Every number on this site comes from a named source with a date, or from
arithmetic you can repeat. This page says where each one comes from, how it is
checked before it is published, and what we do not do. It exists because we ask you to
trust figures about money and courts, and that is not something to ask on faith.
Where the numbers come from
Six sources, all public and all free to check. Nothing on this site rests on a private
dataset, a vendor estimate or a figure recalled from somewhere else.
| What | Source | What we pulled | What we derive from it | Exact reference |
|---|---|---|---|---|
| Consumer complaints | Consumer Financial Protection Bureau, public complaint database | 324,326 debt collection complaints, 6,200,165 credit reporting and 92,805 credit card, received 1 September 2025 to 1 September 2026 | Share by issue, by outcome and by company; complaint rates per 100,000 residents; a 2019–2026 annual series | documented search API |
| Interest rates | Board of Governors of the Federal Reserve System, via FRED | Credit card plans, accounts assessed interest: 22.15%. Personal loans at commercial banks, 24 months: 11.86%. Both 2026-05 | The gap between carrying a balance and refinancing it, month by month; delinquency and charge-off rates by bank size | series TERMCBCCINTNS |
| Payoff arithmetic | Our own calculation, at the Federal Reserve rate above | Months and interest for balances from $2,000 to $30,000 against eleven monthly payment levels, three minimum-payment formulas, four consolidation terms and 29 snowball-versus-avalanche scenarios | Every payoff figure on this site, to the cent, with its assumptions printed beside it | monthly compounding at APR/12; assumptions listed in each article |
| Bankruptcy filings | Administrative Office of the United States Courts, Table F-2 | 608,511 filings in the year to 30 June 2026, across 93 judicial districts, plus the same table for March 2026 and June 2025 | Chapter 13 as a share of consumer filings by district (36.9% nationally, 82.1% to 8.1%); filings per 100,000 residents by state; change on a year earlier | the official XLSX |
| Means test thresholds | Executive Office for United States Trustees, U.S. Department of Justice | Median family income for 55 jurisdictions, for cases filed on or after 15 July 2026 | How many jurisdictions a given income clears; the spread from $30,665 to $88,585 for a one-person household | the means testing index |
| Population | U.S. Census Bureau, Vintage 2024 population estimates | State resident population, used only as a denominator | Everything expressed “per 100,000 residents” | file NST-EST2024-ALLDATA, column POPESTIMATE2024 |
figures it uses, because these sources update on their own schedules.
The distinction that matters: the counts belong to the agencies. The rates,
shares, rankings and per-capita figures are ours — the Consumer Financial
Protection Bureau publishes how many complaints came from Georgia, not how many per
resident; the courts publish filings per district, not what share of them are Chapter 13.
Those derived numbers are the part of this site that does not exist anywhere else, and
they are the part we are most careful with.
Where a rate is assumed rather than measured
Two card rates appear on this site and they are not the same kind of claim, so it is worth separating them here rather than leaving you to notice.
22.15% is a measurement. It is the Federal Reserve’s average rate on credit card accounts assessed interest, the May 2026 observation of series TERMCBCCINTNS, retrieved 2 September 2026. Every figure in a data section on this site is worked at it.
24.99% is an assumption. It is a round, commonly quoted card rate used to work examples in the prose, chosen because it reads cleanly — not observed anywhere and not sourced to anyone. Where a guide uses it, the guide now says so at the point it first appears.
The distinction has consequences, which is why it gets its own heading. At the assumed rate, $500 a month clears $10,000 in 27 months; at the measured rate it clears in 26 and costs $474 less. Neither figure is wrong and neither is a correction of the other. They are not interchangeable, and no number worked at one rate should be set beside a number worked at the other.
How a figure gets published
Four steps, in this order, and a figure that fails any of them does not go out.
- It is fetched, not remembered. Each dataset is built by a script that
records the exact query or file it read and the moment it read it. - It is checked against the source’s own total. When an agency prints a
total, we sum our rows and compare. All three bankruptcy tables reconcile exactly to the
printed Total line; the complaint aggregations sum to the filtered total to within the
0.1% of complaints the bureau records without a state. - It is declared before it is written. Every number in an article’s
data section is tied to a path into the dataset that produced it. An automated check
resolves each one and refuses the article if a single figure in the prose is not
accounted for. That check is why you will not find a rounded-off number here with no
provenance. - Its limits are printed next to it. Each data section ends with what
the figure does not say. A number without its boundary is a promise, and we are
not in a position to make promises.
Three times this process caught us being wrong
The checks above are not theoretical. Here is what they caught while this site’s data
was being built, because a method page that only lists virtues is not worth reading.
- A mislabelled interest rate series. We had a Federal Reserve series
tagged as the credit card rate. It is the rate on 24-month personal loans. The mistake
surfaced because a derived figure came out with the sign reversed, and every series title
on this site is now verified one at a time against its own page. - A renamed category that looked like a trend. The complaints bureau
renamed its credit reporting product in 2023. Querying only the current name returns
zeros for the earlier years — zeros that look exactly like data. Read that way,
Mississippi appears to go from no complaints at all to 166,668
in three years. Our annual series sums both the current and the retired label at every
point; on that basis debt collection complaints went from
46,368 in 2019 to 283,828 in 2025,
which is 6.12 times and is a real change. - A superseded government table. The first version of our means test
figures used the table effective 1 April 2026. A later one, effective 15 July 2026, was
already in force. We compared the two and they carry identical medians — but we
found that out by checking, not by assuming, and the site cites the one in force.
There is also a case where checking confirmed a published figure rather than
overturning it. Our comparison of the debt snowball against the debt avalanche reported a
$998 saving without saying what monthly payment produced it. Rebuilt from scratch, the
figure is right — it corresponds to an extra $275 a month — and the article now
prints that assumption along with the full range, because the same choice is worth
$134 at one payment level and about $1,020 at another.
What we do not do
- We do not backdate anything. Publication dates are the dates things
were published. Where an article has been revised, the revision date changes and the
original does not. - We do not fill in state law tables from memory. Limitation periods,
bankruptcy exemptions and homestead amounts vary by state and several states have changed
theirs. Our articles explain the mechanism and send you to the statute. A stale table
would be worse than no table. - We do not publish a figure we could not verify. Where a number is
needed and no reliable source has it, we say so and give you the route to check. - We do not accept payment for coverage or placement, and no article
here recommends a named company for a fee. How advertising works on this site is on our
advertiser disclosure page.
What this site is not
This is information, not advice. PayoffPath explains how debt, credit
and bankruptcy work in the United States. It does not give individual financial, legal or
tax advice, and reading it creates no professional relationship. What is right for you
depends on your income, your state and the terms of your accounts.
And one gap we are not going to paper over. Content on money and
bankruptcy should carry a named professional review — a practising bankruptcy
attorney for the bankruptcy section, an accredited financial counselor or certified
financial planner for the rest. No such review has taken place yet. The
articles are researched and sourced, their arithmetic is reproducible, and their factual
claims were audited; none of that is the same thing as a credentialed reviewer signing
their name to it, and we are not going to describe it as if it were. Where an article
touches a decision with legal or tax consequences, it says to get advice specific to your
state, and it means it.
Corrections and reproduction
If a figure here is wrong, we want to know, and a correction gets a dated note rather
than a silent edit. Everything on this page can be reproduced from the references in the
table above with no special access. Reach us through the
contact page, and see our
editorial policy for how articles are written and
updated.