A reply is a step, not a confirmed remedy
In Brazil’s published Consumidor.gov.br records finalized during 2025, the mixed books-and-stationery category contained 3,049 rows. Companies had replied in 2,515, or 82.5%. Yet 499 replied rows also carried a consumer assessment of unresolved. Those rows represent 19.8% of replies and 76.2% of all 655 consumer-unresolved rows in the selected category. An answer and a positive consumer assessment therefore describe different stages of complaint handling. A reader deciding whether a problem is finished needs to look beyond the existence of a response.
The issue profile gives that distinction a practical use. Delivery and billing/dispute groups together account for 365 of the 655 assessed-unresolved rows, or 55.7%. More narrowly, the standardized non-delivery/delayed-delivery issue and the refund/delayed-repayment issue together account for 262, exactly 40.0%. These are places to investigate follow-through, rather than evidence that sending another generic acknowledgment will solve a problem. The analysis does not test any intervention or establish why a complaint remained unresolved.
The category includes more than books
The official subject is “Livros e papelaria (material didático, artigos de escritório, etc.)”: books and stationery, including teaching materials and office supplies. Every calculation uses that complete category. The files do not identify which rows concern books, notebooks, other stationery or office goods. They also provide no reliable split between printed books and ebooks. Calling these figures a book-only complaint rate would change the population after measurement and would be unsupported.
Separate categories for news and periodicals, and for subscriptions mixing books with other products such as wine, are excluded. Retailer or publisher names are not used to guess a narrower product. A business can sell several kinds of goods, so its name cannot establish what an individual complaint concerned. The result is deliberately a mixed-category complaint-handling reference. Authors, booksellers and readers can use its process distinctions, but cannot treat it as a picture of the entire Brazilian book market.
The unit is a published row. The source does not provide a stable complaint identifier, and the analysis does not infer unique consumers, transactions or distinct cases. Identical observable rows are retained because separate records can share the same classifications. Repeated platform use by one consumer is also possible. Counts throughout this article refer to rows in the acquired files, even when the surrounding explanation uses the ordinary language of complaint handling.
What the six combinations actually show
Among rows with a company response, 763 were consumer-resolved, 499 consumer-unresolved and 1,253 unassessed. Among rows without a recorded response, the corresponding counts were 67, 156 and 311. These six combinations reconcile to 3,049. They prevent two common shortcuts: treating a response as proof of resolution, and treating the absence of a response as automatic evidence of failure. The 67 no-response rows assessed as resolved are especially clear evidence that these administrative fields are not interchangeable.
There are several possible reasons why response and assessment might differ, but this source does not identify them. A remedy could occur through another channel, a formal answer could promise an action that had not happened, or the fields could reflect other administrative circumstances. Those are possible explanations, not findings from the data. No consumer narratives were read to choose a story, and the analysis does not claim to know what a particular company or consumer did outside the platform.
The response measure answers whether a reply was recorded. The assessment measure answers how the consumer classified the outcome, when feedback was supplied. Neither directly measures the accuracy of an explanation, the amount of money recovered, the time until a replacement arrived, or how much effort the consumer spent. Keeping these boundaries visible makes the counts more useful: they show where an additional outcome check may matter without pretending to describe the entire dispute.
Most missing information is consumer feedback
Across all selected rows, consumers marked 830 resolved and 655 unresolved; 1,564 were unassessed. The shares are 27.2%, 21.5% and 51.3%. Among the 1,485 assessed rows alone, 55.9% were resolved. That conditional percentage describes assessed rows with a recorded outcome evaluation. It cannot be applied to the unassessed half of the file without an assumption about why feedback is missing.
The official platform documentation uses another clearly stated convention for its solution index: resolved plus unassessed rows, divided by all finalized rows. Applying that formula to this exact category and closure year gives 78.5%. This is a reconstruction, not a value read from a live provider dashboard. The convention is already explained by the government and has been discussed elsewhere. Explaining it is necessary context, rather than the original finding of this study.
