Ziggu Research · Belgium · Collected 27 August 2026
Google Review Study: Belgian Property Developers 2026
What 2 971 public Google review texts for 329 Belgian residential property developers said about the sector: where the scores cluster, what buyers actually write about, and how much of the review picture is organised solicitation.
Public Google reviews of Belgian residential property developers were polarised: 89,0% of the 2 971 text reviews collected in August 2026 awarded either one star or five, and only 1,4% awarded three. The tone turned along the buyer journey: 66,1% of reviews touching aftercare were negative, and the most named specific complaint was a promise not kept. The writing itself accelerated too: yearly review volume roughly tripled between 2021 and 2025.
Ziggu collected every public text review on the verified Google Maps profiles of 329 Belgian residential developers. A review profile is not a satisfaction survey: it is a public record of who was moved enough to write. This study measures that record. It counts where the anger and the praise concentrate, what complaints and compliments name in their own words, at which point of the buyer journey the tone flips, whether developers with large review bases really score better, how much of the five-star volume shows the timing signature of organised review campaigns, and how fast the reviewing habit is growing.
Every figure on this page names its own base, and the method section defines each measurement precisely enough to be rebuilt. No developer is named and no review is quoted: a verbatim quote is searchable and would identify the company, and this page reports the sector, not individual firms. Public reviews measure reputation; for what buyers said when they were asked directly, see the Homebuyer NPS Study 2026, built on 1 936 recommendation scores collected through client portals.
The figures at a glance
| Finding | Value | Base |
|---|---|---|
| Text reviews awarding one or five stars | 89,0% | n = 2 971 |
| Average rating across all text reviews | 3,93 | n = 2 971 |
| One-star share | 22,6% | n = 2 971 |
| Negative reviews touching communication and reachability | 44,6% | n = 734 |
| Positive reviews praising staff and attitude | 40,8% | n = 2 195 |
| Aftercare-related reviews that were negative | 66,1% | n = 165 |
| Negative communication reviews naming a promise not kept | 20,5% | n = 327 |
| Negative aftercare reviews placing the break at signing, payment or handover | 47,7% | n = 109 |
| Reviews inside solicitation-like five-star bursts | 18,3% | n = 2 971 |
| Yearly review volume, 2021 to 2025 | 230 → 588 | n = 2 971 |
| Text reviews per reviewed developer per year, 2019 to 2025 | 2,3 → 5,4 | n = 2 971 |
| Developers without a single text review | 36,2% | n = 329 |
Method
The source is the public Google Maps business profile of each developer. A candidate list of 393 Belgian residential property developers was compiled from public sources; 338 were verified as a correct profile match on name and postcode, and after merging nine pairs that turned out to share one Maps profile, 329 unique developers remained. For each verified profile, up to the 200 most recent text reviews were collected on 27 August 2026. Ratings without any written text were not collected, so averages on this page can differ from the profile score visible on Google, which includes them. After removing 124 duplicates created by the profile merges, the dataset holds 2 971 unique text reviews from the 210 developers that had at least one.
Themes were assigned with a multilingual keyword classification (Dutch, French and English) across seven themes: communication and reachability, staff and attitude, the sales experience, finishing quality, price and invoicing, delivery and timing, and aftercare and warranty. The classification is multilabel: a review complaining about poor communication and a late delivery counted for both. Within three subsets (negative communication reviews, negative aftercare reviews and positive staff reviews), specific sub-clusters were then assigned with the same keyword approach. Keyword classification misses irony and implicit complaints, so theme and cluster percentages are robust as an order of magnitude, not to the percentage point. “Negative” means one or two stars (734 reviews), “positive” means four or five (2 195 reviews); the 42 three-star reviews belong to neither group.
A review counts as part of a burst when the same developer received five or more five-star text reviews written within 14 days of each other: the typical trace of a mailing or QR campaign, though a well-timed satisfaction request after handover produces the same trace. Percentages are rounded to one decimal, so a column may not total exactly 100. The analysis was scripted on the raw dataset; every count on this page comes from that script, not from a summary.
