Key findings
- The last complaint narrative in the public archive is dated 11 August 2026. None of the 1,004,609 complaints received from 15 August to 2 October 2026 has one.
- This was the end of a long decline, not a cliff from a healthy base. The share of complaints with a narrative went from 47.5% in 2017 to 22.5% in 2025 and 1.5% in July 2026.
- A second field also stopped: "Company public response" was empty in about 37% of complaints through July 2026, then 62% in August, 87% in September and 99.5% in October.
- No other field we checked changed. Product, issue, company, state, tags, company response to consumer and timeliness stayed as before.
- Counts, product mix, relief rates, state rates and tag comparisons are unaffected. Anything built on narrative text (templates, real cases) is frozen at mid-August 2026.
- We found no documentation explaining either change and do not guess a cause.
Our work on the CFPB database rests on two sources: the main complaint file and the archive of complaint narratives. In 2026 the CFPB stopped publishing new narratives. This investigation measures exactly what changed in the data, month by month, and what that means for anyone analysing it.
What the Data Shows
- The last narrative is dated 11 August 2026. The archive holds about 3.85 million narratives in total. From 15 August to 2 October 2026 the database added 1,004,609 complaints, none with a narrative (938,600 of them credit reporting).
- The share had been falling for years. Complaints with a narrative: 32.5% (2015), 47.5% (2017, the peak), 42.1% (2022), 29.8% (2024), 22.5% (2025), 2.7% for 2026 up to 14 August. By month: 36.3% in January 2025, 11.0% in December 2025, 4.1% in January 2026, 1.5% in July 2026 (10,001 of 669,795), 0 in September.
Share of complaints with a published narrative, by month, 2025–2026. Source: CFPB Narratives Archive; Kibbo Research analysis.
- Credit reporting is part of the explanation, not all of it. Credit reporting makes up about 88% of recent complaints, and in January to July 2026 only 0.13% of its complaints had a narrative. Other families still had much higher shares in those months: debt collection 20.0%, student loans 31.5%, mortgage 38.1%, bank accounts 40.4%, money transfers 42.9%. We did not compute a month-by-month series excluding credit reporting.
- What the stop costs, as an estimate. If each family had kept its January to July 2026 narrative share, the 15 August to 2 October complaints would have carried about 21,300 narratives. Of those, about 5,800 would be debt collection, 4,100 credit and prepaid cards, 4,100 bank accounts. This is an estimate under one assumption, not a count.
- "Company public response" is the second field to stop. The share of complaints with no company public response was between 36.6% and 45.3% from January 2025 to July 2026, then 61.9% (August), 86.8% (September) and 99.5% for the 15,987 complaints received in the first days of October. It declined gradually, not on a single day.
Share of complaints with no company public response, by month, 2025–2026. Source: CFPB Consumer Complaint Database; Kibbo Research analysis.
- Nothing else moved. We compared the empty share of every column month by month. Only "Company public response" jumped by more than 10 points. Product, sub-product, issue, company, state, ZIP code, tags, how the complaint was submitted, date sent to company, company response to consumer and timely response all stayed flat (tags are empty in about 98% of complaints in both July and September 2026).
- What still works. Complaint counts, product mix, relief rates, state rates and per-capita comparisons, and tag-group comparisons do not use narratives and can still be updated. Template detection, the real cases, and any "originals only" cut are frozen at the cutoff.
| Page | Depends on narratives? |
|---|
| Main investigation (templates) | Yes |
| Real cases (9 pages) | Yes: the text comes from a published narrative |
| Debt collection credit-threats note | Partly: narrative share and template split; issue counts do not |
| Monetary relief investigation | Only the "originals only" cut |
| Older Americans and Servicemembers | Only the template share and narrative presence |
| State rates note | No |
| How to write a CFPB complaint (guide) | Yes: cites the template figures |
What We Cannot Say
- Why either field stopped. We found no documentation. We do not speculate.
- Whether it is permanent. The data only run to the start of October 2026.
- That narratives were distributed evenly before. Most narratives were already concentrated outside credit reporting.
- How many narratives were lost. The 21,300 figure is an estimate under a stated assumption.
Method and Limits
- Sources: CFPB Consumer Complaint Database and the narrative archive published through the CFPB's FOIA Electronic Reading Room. Our window ends on 2 October 2026; the database runs to 3 October (1,409 further complaints, mostly credit reporting, not in these figures).
- Narrative share = complaints with a narrative in the archive ÷ complaints received that month. The main database has no narrative column and no "Consumer consent provided?" field in either source, so we cannot check consent.
- Null shares are computed for every main-database column by month, January 2025 to 2 October 2026; flagged jumps are more than 10 percentage points between consecutive months or between the three months before and after 14 August 2026.
- Ten complaint IDs in the narrative archive have no match in the main database file, so 3,851,415 narratives are in the archive and 3,851,405 match a complaint in the main database. We quote the total as "about 3.85 million" to avoid false precision.
- 2024 and 2025 monthly totals reconcile exactly with the official yearly totals.
- Logged as Step 13 in our research log. No company data used.
Related
How to Cite
Kibbo Research (2026). What Stopped Being Published in the CFPB Complaint Database in 2026. getkibbo.com. https://www.getkibbo.com/investigations/cfpb-narratives-stopped-2026