The 51.3 percentage-point distance between 78.5% and 27.2% equals the unassessed share by arithmetic. It is not an estimate of hidden failures, evidence that every silent consumer was satisfied, or proof that the platform’s index is erroneous. The correct reading is that the two indicators answer different questions. One follows the published administrative convention; the other requires an explicit positive consumer assessment. Missing feedback stays missing in the present analysis.
Delivery and billing dominate the unresolved workload
Delivery is the largest official issue group, with 1,008 rows. It contains 294 consumer-resolved, 210 consumer-unresolved and 504 unassessed rows. Companies replied in 775, or 76.9%. Delivery contributes 32.1% of all 655 assessed-unresolved rows. That workload share combines the group’s volume and its observed feedback composition. It does not mean that 32.1% of deliveries failed, because the number of purchases or deliveries is unknown.
Billing and disputes contain 864 rows, with 250 resolved, 155 unresolved and 459 unassessed. Recorded replies cover 86.0% of the group. Its 155 unresolved rows contribute 23.7% of the selected category’s unresolved workload. Together, delivery and billing supply the 55.7% concentration. This gives a complaint team a concrete starting point for reviewing whether promised shipping actions, charge corrections or repayment steps reach a verifiable conclusion.
The contrast also shows why reply coverage alone is an incomplete planning measure. Billing has higher recorded response coverage than delivery, but still contributes a substantial number of unresolved assessments. A team using only unanswered rows as its follow-up queue would miss replied rows that consumers later marked unresolved. The study does not show that a particular follow-up rule is optimal; it shows that these two observable queues are different.
Workload contribution and a within-group percentage also answer separate questions. Delivery’s 210 unresolved rows are 32.1% of all 655 unresolved rows, but 20.8% of its own 1,008 rows. Billing’s 155 are 23.7% of all unresolved rows and 17.9% of its own 864. The first pair helps describe where observed follow-up volume sits. The second pair describes the feedback composition inside each group. Neither supplies the number of successful deliveries or accurate bills outside this selected platform cohort, and neither adjusts for differences in products, companies or consumers. A smaller percentage can still accompany a substantial workload when a group is large.
A more precise view: arrival and repayment
The detailed non-delivery or delayed-delivery issue contains 860 rows: 251 resolved, 177 unresolved and 432 unassessed. Companies replied in 653. Within that issue, 119 replied rows were still assessed unresolved. It is narrower than the delivery group, which also includes incomplete or different products and other delivery classifications. The 177 unresolved rows account for 27.0% of the entire category’s assessed-unresolved total.
The detailed difficulty or delay in returning money, refund or retention-of-funds issue contains 502 rows. Of these, 172 were resolved, 85 unresolved and 245 unassessed; 416 had recorded replies. Its 85 unresolved rows contribute 13.0% of the category total. Combining this issue with non-delivery/delay gives the 262-row, 40.0% concentration. This calculation uses two distinct detailed labels and does not add them to the overlapping group-level figures.
For a consumer, the distinction suggests a useful question after receiving an answer: has the specific missing step happened? An arrival-related answer and a repayment-related answer require different evidence of completion. The source does not verify parcel tracking, bank credits or replacement shipments, so these examples are practical applications rather than measured remedies. They help translate a general response statistic into a check that matches the problem being discussed.
Smaller groups need context, not a league table
Contract and offer issues contain 522 rows, with 128 resolved, 126 unresolved and 268 unassessed. Customer-service issues contain 338, with 82 resolved, 84 unresolved and 172 unassessed. Recorded response coverage is 83.1% and 87.9%, respectively. Their contribution to unresolved workload is 19.2% and 12.8%. Neither group can be judged by response coverage alone, and neither tells us which merchant delivers better overall service.
