Ziggu Research · Portugal · Collected 18 August 2026
Google Review Study: Portuguese Property Developers 2026
What 271 public Google review texts for 45 residential property developers in Portugal said about the buyer experience: how the ratings split, what the complaints named, and when in the last decade the tone turned.
In 271 public Google reviews of 45 residential property developers in Portugal, collected in August 2026, the complaints sat after the sale rather than during it. Of the 48 substantive buyer complaints, 58,3% named after-sales and warranty handling, ahead of construction quality (45,8%) and communication and responsiveness (43,8%); communication was the most widespread, appearing at 12 of the 45 developers. The ratings were sharply split: 182 of the 271 reviews awarded five stars and 74 awarded one, with only fifteen in between. And the negative share rose over the period, from 17,9% of reviews written between 2016 and 2023 to 42,0% of those written from 2024 onwards.
Ziggu collected the public text reviews on the verified Google Maps profiles of residential developers active in Portugal, up to the 30 most recent per profile. 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 how the ratings split, what the complaints named once they were read and coded, when in the last decade the tone changed, and what the praise described instead.
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 the complaints quoted in our internal working paper have been paraphrased for this page rather than reproduced: see the method section for why. This is a small study, and it is not a smaller copy of our Belgian one. It rests on 271 reviews against 2 971, on manual coding against an automated classifier, and on a collection cap of 30 reviews per profile against 200. The two are not comparable measurement to measurement, and no figure here should be read against a figure in the Google Review Study: Belgian Property Developers 2026.
The figures at a glance
| Finding | Value | Base |
|---|---|---|
| Reviews awarding one or five stars | 94,5% | n = 271 |
| One-star share of all reviews | 27,3% | n = 271 |
| One-star share of reviews rated three stars or fewer | 92,5% | n = 80 |
| Average rating across all text reviews | 3,82 | n = 271 |
| Developers with at least one review of three stars or fewer | 30 of 45 | n = 45 |
| Substantive complaints naming after-sales and warranty | 58,3% | n = 48 |
| Substantive complaints naming construction quality | 45,8% | n = 48 |
| Substantive complaints naming communication | 43,8% | n = 48 |
| Developers at which a communication complaint appeared | 12 of 45 | n = 45 |
| Substantive complaints carrying two or more themes | 62,5% | n = 48 |
| Quality complaints that also named a service failure | 81,8% | n = 22 |
| Negative share, reviews written 2016 to 2023 | 17,9% | n = 140 |
| Negative share, reviews written 2024 to August 2026 | 42,0% | n = 131 |
| Positive reviews mentioning people or service | 41,4% | n = 191 |
| Positive reviews mentioning deadlines or punctuality | 2,1% | n = 191 |
| Median length, negative against positive reviews | 177 / 69 characters | n = 80 / 191 |
| Five-star reviews shorter than 80 characters | 59,9% | n = 182 |
| Share of all negative reviews at the five most-criticised profiles | 48,8% | n = 80 |
Method
The starting population was a list of 110 companies active in residential development in Portugal, compiled from new-build listing portals and business databases and deduplicated by company domain. For each, the official Google Maps profile was identified and manually verified. Profiles were excluded when none could be found (38), when the profile carried no written reviews (9), when it belonged to a different entity (3), when the company name could not be read reliably (6), and when the profile turned out to be a hotel, resort or rental operation rather than a residential developer (8, holding 114 review texts). One company appeared twice under two domains with an identical review set; the duplicate was merged. The result: 45 verified developer profiles and 271 public reviews with written text, collected on 18 August 2026 at up to the 30 most recent reviews per profile, sorted newest first. Reviews span June 2016 to August 2026, and 80,8% were written from 2022 onwards. Ratings without written text were not collected, so averages here differ from the profile score visible on Google, which includes them.
All 80 reviews rated three stars or fewer were read in full and coded by hand into five complaint themes. The coding is multilabel: a review naming both an unresolved defect and an unanswered email counts for both. Fourteen were set aside as falling outside the buyer relationship: planning and environmental objections, construction-site nuisance, tenancy disputes, complaints from business partners or investors, and remarks about the review profile itself. Eighteen carried no codable content beyond a verdict (“Péssimo”, “Não gostei”). That leaves 48 substantive complaints as the base for the theme figures. Percentages on a base of 48 carry a margin of roughly ±14 percentage points at 95% confidence for a share near 50%, so the theme figures rank the themes; they do not measure them to the point. Positive reviews were not coded by hand: the two figures reported for them come from a keyword scan across Portuguese and English, which is a coarser instrument and is labelled as such where it appears.
