The Permanent Record / Dossier · Disputes & Correction
The Two-Digit Dispute
You write a letter. You attach the cancelled check, the settlement agreement, the letter from the collections agency saying the account was closed. You mail it to the credit bureau, and you assume that what you sent is what the creditor will read.
It is not. The dispute travels through a private network called e-OSCAR, built in 1993 and owned by the credit bureaus themselves. What arrives at the furnisher is an electronic form carrying one or two numeric codes drawn from a list of twenty-nine, plus, in a minority of cases, up to 255 characters of free text. The documents stay behind. In 2011 the bureaus added any free-form text at all to 26 percent of those transmissions, and the system did not accept attachments.
Join the discussion →
The Fair Credit Reporting Act gives you a right that sounds absolute. Dispute an item, and the credit bureau must conduct a reasonable reinvestigation, generally within thirty days. It has five business days to forward the dispute to whoever furnished the information, and the notice it sends must include, in the statute's words, all relevant information regarding the dispute that the agency has received from the consumer. If the reinvestigation finds the data inaccurate, incomplete, or unverifiable, the bureau must delete it.
None of that describes a process. It describes an outcome the law requires, and the process that grew up to produce it was designed by the industry being regulated. Understanding what actually happens to a dispute means looking at that machinery rather than at the statute, because the two are not the same shape.
A Network Owned by the Defendants
Trade line disputes move through e-OSCAR, the Online Solution for Complete and Accurate Reporting. It began in 1993 as a system run by the Associated Credit Bureaus, now the Consumer Data Industry Association. Four companies built it and own it: Equifax, Experian, Innovis, and TransUnion. The internet-based version dates to 2001, and in 2006 the CDIA created a separate entity, the Online Data Exchange, to operate it. By 2011, sixteen thousand furnishers were connected to the bureaus through it.
The volume it carries is not small. In 2011 the three nationwide bureaus received roughly eight million consumer contacts disputing the completeness or accuracy of something in a file. Because a single contact often disputes more than one item, those contacts generated somewhere between thirty-two and thirty-eight million individual reinvestigations. That was a low year. In 2007, when mortgage activity was at its peak, the bureaus received disputes on forty-seven to fifty-three million items. Consumers filed 42 percent of those disputes online, 44 percent by mail, and 13 percent by phone.
Everything that follows is a story about compression. Eight million human accounts of what went wrong have to be turned into something a furnisher's clerk can act on in seconds.
Twenty-Nine Codes
When a dispute is filed online, the consumer must select one or two reason codes from a list of twenty-nine. The list is not secret and it is not long. It includes "Not his/hers." It includes "Belongs to another individual with same/similar name." It includes "Claims paid the original creditor before collection status or paid before charge-off," "Not liable for account (i.e. ex-spouse, business)," "Disputes dates of last payment/date opened/date of first delinquency/date closed," and, at the end, the catch-all: "Consumer states inaccurate information." For disputes filed by mail or phone, the consumer never sees the list at all; a bureau representative reads what was sent and assigns the codes they deem appropriate.
The bureau then runs its own internal screen before anything is forwarded. It uses proprietary decision rules to see whether it can resolve the dispute without involving the furnisher — rejecting resubmitted disputes as frivolous, accepting documentation it can reasonably verify as authentic, correcting identifying information itself. In 2011 the bureaus resolved or rejected an average of 15 percent of the disputes they received this way, and the CFPB reported that it did not know what share of those resolutions favored the consumer. One bureau separately reported that roughly 16 percent of disputes produce no e-OSCAR transaction at all, because the consumer had submitted an identical dispute recently and the furnisher had already verified the data.
What survives that screen is packaged into an Automated Consumer Dispute Verification — an ACDV — and sent to the furnisher. The ACDV carries the codes. Supplementing them, it can carry up to 255 characters of consumer-supplied free text. In 2011 the bureaus added free-form text to 26 percent of their e-OSCAR transmissions on average, a figure that varied by bureau depending in part on whether their online dispute form even contained a text box.
The cancelled check does not travel. Consumers can mail supporting documents to a bureau, but the CFPB found that the bureaus generally do not pass them to furnishers, and that e-OSCAR did not permit consumer documents to be forwarded as attachments. Industry sources named technological limits, the difficulty of judging whether a consumer's document is authentic, and privacy concerns. The bureaus' position was that most disputes can be fairly and completely summarized using the numeric codes.
The statute says the bureau must forward all relevant information it received from the consumer. The system it built to do that has room for two numbers and a text field the size of a short paragraph.
A decade later the regulator addressed the gap directly. In Consumer Financial Protection Circular 2022-07, published November 10, 2022, the CFPB wrote that while no provision affirmatively requires a bureau to forward original paper copies, it would be difficult for a consumer reporting agency to prove they provided all relevant information if they fail to forward even an electronic image of documents that constitute a primary source of evidence. The same circular told bureaus and furnishers they may not condition an investigation on a consumer using a specific format or supplying a specific attachment.
