Every tool on this site produces a number that somebody may act on. This page explains where those numbers come from, how they are verified, how often they change, and — just as important — where they stop being reliable.
There is a fair question behind every free calculator on the internet: who worked this out, and should I believe it? Most sites never answer it. This page is our answer, written in enough detail that you can check our reasoning rather than take it on faith.
A calculator is a claim. When a mortgage tool returns a monthly payment of $2,061.25, it is asserting that a particular formula was applied to particular inputs under particular assumptions. If any of those three is wrong, the output is wrong in a way that looks completely authoritative — clean typography, two decimal places, no visible doubt.
That is the risk we take seriously. A wrong figure on a converter costs somebody a mild inconvenience. A wrong figure on a mortgage, tax or medication-dosage tool can cost real money or worse. So the standards below are not uniform across the site: the arithmetic behind a retirement projection is checked more aggressively than the arithmetic behind a pizza-size comparison, and we say so openly rather than implying that everything here carries the same weight.
Roughly a third of our tool pages cite an outside authority directly, and the citation appears at the foot of the formula section rather than buried in a footer. We use two different words there, and the difference is deliberate.
“Source” means the linked body actually defines the rule the page explains. When our BMI page cites the CDC, it is because the weight categories shown are the CDC’s categories. When the required-minimum-distribution page cites the IRS, it is because the life-expectancy divisor comes from the IRS Uniform Lifetime Table. When the paper-size converter cites ISO 216, it is because the √2 aspect ratio of the A series is defined in that standard.
“Reference” means something weaker and we did not want to blur the two. On our state-by-state mortgage and paycheck pages, the property-tax rates and median home values come from our own compiled dataset. Linking a federal agency next to those figures would imply the agency published them, which it did not. So those pages say “Reference” and point to the authority that governs the subject — the body you would consult to go deeper — without claiming it as the origin of our number.
The authorities we lean on most are the ones you would expect for United States figures: the Internal Revenue Service for tax treatment and contribution limits, the Social Security Administration for payroll tax rates and the annual wage base, Freddie Mac’s Primary Mortgage Market Survey for benchmark mortgage rates, the Consumer Financial Protection Bureau for lending rules such as when private mortgage insurance can be cancelled, the Bureau of Labor Statistics for the Consumer Price Index, the Federal Reserve for interest-rate series and household wealth data, and the CDC and NIH for health reference ranges. For measurement we use NIST and the BIPM, whose definitions underpin every unit conversion on the site. For document and text tools we cite the actual specifications — ISO 32000 for PDF, ECMA-376 for Office Open XML, the Unicode Standard for character encoding, and the relevant IETF RFCs for data formats.
Every one of those links was fetched and confirmed to return a working page before it was published, and the page title was checked against the claim being made. That sounds obvious. It is not: during this work we found a candidate government URL that redirected to an entirely unrelated article, which would have shipped as a broken citation if we had trusted the status code alone. We also dropped one citation entirely — the international standard governing Morse code timing — because we could not reach the publisher’s site to confirm the address. The page names the standard in the text and links nothing, which we consider the honest outcome. A citation you have not verified is worse than no citation, because it borrows authority it has not earned.
Every worked example on this site is computed, not typed. That distinction matters more than it sounds, because plausible-looking arithmetic written by hand is wrong far more often than anyone expects, and it is wrong in ways that survive proofreading — a digit transposed, a rate applied monthly instead of annually, a total that nobody re-adds.
Some real examples of what that catches. Our mortgage page originally said the crossover point, where a payment starts putting more into principal than interest, arrives “around year 18”. Recomputing the full amortisation schedule put it at month 233 — year 20. On an auto-loan comparison we described the gap between a 6% and a 12% rate on $30,000 over five years as about $85 a month and $5,100 in total; the true figures are $87.35 and $5,240.96. A future-value growth factor was published as 5.73947 when the correct value is 5.72542.
Two corrections went further than a digit and changed the conclusion of the paragraph they sat in. On a credit-card payoff example, we had assumed a 2% minimum payment covers the interest; recomputing showed that at typical card rates it does not, so the balance grows while the borrower pays on time every month. On a car-financing comparison, we had picked the wrong option outright — cash back beats low-interest financing, but only below roughly 36 months, and the ranking flips after that. Neither error would have been caught by reading the text carefully. Both were caught by making a computer do the sum.
Most sites type a date into a template and forget it. Ours is derived: the date shown on a page is taken from when the code that produces that page last changed. It moves when the page moves and it cannot silently rot.
This is a deliberate correction of something we got wrong ourselves. For a long stretch the site displayed a hardcoded “Last updated June 2026” on hundreds of pages regardless of what had happened to them, and twenty pages were still advertising a date from the previous January. Worse in its own way, a set of data pages emitted a machine-readable modification date that always resolved to the current day, so every page claimed to have been edited the moment you loaded it. That is not a freshness signal, it is noise dressed as one, and search engines are entitled to treat it as such. Both are now fixed, and the date a reader sees matches the date in the structured data on every page.
What the date does not mean is that every underlying figure was re-checked that month. Tax thresholds, contribution limits and benchmark rates change on their own schedule, set by the bodies that publish them, not by us. Where a figure is time-sensitive we name its year in the text so you can see at a glance whether it has been overtaken.
Most of our traffic is American and most of our tax and payroll logic is built for the United States. Plenty of visitors are not, and we would rather be straightforward about what travels well and what does not than let a Canadian, British, Irish, Australian, European or Singaporean reader assume a US model applies to them.