Reviews are paraphrased on this page, not quoted. The wording of every passage has been rewritten and identifying details removed, so that no passage can be traced back to a company or a reviewer through a search engine. The meaning of each is preserved. In a market of 45 developers a verbatim quotation identifies the company within one search, and often the reviewer with it, and this page reports the market rather than any firm in it. The paraphrases therefore carry no evidential weight and are marked as paraphrases wherever they appear: the coded counts carry the findings. Percentages are rounded to one decimal, so a column may not total exactly 100. Every count on this page was computed from the raw dataset by script, and the manual coding is recorded review by review so it can be re-run.
Finding 1
How did the ratings split?
Of 271 text reviews, 67,2% awarded five stars and 27,3% awarded one. Fifteen reviews in total, 5,5%, sat anywhere in between. Of the 80 reviews rated three stars or fewer, 92,5% were one-star: buyers did not use a public review to register mild disappointment. They used it once other channels had failed.
The two groups also wrote differently. The median negative review ran 177 characters, the median positive review 69, and 109 of the 182 five-star reviews were shorter than 80 characters. Complaint itemised: dates, unanswered emails, lists of defects. Praise summarised. Negative experience was also concentrated rather than spread: the five most-criticised profiles held 48,8% of all negative reviews, and 30 of the 45 developers carried at least one.
The star rating attached to every collected review, counted per star value over the deduplicated set. Review length is the character count of the collected text.
n = 271 / 271
- Three stars or fewer 80 of 271 · 29,5%
- Four or five stars 191 of 271 · 70,5%
Number of reviews at each star value. Bars darken with the rating.
| Rating | Reviews | Share |
|---|---|---|
| 1 star | 74 | 27,3% |
| 2 stars | 3 | 1,1% |
| 3 stars | 3 | 1,1% |
| 4 stars | 9 | 3,3% |
| 5 stars | 182 | 67,2% |
| Total | 271 | 100% |
A bimodal profile is what public review channels produce, not a property of this sector. People write when they are delighted or when they feel ignored, and the middle of the satisfaction curve stays silent. The split describes who wrote, not how often things went wrong.
Finding 2
What did the complaints name?
Of the 48 substantive complaints, 58,3% named after-sales and warranty handling, ahead of construction quality and finishing (45,8%) and communication and responsiveness (43,8%). Delivery delays followed at 20,8% and money at 4,2%. Communication was not the largest theme but it was the most widespread: it appeared at 12 of the 45 developers, more than any other, against eleven for after-sales and ten for quality.
The themes rarely arrived alone. 62,5% of the substantive complaints carried two or more, and the pairing was consistent: of the 22 complaints that named a construction defect, 18 also named a service failure around it. Only four complaints in the whole set, 8,3%, described a defect without also describing what happened when it was reported. In the reviewers’ own accounts the defect was the trigger, and the silence afterwards was the grievance.
Share of the substantive complaints in which a theme occurs, coded manually, multilabel. Developers concerned counts the distinct profiles at which a theme appeared at least once.
n = 48 substantive complaints of 80 reviews rated three stars or fewer
Substantive complaints naming each theme, n = 48. Multilabel, so shares do not total 100.
| Theme | Complaints | Share (n = 48) | Developers concerned |
|---|---|---|---|
| After-sales & warranty handling | 28 | 58,3% | 11 |
| Construction quality & finishing | 22 | 45,8% | 10 |
| Communication & responsiveness | 21 | 43,8% | 12 |
| Delivery delays & missed deadlines | 10 | 20,8% | 4 |
| Deposits, refunds & payments | 2 | 4,2% | 2 |
After-sales and warranty. The most consistent storyline in the data: the purchase went well and everything after it did not. Reviewers described defect reports that went unanswered for months, warranty claims contested rather than resolved, and a service relationship that ended at the notary. Several said they had begun or were preparing legal action, and described an escalation that started with silence rather than with the defect.