Finding 1
How polarised were the reviews?
Of 2 971 text reviews, 66,4% awarded five stars and 22,6% awarded one. The middle barely existed: 1,4% gave three stars. The average of 3,93 therefore described almost nobody: a review profile in this sector was a ratio between two extreme groups, the delighted and the furious, with silence in between.
The two groups also wrote differently. The median negative review ran 267 characters; the median five-star review, 124. Anger itemised: dates, unanswered emails, defect lists. Praise summarised. A single detailed one-star account therefore carried more information, and often more persuasive weight, than several short five-star lines next to it.
The star rating attached to every collected text review, counted per star value over the full deduplicated set.
n = 2 971 / 2 971
- Negative (1–2 stars) 734 of 2 971 · 24,7%
- Middle (3 stars) 42 of 2 971 · 1,4%
- Positive (4–5 stars) 2 195 of 2 971 · 73,9%
Number of text reviews at each star value. Bars darken with the rating.
| Rating | Reviews | Share |
|---|---|---|
| 1 star | 670 | 22,6% |
| 2 stars | 64 | 2,2% |
| 3 stars | 42 | 1,4% |
| 4 stars | 222 | 7,5% |
| 5 stars | 1 973 | 66,4% |
| Total | 2 971 | 100% |
Finding 2
What did reviewers write about?
Negative reviews were about silence: 44,6% of them touched communication and reachability, more than any other theme, ahead of the sales experience (36,2%) and finishing quality (31,5%). Positive reviews were about people: 40,8% praised staff and attitude. The same relationship read from both ends: when it worked, buyers named the person who answered; when it failed, they described not being answered.
56,5% of negative reviews touched at least one of communication, delivery information or aftercare: the themes that live in how a developer keeps its buyers informed. Only 6,5% raised finishing quality without touching any of those three. A complaint about a defect almost never stood alone; what turned it into a one-star review was, in the reviewer’s own telling, what happened after it was reported.
Share of reviews in which a theme occurs, per multilingual keyword classification, multilabel. Negative = 1–2 stars, positive = 4–5 stars.
n = 734 negative · 2 195 positive
Themes in negative reviews
1–2 stars, n = 734. Multilabel, so shares do not total 100. 22,2% of negative reviews matched no theme.
Themes in positive reviews
4–5 stars, n = 2 195. Multilabel, so shares do not total 100.
| Theme | In negative reviews (n = 734) | In positive reviews (n = 2 195) |
|---|---|---|
| Communication & reachability | 44,6% (327) | 33,3% (730) |
| Staff & attitude | 16,1% (118) | 40,8% (895) |
| Sales experience | 36,2% (266) | 21,1% (463) |
| Finishing quality | 31,5% (231) | 14,9% (328) |
| Price & invoicing | 27,7% (203) | 8,3% (183) |
| Delivery & timing | 23,0% (169) | 9,3% (205) |
| Aftercare & warranty | 14,9% (109) | 2,5% (55) |
A theme share is an association, not a diagnosis. Better communication does not repair a badly built wall, and 31,5% of negative reviews did name finishing quality. What the data shows is narrower: in the reviewers’ own accounts, defects mostly became one-star reviews in combination with silence about them.
Finding 3
What exactly did the complaints and the praise name?
Inside the 327 negative communication reviews, the most named specific failure was a promise not kept (20,5%), ahead of total silence: no answer at all (16,2%), no updates on status or planning (12,8%), having to chase repeatedly (8,0%) and being ignored as a prospective buyer (6,7%). Inside the 109 negative aftercare reviews, almost half (47,7%) placed the moment the service broke at signing, payment or handover. Inside the 895 positive staff reviews, 61,9% praised professionalism and 40,0% friendliness.
Read together, the three lists describe the same job from three sides. What buyers punished was procedural: promises that evaporated, questions that died, status that stayed dark, service that stopped once the money was in. What buyers rewarded was the mirror image: professional, friendly people who answered fast and knew their file. None of the top clusters on either side concerned the building itself.