Unit conversion is universal by construction. The metre, kilogram, second, ampere, kelvin, mole and candela are defined identically everywhere, and every factor-based converter on this site resolves through those definitions, so a length, mass, pressure, energy, viscosity or concentration conversion is as valid in Dublin or Singapore as in Denver. The same is true of the mathematics and geometry tools, the date and time calculators, and the general statistical functions.
Most health formulas are also jurisdiction-neutral, because they describe bodies rather than laws. Body mass index, basal metabolic rate through the Mifflin-St Jeor equation, body-fat estimation, pace and macronutrient splits do not change at a border. What can change is the reference range a clinician uses to interpret the result, and the units you are used to reading it in — British readers often think in stones for body weight, and much of the world uses millimoles per litre for blood glucose where the United States uses milligrams per decilitre. Our converters handle the units; local clinical guidance governs the interpretation.
The core financial mathematics travels too. Compound interest, present and future value, annuity accumulation and loan amortisation are arithmetic, not policy. A payment calculated from a principal, a periodic rate and a number of periods is correct in any currency, which is why our amortisation and compound-interest tools are safe to use anywhere provided you supply the right rate and know how your lender compounds it.
Anything touching tax, payroll or government benefits is US-specific and should not be used elsewhere. That includes federal and state income tax, take-home pay and withholding, Social Security and Medicare contributions, 401(k) and IRA limits, required minimum distributions, capital-gains treatment, and sales tax. The underlying structures have no equivalent abroad: the United Kingdom operates PAYE with National Insurance, Ireland adds PRSI and the Universal Social Charge, Canada runs CPP and EI alongside federal and provincial tax, Australia uses PAYG with a legislated superannuation guarantee, Singapore has the Central Provident Fund, and most of Europe applies value-added tax where the United States applies sales tax at the point of sale. A US paycheck calculator does not approximate any of these; it models a different system.
Mortgages deserve their own warning, because they look transferable and are not. In Canada, fixed-rate mortgage interest is compounded semi-annually rather than monthly, which means a US calculator will return a payment that is close enough to look right and wrong enough to matter over a full term. British mortgages are typically fixed for an initial period and then revert to a variable rate, so a single-rate amortisation covering the whole term describes a product that is rarely sold. Australian loans commonly pair with offset accounts that change the effective interest calculation entirely. Across the European Union and Ireland, lenders quote an APRC under the Mortgage Credit Directive, which is defined differently from the US APR and is not directly comparable. Property purchase costs diverge even further — stamp duty in the UK and Ireland, land transfer taxes by province in Canada, additional buyer’s stamp duty in Singapore — none of which appear in our US closing-cost model. Our loan-specific tools such as FHA, VA and USDA calculators describe United States government programmes and have no counterpart abroad at all.
A few things trip up international readers regardless of which tool they use. Dates are written month-first in the United States and day-first almost everywhere else, so 03/04 is ambiguous unless you know the convention. Decimal separators differ — a comma in much of continental Europe where English-speaking countries use a point — and our inputs expect the point. Fuel economy runs in miles per gallon in the United States, litres per 100 kilometres across Europe and Canada, and the two are reciprocal rather than proportional, so they cannot be averaged the same way. Even the gallon is not one unit: the US liquid gallon is about 3.785 litres against the imperial gallon’s 4.546, a difference of roughly 20% that quietly wrecks any comparison made without checking. Pay frequency varies too, with fortnightly cycles common in Australia and Ireland where monthly dominates in Britain and biweekly in the United States.
Financial, tax and health tools on this site produce educational estimates. They are not financial, tax, legal or medical advice, they do not know your circumstances, and they are not a substitute for a professional who does. Pages in those categories now carry that notice directly rather than hiding it in the terms of use.
We also mark our own novelty tools as novelty tools. The love, zodiac and Chinese zodiac calculators, the countdown and the random-number generator are entertainment, and they say so where a source would otherwise appear. They previously carried the same “peer-reviewed formulas and official government data” line as our tax tools, which was simply untrue, and correcting it mattered more to us than the tidiness of a uniform footer. Two related tools sit in between and are labelled accordingly: the pizza comparison is real geometry rather than research, and the dog and cat age calculators use the epigenetic ageing curve published by Wang and colleagues in Cell Systems in 2020 — genuine peer-reviewed work, and a considerably better model than multiplying by seven.
The PDF, image, barcode and text utilities run entirely inside your browser. Files are read locally through standard browser APIs, processed in your device’s memory, and never uploaded to us or to anyone else. There is no server-side copy because there is no upload, which is the point: a contract, a passport scan or a signed form should not leave the machine it was opened on to be merged, split or compressed.
This design has honest limits. Everything happens in the memory available to a browser tab, so very large files can exhaust it — a single twelve-megapixel photograph occupies roughly 34 MB once decoded — and heavy work such as optical character recognition is slower than a paid desktop application backed by a server. We consider that an acceptable trade for not handling your documents at all.
We would rather be told. If a figure here looks wrong, or a source no longer supports the claim beside it, please get in touch and describe what you were calculating and what you expected. Corrections are made to the underlying tool rather than patched into the text, so the fix propagates to every page that uses the same logic, and the page’s update date moves with it automatically.
Errors found in our own review are corrected the same way and are not quietly overwritten; several are described by name on this page precisely because a methodology statement that claims a perfect record is not worth reading.
Figures come from named official sources, and the difference between a source and a reference is stated rather than blurred. Links are verified before publication and dropped when they cannot be. Worked examples are computed rather than written. Update dates are derived from the code that produces each page, and match the machine-readable date search engines read. Tools that are US-specific are identified as such, and the ones that are entertainment say so. None of this makes a free calculator infallible. It does mean that when one of ours is wrong, you can find out why.