They were excellent at selling and at promising. After the sale that changed: no replies, and almost nothing put right.
Paraphrased, not quoted. Wording rewritten and identifying details removed. Residential developer, 2024.
I have written about the same urgent fault five times in six months and had no reply at all.
Paraphrased, not quoted. Wording rewritten and identifying details removed. Residential developer, 2025.
Communication and responsiveness. Not the largest theme, but the one that reached the most companies, and the one that turned other problems into grievances. Of the 28 after-sales complaints, 9 also named unanswered calls and emails or vague and shifting information as what escalated the matter, and 5 of the 10 delay complaints did the same.
The information we were given about how the development was progressing was vague, kept changing, and told us very little.
Paraphrased, not quoted. Wording rewritten and identifying details removed. Residential developer, 2025.
Construction quality and finishing. Complaints described water infiltration, defective installations, poor finishing, and common areas that did not match what was sold. Almost every one of them doubled as a service complaint.
What was advertised and what was actually built are not the same thing, and the faults we reported are not being fixed.
Paraphrased, not quoted. Wording rewritten and identifying details removed. Residential developer, 2025.
Delivery delays. Concentrated at four developers, but the longest reviews in the set. Buyers counted the months and years and listed the deadlines announced and missed. What they objected to was rarely the delay itself: it was learning about it late, informally, or not at all.
Deadlines are not met and the dates keep changing, with no formal notice and no proper explanation.
Paraphrased, not quoted. Wording rewritten and identifying details removed. Residential developer, 2026.
Deposits, refunds and payments. The smallest theme, at two complaints, and no quantitative claim rests on it. It is reported because these were the fastest to escalate: both invoked consumer authorities or the courts.
A base of 48 ranks themes; it does not measure them. The margin is roughly ±14 percentage points, so the gap between quality at 45,8% and communication at 43,8% is not a real difference, and a second coder drawing the line differently on which reviews count as buyer-related would move every share by a few points. What survives that uncertainty is the ordering at the top and the bottom, and the pairing: complaints about the building almost never travelled alone.
Finding 3
When did the reviews turn negative?
The negative share roughly doubled across the period. Of reviews written between 2016 and 2023, 17,9% awarded three stars or fewer; of those written from 2024 to August 2026, 42,0% did. The shift was not gradual: it stepped up between 2023 (22,8%) and 2024 (48,1%) and stayed there through 2025 (46,8%), easing to 35,1% in 2026 to date.
Those are the years in which units sold off-plan during Portugal’s 2019 to 2022 development boom were handed over and their first defects surfaced. The dataset cannot demonstrate that link: it holds no project dates, no handover dates and no delivery volumes, only the date each review was written. What it shows is the timing, and the timing is consistent with a delivery wave arriving.
Reviews grouped by the year they were written; per group, the share awarding three stars or fewer. 2026 runs to 18 August.
n = 271 / 271
Share of each period’s reviews awarding three stars or fewer. Darker bars mark the higher shares.
| Period | Reviews | Three stars or fewer | Negative share |
|---|---|---|---|
| 2016–2021 | 52 | 7 | 13,5% |
| 2022 | 31 | 5 | 16,1% |
| 2023 | 57 | 13 | 22,8% |
| 2024 | 27 | 13 | 48,1% |
| 2025 | 47 | 22 | 46,8% |
| 2026 (to 18 Aug) | 57 | 20 | 35,1% |
| Total | 271 | 80 | 29,5% |
Read this as a direction, not a measurement. The yearly subsamples are small, 27 reviews in 2024, and the collection method takes the 30 most recent reviews per profile, which loads the sample towards recent years and can drop older reviews at the busiest profiles entirely. Both effects push in the same direction as the finding, so the size of the rise is unreliable even though its presence is visible in every year from 2024 on.
Finding 4
What did the praise describe?
Not the same phase of the relationship. A keyword scan across the 191 positive reviews found that 41,4% mentioned people or service: professionalism, attentiveness, the sales team, often by name. 2,1% mentioned deadlines or punctuality. In the 80 negative reviews the proportions inverted for timing, at 13,8%, while quality terms appeared in 30,0% against 12,6% of positive reviews.
