
The method, briefly
- 01UniverseThe data set of the Mersin / State of Digital 2026 study: the 19,615 local businesses visible as active on Google. No extra API call was made for this report; all of it was produced from the existing data.
- 02The empty-value policyThe 3,491 businesses with no review count were not quietly turned into zeroes. The concentration measures were calculated twice: with those rows excluded (the default, n = 16,124, Gini 0.8433) and with them counted as zero (n = 19,615, Gini 0.8712).
- 03District assignmentThe administrative_area_level_2 field in Google's address components. The table holding the district the query was made in was not used: that table counts a business more than once. The 130 records that could not be assigned were not estimated but kept in a bucket of their own.
- 04Sector correctionIn the parent study the sector is derived from the search term that found the business; that distorts a ranking by review volume. The sector of 418 businesses was corrected against Google's primaryType field, and every change was recorded with its reason.
- 05Separating the anchor venuesThe 381 venues such as shopping centres, terminals, airports, parks, marinas, museums and chain supermarkets are reported separately. Every sector table was produced twice: with all businesses and with these venues excluded.
- 06Thresholds and verificationSectors entered the comparison only at n ≥ 50; 67 sectors passed that threshold. Every rate in the report is checked at build time by re-deriving it from the count and base printed beside it.
Sources
- 01Google Places API (New) business profilesThe primary data source; name, address, coordinates, rating, review count, primary type (primaryType), business status and the website field. Fieldwork 2026·08·25. No extra API call was made for this report.developers.google.com
- 02Nixeny Dijital · Mersin / State of Digital 2026The same universe, the same field date. The digital presence classification, the district assignment and the 19,615-business base this report uses come from that study. There the review count was an auxiliary variable; here it is the subject itself.nixeny.com
- 03The sector correction logThe sector of 418 businesses was corrected against Google's own primaryType field, 403 of them by an automatic rule. Every changed row is recorded with its place_id, its old sector, its new sector and the reason, and is shared on request.
- 04The formulas usedGini: 2·Σ(i·xᵢ)/(n·Σx) − (n+1)/n over the ascending vector. Lorenz: 101 cumulative points in 1% steps. HHI: the sum of the squared shares, on a 0-10,000 scale. Zipf: a least-squares fit of log(reviews) = constant − α·log(rank). Log-normal: maximum likelihood on the logarithm of the positive values, implied Gini erf(σ/2). Hill: k / Σ(ln xᵢ − ln x₍k₊₁₎) over the top 1%.
No comparison with any outside data set was made for this report. Placing the concentration of reviews beside an inequality measurement from another field — an income distribution, say — shows different units on the same scale; this report does not do that. The base of every number here is this study’s own universe.
Colophon
Conducted and published by Nixeny Dijital, Mersin. Data collection, the sector correction, the statistics and the report design are Nixeny’s own. The universe and the digital presence classification are shared with the Mersin / State of Digital 2026 study.
The method, the raw data sets, the sector correction log and the verification records are shared on request. For corrections and questions: info@nixeny.com