Within each subset, sub-clusters assigned with the same multilingual keyword approach, multilabel: a review can carry several clusters, and percentages are shares of the subset, not of all reviews.
n = 327 · 109 · 895
Inside negative communication reviews. The specifics, where reviewers gave them:
Specific failures named inside negative communication reviews (n = 327). Multilabel; 52,3% named no specific mechanism beyond poor communication or reachability.
| What the review named | Reviews | Share |
|---|---|---|
| A promise or commitment not kept | 67 | 20,5% |
| Total silence: no answer, unreachable | 53 | 16,2% |
| No updates on status, progress or planning | 42 | 12,8% |
| Had to chase: repeated calls and emails | 26 | 8,0% |
| Ignored as a prospective buyer | 22 | 6,7% |
| No specific mechanism named | 171 | 52,3% |
Inside negative aftercare reviews. Where the service broke, in the reviewers’ telling:
Sub-clusters inside negative aftercare reviews (n = 109). Multilabel; 19,3% matched none of these.
| What the review named | Reviews | Share |
|---|---|---|
| Service broke after signing, payment or handover | 52 | 47,7% |
| Reported defects stayed unresolved | 33 | 30,3% |
| Problem pointed to contractors or suppliers | 31 | 28,4% |
| Waiting for months or years | 19 | 17,4% |
| A warranty dispute | 18 | 16,5% |
Inside positive staff reviews. What earned the praise:
Sub-clusters inside positive staff-and-attitude reviews (n = 895). Multilabel; 7,7% matched none of the measured clusters.
| What the review praised | Reviews | Share |
|---|---|---|
| Professionalism | 554 | 61,9% |
| Friendliness and warmth | 358 | 40,0% |
| Speed of response | 242 | 27,0% |
| Expertise and advice | 184 | 20,6% |
| Personal approach and listening | 130 | 14,5% |
These clusters are a heuristic on top of a heuristic. The subsets come from the theme classification and the clusters from keyword lists within them, so the error compounds: treat the ordering as reliable and the exact shares as indicative. Half of the negative communication reviews named no specific mechanism at all, and the aftercare base is 109 reviews: enough to rank, too few for precision.
Finding 4
When in the buyer journey did reviews turn negative?
Order the themes by where they sit in the buyer journey and the tone flipped along the way. Of reviews touching staff and attitude, 11,6% were negative. For the sales experience the share was 36,0%, for communication during the project 30,6%, for delivery and timing 44,4%, for price and invoicing 51,8%, and for aftercare and warranty 66,1%. Reviews about the phase before the signature leaned positive; reviews about everything after the keys leaned negative, aftercare most of all.
Nine years earlier, our Belgian Homebuyer Survey 2017 found the same cliff with the opposite instrument. In that survey of 242 new-build buyers, 76,4% were satisfied with the sales approach and 54,7% with updates during construction, and the sector’s Net Promoter Score was −42. A survey in 2017 and public reviews in 2026 are two independent measurements, and they agreed on the shape: the experience thinned out after the signature, and the reputation was settled after the keys.
Per theme: the share of all reviews touching that theme (any star value) that awarded one or two stars. Journey position is an interpretation of the theme, not a field in the data.
bases per theme in the table
Share of theme-touching reviews that awarded one or two stars, ordered roughly along the buyer journey.
| Journey phase | Theme | Reviews touching it | Share negative |
|---|---|---|---|
| Before the signature | Staff & attitude | 1 019 | 11,6% |
| Before the signature | Sales experience | 738 | 36,0% |
| During construction | Communication & reachability | 1 067 | 30,6% |
| Around handover | Delivery & timing | 381 | 44,4% |
| After the keys | Price & invoicing | 392 | 51,8% |
| After the keys | Aftercare & warranty | 165 | 66,1% |
The gradient was not perfectly monotonic. Reviews touching the sales experience were negative slightly more often than reviews touching communication, largely because unhappy buyers retold the sales promises in their complaint. The direction over the journey as a whole is the finding; the ordering of two adjacent themes is not.