What "Modified" Means
The furnisher's side of the transaction is narrower still. A representative reviews the company's own electronic record of the account and picks one of four responses: verify the account as accurate, modify the trade line as indicated, delete the account, or delete the account due to fraud. There is no fifth option for "the consumer's document contradicts our file," because the consumer's document is not in front of them.
The outcome distribution looks, at first glance, favorable. Over a 120-day period in 2012 the CDIA reported that 22 percent of furnisher responses verified the original data as accurate, 61 percent modified a trade line or other item, 13 percent deleted one, and 0.5 percent deleted one for fraud. A further 4 percent of disputed trade lines were deleted or modified because the furnisher missed the statutory deadline. Sixty-one percent modified sounds like a system that mostly agrees with the person complaining.
The CFPB attached the qualification in the same paragraph. Many larger furnishers automatically refresh a trade line with the latest account information on receiving a dispute — a new balance, a new status — regardless of whether they think anything was wrong. A modification, the Bureau wrote, may not necessarily reflect concurrence with the consumer's dispute. The response code that dominates the statistics is the one that is hardest to read.
The historical series makes the drift visible. In the first three quarters of 2002, according to GAO testimony, 46 percent of responses verified the data as reported and 27 percent modified it. By the first five months of 2004 that had flipped to 22 percent verified and 54 percent modified. By 2011 it was 22 and 61. Over the same period, deletions caused by a furnisher simply not answering within thirty days fell from 16 percent to 4 percent. The furnishers got faster and the modifications got more common, and neither number tells you whether anyone's file got more accurate.
The Records That Skip the Network Entirely
Public records take a different path. A dispute about a judgment or a lien is not investigated through e-OSCAR at all. The bureaus instead re-collect the record from the government source, or contract that work to LexisNexis Risk Data Retrieval Services, which was pulling one to two million public records a year at the bureaus' request. The bureaus send LexisNexis one of up to twenty-four reason codes. In 60 to 70 percent of those verification requests the consumer's assertion is that the record is not theirs; in another 20 to 25 percent, that the judgment or lien was paid.
LexisNexis sends a data collector to re-check the record and reports back one of three statuses: changed, unchanged, or unable to verify. It says it locates the record in 99.5 percent of queries. But as the CFPB set out plainly, LexisNexis does not verify the content of the underlying public record as accurate, and does not determine whether the bureau correctly linked the record to the consumer in the first place. The single most common complaint — this is not mine — is the one question the verification step is not designed to answer.
Who Disputes, and What Happens Next
In October 2021 the CFPB published an analysis by Ryan Sandler drawn from its Consumer Credit Panel, a longitudinal sample of roughly five million de-identified credit records from one nationwide bureau. It tracked the compliance condition code — the field a furnisher uses to mark that an account is in dispute — on accounts opened between 2012 and 2019.
Disputes are rare and unevenly distributed. Of accounts opened in that window, 0.75 percent of auto loans, 0.81 percent of student loans, 2.1 percent of general purpose credit cards, and 0.5 percent of retail cards ever carried a dispute flag. Collections items, by contrast, were disputed at nearly 5 percent. Consumers with dispute flags were younger and had substantially lower scores than those without, and the geography was stark: 2.8 percent of auto loans belonging to residents of majority-Black census tracts carried a dispute flag, against 0.8 percent in majority-White tracts. For credit cards the split was 4.7 against 2.5. Holding age, score and tract median income constant narrowed the gap but did not close it.
Two other findings from that data are worth holding onto. The first is timing: accounts that were eventually disputed had already been sitting on the report for a long time — an average of 856 days for auto loans and 1,026 days for student loans before the flag appeared. The second is duration. The statute gives the bureau thirty days. In practice, the CFPB found, an initial dispute flag frequently remains for significantly longer than thirty days.
The Complaint Channel as Evidence
There is a second window into how the process performs, because Section 611(e) of the FCRA requires the CFPB to report annually on the complaints it transmits to the three bureaus. Between October 2021 and September 2022 it sent them more than 488,000, of which about 65 percent were the covered kind — incorrect information, or a problem with a prior investigation, where the consumer had already tried to fix it directly.
These are people who have already been through the machine. More than 93 percent of consumers complaining about a nationwide bureau reported having previously attempted to fix the problem with the company, against 86 percent for furnishers and under 85 percent for other reporting agencies. And in 2021, the Bureau found, the rate at which the three companies reported providing any relief in response to those complaints fell to less than 2 percent, down from nearly 25 percent in 2019 — the result, it concluded, of the bureaus closing complaints faster on unsubstantiated conclusions of third-party involvement and referring others into the dispute channel without reporting what happened. Equifax attributed its reduced complaint-handling staffing to increased automation. In August 2022, nearly 64 percent of complaints were still taking thirty days or more.
What the process produces, when it works, is also measurable. The FTC's Section 319 study, published in December 2012, walked 1,001 participants through their own reports. Twenty-six percent identified at least one potentially material error, and 262 filed disputes covering 572 of the 2,968 reports examined. The bureaus modified something on 399 of those reports, and 206 consumers — 21 percent of the sample — got a modification of some kind. But only 97 of the 262 disputants, 37 percent, had every one of their disputes addressed. Over half of the people who got a modification still had information on a report that they said was wrong.