The personal-data and privacy group is smaller: 162 rows, with 30 resolved, 51 unresolved and 81 unassessed. Among its 81 assessed rows, 37.0% were resolved. That is a descriptive flag for closer examination of this selected feedback, not a finding that privacy complaints are inherently harder to solve. The composition of consumers, companies and issues is unadjusted. The analysis does not inspect the content of personal-data complaints or publish any individual information.
Quality-defect issues contain 122 rows, with 41 resolved, 18 unresolved and 63 unassessed. Thirty-three rows in smaller groups are pooled for presentation. A lower count does not establish a safer product class, and a higher conditional resolution percentage does not establish better products. These are complaint records from participating platform users, not a representative comparison of products purchased in Brazil.
A practical way to read a complaint outcome
Start with the same question for each statistic: what event was observed, and on which base? A response rate concerns recorded replies. A consumer-confirmed resolution share on all rows requires a positive assessment. An assessed-only resolution percentage excludes rows without feedback. The reconstructed platform-style indicator includes unassessed records in its numerator. Each can be useful if the description accompanies the number; replacing one with another changes its meaning.
For an individual unresolved problem, keep the distinction between an acknowledgment, a proposed remedy and a completed remedy clear in your own records. When supplying feedback, describe the actual result through the platform’s available outcome fields. This is a general documentation practice, not legal advice or a promise that a platform can compel a particular result. The analysis does not establish the best escalation route for an individual dispute.
For complaint handlers, a useful review can maintain separate counts for unanswered records, replied records later assessed unresolved, and records without consumer feedback. The 2025 mixed-category file shows why those queues should not be collapsed. Delivery and billing are sensible starting points for an operational review because they contain more than half the observed unresolved assessments. Any claim that a changed process improves outcomes would require a separate evaluation, with suitable comparison and follow-up data.
The joint fields can also be audited within individual issue groups. Delivery has 233 rows without a recorded reply and another 144 with a reply but an unresolved consumer assessment. Billing has 121 without a reply and another 119 replied-but-unresolved rows. Each pair is mutually exclusive because a row cannot simultaneously have response status S (sim, yes) and N (não, no). The numbers describe historical finalized records, not an active service backlog, distinct consumers or verified outstanding obligations. They nevertheless show what a reply-only completion measure leaves outside view. The no-reply part includes 31 delivery and 14 billing rows reported resolved, so a flag is not proof of unfinished work.
A simple addition of all unresolved rows to all unanswered rows would double-count their intersection. Delivery has 66 rows that are both unanswered and consumer-unresolved; billing has 36. The union of those two observable flags is therefore 377 delivery rows and 240 billing rows, obtained either by subtracting that overlap or by adding the mutually exclusive categories above. Unassessed rows without a reply are included in these unions because of their reply status, not because missing feedback is interpreted as failure. These are examples of an auditable classification rule, not estimates of how much work a team should perform today. A future operating review would need current records, the content of the promised actions and evidence of actual completion before deciding what requires attention.
How the files were checked
All twelve official finalized-month archives for 2025 were acquired through the publisher’s publicly declared download route and fingerprinted. They contain 2,328,863 published rows across all subjects. An exact subject match selects 3,049. Every closure date falls within its file’s month. January through December are all present. This is a closure-year cohort, so a record finalized in 2025 may concern a purchase or complaint initiated earlier. Registration-date or purchase-season conclusions are unavailable.
The protocol and reporting thresholds were fixed before joint outcomes were calculated. Issue groups with at least 50 rows and detailed labels with at least 100 are shown separately; smaller labels are pooled at their respective level. All response, feedback and finalization-status values reconcile without unexpected states. Independent reconstruction reproduced the safe aggregate cube and results from the twelve original files.
September’s file adds two columns beyond the common header, including a possible consumer narrative field. A documented schema amendment preserves only the seven preselected safe analysis columns. Extra values are not examined or exported. No company names, demographic fields, cities, consumer scores or narratives appear in the public aggregates. The downloadable code and tables reproduce the calculations without redistributing source rows.