Finding 5
Did developers with more reviews score better?
Consistently. The 128 developers with one to nine text reviews averaged 3,63, with 28,5% one-star reviews. Every size class up scored higher, to 4,67 and a 6,0% one-star share for the three developers with a hundred or more. The gradient was monotonic across all five size classes.
Volume worked as a buffer: a company with hundreds of reviews was not defined by its angriest buyer, a company with four was. The gradient was not entirely organic either; Finding 6 returns to what solicitation added at the top of it.
Developers grouped by their number of collected text reviews; per group, the review-weighted average rating and the share of one-star reviews.
210 developers with at least one text review
Share of one-star reviews per size class. The bigger the review base, the smaller the one-star share.
| Size class | Developers | Reviews | Average rating | One-star share |
|---|---|---|---|---|
| 1–9 reviews | 128 | 403 | 3,63 | 28,5% |
| 10–24 reviews | 48 | 782 | 3,70 | 27,7% |
| 25–49 reviews | 16 | 516 | 3,83 | 26,2% |
| 50–99 reviews | 15 | 901 | 4,02 | 20,1% |
| 100+ reviews | 3 | 369 | 4,67 | 6,0% |
An association, not a mechanism. Developers who built large review bases differ from the rest in more ways than volume: they are bigger firms, they ask for reviews systematically, and they may also serve buyers better. This data cannot separate those. And the top class holds three companies: enough to show a direction, too few for precision.
Finding 6
How much of the review picture was organised?
Three timing and content signals were scored per review. 543 reviews (18,3%) sat inside a burst window: five or more five-star reviews for the same developer within 14 days, the trace of a mailing or QR campaign. 412 (13,9%) were five-star reviews under 80 characters matching no concrete theme. Two of the 329 profiles held a perfect 5,0 across more than 30 reviews. 111 reviews (3,7%) triggered at least two signals: the strict lower bound. The honest reading is a range: somewhere between 3,7% and 18,3% of the sector’s reviews showed the signature of organised solicitation.
The practice was concentrated and it was growing. Of the 82 developers with ten or more reviews, eight had more than half of their reviews inside burst windows, up to 76% at one company. The signals also concentrated at the top of the size classes from Finding 5: 65,0% of the reviews of the three developers with a hundred or more sat inside burst windows, and excluding those, their average fell from 4,67 to 4,05: still first, by a much smaller margin. And the burst share of each year’s reviews jumped from 7% or less before 2022 to 28–30% in 2025 and 2026. The sector’s apparent improvement in those years, an average of 4,09 in 2025 and 4,17 in 2026 against 3,65 in 2023, largely dissolved once burst reviews were excluded: 3,74 and 3,81. Most of the recent rise in scores was solicited, not earned.
Burst: five or more five-star reviews for one developer within 14 days of each other. Contentless: five stars, under 80 characters, no theme matched. Perfect profile: average exactly 5,0 over 30 or more reviews. Each signal is heuristic; none proves intent at any individual firm.
n = 2 971 / 2 971
Share of each year’s reviews that sat inside a five-star burst window. 2026 runs to 27 August.
| Measure | All text reviews | Excluding burst reviews |
|---|---|---|
| Reviews | 2 971 | 2 428 |
| Average rating | 3,93 | 3,69 |
| Five-star share | 66,4% | 58,9% |
| One-star share | 22,6% | 27,6% |
| One-or-five-star share | 89,0% | 86,5% |
| Average, reviews from 2025 | 4,09 | 3,74 |
| Average, reviews from 2026 | 4,17 | 3,81 |
A burst is not proof of manipulation. A developer that hands over an apartment block and asks its happy buyers for a review in that week produces exactly this trace, legitimately. The signals measure organisation of the asking, not honesty of the answers. No individual company can be judged on them, which is one reason this page names none.
Finding 7
How fast was the reviewing habit growing?