Praise and complaint therefore described two different halves of the same relationship. Developers earned public credit for the buying experience and public criticism for the owning experience. Read alongside Finding 2, where 18 of 22 defect complaints also named a service failure, the complaint themes are all information problems before they are construction problems: unanswered messages, unmanaged expectations, invisible progress, unresolved defects.
A keyword scan in Portuguese and English over three term groups: people and service, deadlines and punctuality, quality and finishing. A review counts once per group. This is a coarser instrument than the manual coding used in Finding 2 and it misses irony, negation and implicit praise.
n = 191 positive · 80 negative
Term groups appearing in the 191 reviews rated four or five stars. Keyword scan, not manual coding.
| Term group | In positive reviews (n = 191) | In negative reviews (n = 80) |
|---|---|---|
| People & service | 41,4% (79) | 32,5% (26) |
| Quality & finishing | 12,6% (24) | 30,0% (24) |
| Deadlines & punctuality | 2,1% (4) | 13,8% (11) |
Part of the positive volume looked organised rather than spontaneous. Three profiles holding eight or more reviews consisted entirely of five-star ratings with no critical voice, and 109 of the 182 five-star reviews were shorter than 80 characters. One reviewer alleged directly that a developer had asked suppliers and staff to post favourable reviews. This is the pattern an invited-review campaign leaves, and a well-timed request after handover leaves the same trace legitimately. It is also why this study analyses complaint content and does not rank developers or interpret star averages.
Cross-check
Does a second channel show the same pattern?
Qualitatively, yes. Portal da Queixa, Portugal’s largest consumer-complaints platform, summarises its civil-construction category around the same themes this dataset produced: delays in completing works, lack of communication, and poor service execution, with dissatisfaction about the transparency and effectiveness of construction companies. On its Satisfaction Index, most listed construction brands, including some of the country’s largest builders, sit in the weak band.
This is not our data and it carries no figure on this page. Coverage of residential developers on that platform is too thin for quantitative use: individual developer profiles typically hold a handful of complaints. It appears here for one purpose only, which is to show that the themes found on Google Maps are not an artefact of a single channel.
What this study does not show
- It is small, and it is a snapshot. 271 reviews across 45 developers, collected on 18 August 2026. Every figure describes the public record on that date.
- It is not comparable to our Belgian study. Different scale (271 reviews against 2 971), different collection cap (30 per profile against 200), and different method (manual coding against an automated multilingual classifier). No figure on this page should be read against a figure in the Google Review Study: Belgian Property Developers 2026.
- Reviews are not a satisfaction survey. Dissatisfied buyers write more, and write longer, and part of the positive volume shows the shape of invited reviews. This measures the content and shape of public dissatisfaction, not how often buyers are dissatisfied. For satisfaction measured by asking, see the Homebuyer NPS Study 2026.
- The complaint coding is manual and it is one coder’s judgement. The base of 48 carries a margin of roughly ±14 percentage points, and the boundary between a buyer complaint and a tenancy, investor or planning complaint is a judgement call that moves the base. Treat the ordering as the finding.
- The paraphrases are not evidence. Every passage on this page has been rewritten and stripped of identifying detail so it cannot be traced by search. That protects the companies and the reviewers, and it means the passages cannot be verified against the source. The coded counts, not the paraphrases, carry the findings.
- The collection cap biases the timeline. Taking the 30 most recent reviews per profile loads the sample towards recent years, so Finding 3 shows a direction and not a rate of change.
- Negative experience is concentrated. Five profiles hold 48,8% of all negative reviews. This study draws no conclusion about how often things go wrong in the market, only about what goes wrong when buyers feel compelled to say so in public.
- Reviewers are not verified buyers. Google does not check who writes. Some reviews are visibly written by suppliers, partners, tenants or passers-by; those identified as such were excluded from the coding, but the star distribution includes them.
- No developer is named. The dataset holds company names and full review texts. This page reports the market in aggregate, because in a set of 45 developers a named example is a public accusation.
- Ziggu is the author. This is our own desk research, published by a company that sells client-communication software to the sector it studied. The definitions, bases and exclusions are on this page so the figures can be rebuilt rather than taken on trust.
Citing this research
Ziggu (2026). Google Review Study: Portuguese Property Developers 2026. Analysis of 271 public Google reviews with written text for 45 residential property developers in Portugal, collected 18 August 2026, with 48 substantive complaints coded manually into five themes.