What remains unknown
These are complete counts of the selected rows in the acquired published files, not a probability sample of purchases or a census of consumer problems. No sampling interval is supplied. That does not eliminate classification error, platform selection, repeated participation, missing feedback, possible publication duplication or later provider revisions. Without identifiers, identical rows cannot safely be treated as duplicates, and without sales counts there is no buyer complaint rate.
The mixed subject prevents a book-only or format-specific conclusion. The lack of response content prevents judgment about the quality of an answer. Missing feedback prevents assignment of an actual outcome to half the records. The useful finding is narrower and actionable: many consumer-unresolved rows already had replies, and delivery and billing concentrate much of that observed workload. A reply is worth recording, but a review of whether the problem was actually resolved needs its own evidence.
Replies and consumer feedback are separate results
Consumer feedback within each issue group, with recorded response share alongside. Published rows, not unique consumers.
Full chart and source
Scroll horizontally to inspect the chart. ↔
Chart and values
Scroll horizontally to see all columns.
| Issue | Rows n | Response % | Reported resolved n | Reported unresolved n | Unassessed n | Share of all unresolved % |
|---|---|---|---|---|---|---|
| All selected rows | 3,049 | 82.5 | 830 | 655 | 1,564 | 100.0 |
| Delivery | 1,008 | 76.9 | 294 | 210 | 504 | 32.1 |
| Billing / disputes | 864 | 86.0 | 250 | 155 | 459 | 23.7 |
| Contract / offer | 522 | 83.1 | 128 | 126 | 268 | 19.2 |
| Customer service | 338 | 87.9 | 82 | 84 | 172 | 12.8 |
| Personal data / privacy | 162 | 80.2 | 30 | 51 | 81 | 7.8 |
| Quality defects | 122 | 89.3 | 41 | 18 | 63 | 2.7 |
| Other groups | 33 | 81.8 | 5 | 11 | 17 | 1.7 |
All rows finalized in 2025 in the exact mixed books-and-stationery subject. Not unique consumers, all buyers, book-only outcomes or merchant quality. Group/detail levels overlap and must not be added. Original Portuguese labels and full precision are in downloads. The all-rows summary is not additive to its groups.
Data & downloads
Scroll horizontally to see all columns.
| Company response | Consumer feedback | Published rows |
|---|---|---|
| Reply recorded | Reported resolved | 763 |
| Reply recorded | Reported unresolved | 499 |
| Reply recorded | Unassessed | 1,253 |
| No reply recorded | Reported resolved | 67 |
| No reply recorded | Reported unresolved | 156 |
| No reply recorded | Unassessed | 311 |
All rows finalized in 2025 in the exact mixed books-and-stationery subject. Not unique consumers, all buyers, book-only outcomes or merchant quality. Group/detail levels overlap and must not be added. Original Portuguese labels and full precision are in downloads. The all-rows summary is not additive to its groups.
Scroll horizontally to see all columns.
| Issue | Rows n | Response % | Reported resolved n | Reported unresolved n | Unassessed n | Share of all unresolved % |
|---|---|---|---|---|---|---|
| All selected rows | 3,049 | 82.5 | 830 | 655 | 1,564 | 100.0 |
| Delivery | 1,008 | 76.9 | 294 | 210 | 504 | 32.1 |
| Billing / disputes | 864 | 86.0 | 250 | 155 | 459 | 23.7 |
| Contract / offer | 522 | 83.1 | 128 | 126 | 268 | 19.2 |
| Customer service | 338 | 87.9 | 82 | 84 | 172 | 12.8 |
| Personal data / privacy | 162 | 80.2 | 30 | 51 | 81 | 7.8 |
| Quality defects | 122 | 89.3 | 41 | 18 | 63 | 2.7 |
| Other groups | 33 | 81.8 | 5 | 11 | 17 | 1.7 |
All rows finalized in 2025 in the exact mixed books-and-stationery subject. Not unique consumers, all buyers, book-only outcomes or merchant quality. Group/detail levels overlap and must not be added. Original Portuguese labels and full precision are in downloads. The all-rows summary is not additive to its groups.