Yearly review volume roughly tripled in four years: from 230 text reviews in 2021 to 588 in 2025, with 2026 running at 469 by 27 August, a pace of about 716 over a full year. 42,1% of all reviews in the dataset were less than two years old. The growth ran on both sides of the ledger: between 2021 and 2025, positive reviews grew from 167 to 457 a year and negative reviews from 57 to 127.
The growth came from depth, not breadth. The number of developers reviewed in a year rose about 40%, from 77 in 2021 to 108 in 2025, while volume grew far faster: the average reviewed developer went from 2,3 text reviews a year in 2019 to 5,4 in 2025. Fewer newcomers entered the review economy each year: 34 developers received their first text review in 2020, 13 in 2025 and 9 in 2026 to 27 August, bringing the total with at least one review to 210 of 329. The writing concentrated on companies already being reviewed.
The sector nonetheless remained thin. The median reviewed developer held 5 text reviews in total, 96 of the 210 held four or fewer, and 119 of the 329 (36,2%) had no text review at all, 65 of them (19,8%) not even a star rating. For most of the sector one review still moved the profile, and for a third, the next review, good or bad, would be the profile.
Unique text reviews counted by the year of their review date; per year, the number of distinct developers receiving at least one, and the ratio of the two. Ten reviews from 2015 and 2016 are included in totals but omitted from the charts.
n = 2 971 / 2 971
Text reviews per year of writing. 2026 covers roughly eight months; at that pace the full year lands around 716.
Average number of text reviews per developer reviewed in that year. The habit deepened faster than it spread.
| Year | Text reviews | Developers reviewed | Reviews per developer | Negative | Positive |
|---|---|---|---|---|---|
| 2019 | 139 | 60 | 2,3 | 32 | 98 |
| 2020 | 187 | 80 | 2,3 | 57 | 125 |
| 2021 | 230 | 77 | 3,0 | 57 | 167 |
| 2022 | 377 | 94 | 4,0 | 87 | 287 |
| 2023 | 360 | 110 | 3,3 | 117 | 240 |
| 2024 | 462 | 111 | 4,2 | 130 | 330 |
| 2025 | 588 | 108 | 5,4 | 127 | 457 |
| 2026 (to 27/8) | 469 | 92 | 5,1 | 93 | 373 |
What this study does not show
- It is a snapshot. Every figure describes the public review picture as collected on 27 August 2026. Scores and volumes change daily.
- Reviews are not a satisfaction survey. People with strong experiences write; everyone else stays silent. The 89,0% polarisation is partly an artefact of that self-selection. This page measures the public record, not the true distribution of buyer satisfaction; for that, see the Homebuyer NPS Study 2026 and the Belgian Homebuyer Survey 2017.
- Text reviews only, capped at 200 per profile. Star-only ratings were not collected, so averages here differ from the profile score visible on Google.
- The theme classification is heuristic. Keyword matching in three languages misses irony and implicit complaints. Theme shares are reliable as an order of magnitude, not to the percentage point, and the sub-clusters in Finding 3 sit on top of that classification and inherit its error twice.
- The solicitation signals prove no intent. A burst can be a legitimate, well-timed request after a delivery. The range of 3,7% to 18,3% describes patterns in timing and content, and no individual developer can be judged on it.
- Reviewers are not verified buyers. Google does not check who writes; some reviews are visibly written by business partners, suppliers or acquaintances rather than customers.
- Associations are not mechanisms. Developers with large review bases, and developers who solicit reviews, differ from the rest in many ways at once.
- Ziggu is the author. This is our own research, published by a company that sells software to the sector it studied. The definitions, bases and pipeline are on this page so the figures can be rebuilt rather than taken on trust.
- No developer is named and no review is quoted. The dataset holds names and full texts; the published page reports aggregates only, because a verbatim quote is searchable and would identify the company and the reviewer.
Citing this research
Ziggu (2026). Google Review Study: Belgian Property Developers 2026. Analysis of 2 971 public Google review texts for 329 residential property developers in Belgium, collected 27 August 2026.