And Then You're Dead
The dispute right in the statute is a right to have someone look again. The system built to deliver it converts a person's account of their own life into two numbers off a fixed list, forwards them to the company that made the entry, and accepts that company's answer. The check you photocopied never leaves the building. The most common thing a consumer says — this account is not mine — is routed either to the furnisher who already believes it is, or to a vendor whose job explicitly excludes deciding whether the record was attached to the right person. Thirty days later the file is whatever the furnisher said it was, and the only durable trace of the argument is a compliance condition code that tends to outlast its own deadline.
That's it. That's the whole thing.
Sources
- The e-OSCAR network's 1993 origin under the Associated Credit Bureaus, its ownership by Equifax, Experian, Innovis and TransUnion, the 2001 internet-based system and the CDIA's creation of the Online Data Exchange in 2006, the 16,000 connected furnishers in 2011; the five-step dispute process; the list of 29 dispute reason codes and the requirement that an online disputer select one or two; the ACDV form and its 255-character free-text field, added to 26% of 2011 transmissions; the finding that bureaus generally do not forward consumer documents and that e-OSCAR did not permit attachments; the four furnisher response types; the CDIA's 2012 120-day outcome distribution (22% verified, 61% modified, 13% deleted, 0.5% fraud-deleted, 4% deleted or modified for non-response) and the 2004 and 2002 comparison figures; the caution that large furnishers auto-refresh trade lines on receipt so a modification may not reflect agreement; the ~8 million 2011 dispute contacts, 32–38 million reinvestigations, 47–53 million disputed items in 2007, and the 42%/44%/13% online/mail/phone split; the 15% of disputes resolved or rejected internally and one bureau's 16% duplicate-suppression figure; the Q4 2011 dispute-type mix (33% "not mine," 15% inaccurate information, 4% balance, 4% unknown collections); and the public-record path via LexisNexis Risk Data Retrieval Services, its 1–2 million annual record pulls, up to 24 reason codes, the 60–70% "not his/hers" and 20–25% "already paid" split, the 99.5% location rate, and the explicit statement that LexisNexis does not verify the underlying record's content or whether the bureau linked it to the right consumer — Consumer Financial Protection Bureau, "Key Dimensions and Processes in the U.S. Credit Reporting System: A review of how the nation's largest credit bureaus manage consumer data," December 2012
- The Consumer Credit Panel dispute-flag analysis: the ~5 million de-identified record sample; dispute-flag rates for accounts opened 2012–2019 (0.75% auto loans, 0.81% student loans, 2.1% general purpose credit cards, 0.5% retail cards, and almost 5% of collections items); the census-tract table (auto 2.8% majority-Black vs 0.8% majority-White; student 2.4% vs 0.9%; credit cards 4.7% vs 2.5%; retail 1.4% vs 0.6%) and the note that controlling for age, score and tract median income shrinks but does not eliminate the gap; the average days reported before a dispute flag (856 auto, 1,026 student, 767 credit card, 741 retail); the pre-dispute score and delinquency figures; the finding that disputers skew younger and lower-scoring; and the observation that an initial dispute flag frequently remains far longer than the statutory 30 days — Ryan Sandler, Consumer Financial Protection Bureau, "Consumer Credit Trends: Disputes on Consumer Credit Reports," October 2021
- The more than 488,000 complaints transmitted to the three nationwide bureaus between October 2021 and September 2022 and the ~65% that were covered complaints; the finding that reported relief fell to less than 2% of complaints in 2021 from nearly 25% in 2019 and the reasons the CFPB gave for it; the 93%/86%/85% prior-contact comparison across bureaus, furnishers and other reporting agencies; Equifax's attribution of reduced complaint-handling staffing to increased automation; and the finding that nearly 64% of complaints took 30 days or more as of August 2022 — Consumer Financial Protection Bureau, "Annual report of credit and consumer reporting complaints," January 2023
- The study of 1,001 participants in which 26% identified at least one potentially material error, 262 filed disputes covering 572 of 2,968 reports, 399 reports were modified, 206 consumers (21%) received some modification, 97 disputants (37%) had all disputes addressed, and over half of consumers who received a modification still had information on a report they alleged was inaccurate; also the 64.3% modification rate for disputed header information — Federal Trade Commission, "Report to Congress Under Section 319 of the Fair and Accurate Credit Transactions Act of 2003," December 2012
- The statement that a consumer reporting agency would find it difficult to prove it provided all relevant information if it fails to forward even an electronic image of documents constituting a primary source of evidence, and the prohibition on conditioning an investigation on a specific format or attachment — Consumer Financial Protection Bureau, Consumer Financial Protection Circular 2022-07, "Reasonable investigation of consumer reporting disputes," November 10, 2022
- Photo: "Kansallisarkisto Helsinki Rauhankatu 17 luettelohuone kortistokaappi", the catalogue room of the National Archives of Finland, photographed February 23, 2026 by Fuje23, CC BY-SA 4.0, via Wikimedia Commons
Comments
Loading…