Scroll horizontally to see all columns.
| Issue | Rows n | Response % | Reported resolved n | Reported unresolved n | Unassessed n | Share of all unresolved % |
|---|---|---|---|---|---|---|
| Non-delivery / delayed delivery | 860 | 75.9 | 251 | 177 | 432 | 27.0 |
| Other detailed issues | 823 | 87.6 | 181 | 182 | 460 | 27.8 |
| Refund / delayed repayment / retained funds | 502 | 82.9 | 172 | 85 | 245 | 13.0 |
| Unfulfilled offer / service not supplied / misleading sale or advertising | 378 | 81.0 | 101 | 94 | 183 | 14.4 |
| Difficulty contacting / slow service | 191 | 88.0 | 49 | 40 | 102 | 6.1 |
| Incomplete product / different from order | 148 | 82.4 | 43 | 33 | 72 | 5.0 |
| Poor service (discourtesy / unpreparedness / embarrassment) | 147 | 87.8 | 33 | 44 | 70 | 6.7 |
All rows finalized in 2025 in the exact mixed books-and-stationery subject. Not unique consumers, all buyers, book-only outcomes or merchant quality. Group/detail levels overlap and must not be added. Original Portuguese labels and full precision are in downloads. The all-rows summary is not additive to its groups.
Evidence & method
How we calculated it
- All twelve official finalized-month 2025 archives; exact subject match; published row unit with no deduplication.
- Seven safe categorical fields only. September’s extra fields were discarded under a documented schema amendment before outcomes.
- Groups >=50 and detailed issues >=100 shown separately. Percentages derive from unweighted counts; no causal inference or sampling intervals.
- The 78.5% platform-style convention is reconstructed as resolved plus unassessed over all rows; consumer-reported resolved among assessed is 55.9%.
What the data cannot tell us
- The category mixes books with stationery, teaching materials and office supplies; no product or format subtype.
- Selective companies and platform users; no purchase denominator, stable complaint ID, unique-person count or merchant adjustment.
- 51.3% unassessed. Feedback is self-reported; a reply is not independent proof of a remedy and no reply is not automatic failure.
- Closure-year cohort, not purchase or filing year. Coding errors, indistinguishable rows, publication repetition and later revisions remain possible.
Original sources
- Consumidor.gov.br / Senacon: official public finalized-month data
Twelve actual primary archives, January–December finalized 2025, acquired through the declared public route; hashes and URLs are downloadable.
Accessed October 31, 2026
- Consumidor.gov.br: current official FAQ
Current official FAQ explicitly defines the solution convention. It is not equivalent to positive consumer feedback.
Accessed October 31, 2026
- Consumidor.gov.br: current indicator methodology
Current methodology separates reply and consumer assessment, and selects finalized dates.
Accessed October 31, 2026
- Senacon: Boletim Consumidor.gov.br 2025
The official annual bulletin supplies the same 2,328,863 all-subject rows and broad-sector prior art. The limited contribution here is an exact-category issue-by-response-by-feedback profile.
Accessed October 31, 2026
Cite & reuse this study
Alisson Bento. 2026. Brazil’s books-and-stationery complaints: beyond the reply. Consumidor.gov.br / Senacon, 2025. https://alissonbento.com/research/brazil-books-stationery-complaint-followthrough-2025/
Source data and definitions: Consumidor.gov.br / Senacon, Ministry of Justice and Public Security. Original aggregate analysis, figures, code and explanation: Alisson Bento. Reuse original materials with attribution and a link, retaining mixed scope and limitations. No source rows or government endorsement are supplied.
Corrections or editorial questions