Corrections & Changes

The host served both /about and /about.html with a 200, so every page was being discovered and crawled twice. The canonical tag meant the right version was indexed, but the duplicate still consumed a crawl and appeared in Search Console as an alternate page. Extensionless paths now redirect to their .html version, generated from the site itself so new pages are covered automatically. This was previously left alone because a redirect could have caused a loop if the host redirected the other way; crawl data confirmed both forms returned 200, so no loop is possible, and the generator checks for one before writing.

Twenty-nine tool pages were missing the WebApplication entry that the other seventeen carried, so search engines had no machine-readable statement that those pages are free interactive tools. All forty-six now declare it, with the page URL, description, publisher and a zero price, and every JSON-LD block on the site validates.

Twenty-five pages gained a section naming the specific errors that arise with that particular tool, bringing the site to 46 of 46. Each names a checkable mistake rather than offering general caution: that eleven stone four is 158 pounds and not 11.4 stone; that an Imperial fluid ounce is smaller than a US one even though the Imperial gallon is larger; that a refinance breaking even in fifteen months can still cost thousands more because the term reset; that sales tax in most US states is charged after the trade-in allowance rather than before it; that a 13 mm spanner on a half-inch bolt leaves enough play to round the head. No paragraph is repeated across any two pages.

Amortization, mortgage, loan, auto loan, extra payment, debt payoff, compound interest, investment, retirement and rent vs. buy now export their full schedule as a spreadsheet file — 360 rows for a thirty-year mortgage, every payment split into interest, principal, PMI and remaining balance, with a header block recording the inputs the file was built from. Rows are generated from the calculation engine at the moment you click, not scraped from the table on screen, so the file carries full precision rather than rounded display strings. The file is assembled in your own browser and never uploaded, which keeps intact the property that figures you enter never leave your device.

All 190 inputs across the site now carry an explanation, up from a handful on the mortgage page. Each one says what the field is and, where there is one, the mistake people make with it — that two stacked discounts of 20% and 10% come to 28% rather than 30%, that 11 stone 4 is not 11.4 stone, that resetting a part-paid loan to a fresh full term lowers the payment while raising the total. Help is delivered as a real button with aria-expanded and a linked panel rather than a title attribute, because title text cannot be reached by keyboard, does not appear on touch devices, and is announced unreliably by screen readers.

Only six of the forty-six tools had any visual; forty-three do now. Chart type follows what the tool computes rather than being applied uniformly: loans get a falling balance, growth tools stack contributions against the growth they generate, tax tools break income into each deduction and the remainder, and each converter shows how big one unit is against the other. The three number-to-words converters were deliberately left alone, because a chart there would be decoration rather than information. Charts are inline SVG with no library, so they add about six kilobytes in total, print correctly, and carry an aria-label describing what they show rather than being invisible to a screen reader.

An audit found eight checkboxes with no accessible name — a screen reader announced them as "checkbox, unchecked" with nothing else — and three pages that skipped a heading level. Both are fixed. The mortgage calculator was then rebuilt: down payment and property tax can be entered as a percentage or a dollar amount, extra payments and annual cost increases are modelled, PMI now stops automatically once the balance reaches 78% of the purchase price, payment numbers are shown as calendar dates with a payoff month, monthly and lifetime costs appear side by side, and a biweekly comparison shows what twenty-six half-payments a year would save. Inline help was added to every significant input, print buttons to forty-seven pages, and a full calculator strip to the foot of every tool page.

All forty-six tools can be placed on any website with a single line of HTML, free and without registration. Each embed is generated by stripping the real page rather than rebuilding it, so an embed can never drift from the tool it mirrors. Advertising, analytics and structured data are removed, and the embeds are marked noindex and kept out of the sitemap so they cannot compete with the pages they came from. A discovery worth recording: the site sends X-Frame-Options SAMEORIGIN and frame-ancestors 'self' on every path, which would have blocked third-party framing entirely — that is now relaxed only for the embed directory, so every other page keeps its clickjacking protection.

Breadcrumb structured data had been present on 74 pages but was invisible to visitors; the trail is now shown on the page, generated from the same data as the schema so the two cannot disagree. Sixty-four pages that previously declared a flat trail are now correctly nested under their category hub. Search was only on the home page and is now in the header of every page, as a plain form that works without JavaScript. A contact page replaces the bare email link, covering corrections, calculator requests, and what the address cannot help with. A new A–Z index lists all forty-six tools alphabetically with a one-line description of each, generated from the pages themselves so it cannot fall out of date.

The remaining thin articles were roughly doubled and each gained a frequently-asked-questions section. New material covers the bracket myth (a raise never reduces take-home pay through tax brackets — a single filer keeps $82,930 at $95,000, $86,830 at $100,000 and $90,730 at $105,000, so the real risk is benefit cliffs rather than brackets), the minimum payment trap ($6,000 at 22.99% takes 167 months and $14,033.26 in interest at a $120 minimum, against 46 months and $3,011.78 at a fixed $200), compounding fee drag (a one percent annual fee costs $107,727.98 over thirty years on $500 a month at 7% — nearly eighteen percent of the final balance), closing costs and PMI removal thresholds, escrow as the usual reason a fixed-rate mortgage payment changes, and why APR misleads on a loan repaid early. Every figure was computed and checked before publication.

Forty-six tools presented as cards had become a long scroll. The homepage now lists them as category columns of plain links, so a named tool can be found in one glance, with each category heading linking to its own hub page. A new Everyday Math hub completes that set. The search box now filters the list and hides any category with no matches.

Weight, length, volume, area, speed, temperature, energy, pressure and data storage. Conversion logic lives in a new units.js rather than finance.js, so a fault in one cannot take down the other. Most factors are exact by definition rather than measured — 1 inch is 2.54 cm and 1 pound is 0.45359237 kg under the 1959 International Yard and Pound Agreement — and every one was verified against its definition and round-tripped. The pages that need it carry a US or Imperial toggle: the two gallons differ by 20% and, counterintuitively, the two fluid ounces differ by 4% in the opposite direction. Cups to grams asks which ingredient you mean, because volume and mass have no fixed relationship.

Sales tax (both directions), inflation, interest rate, amortization schedule, investment, and retirement. The interest rate tool solves the amortization formula backwards by bisection, since it cannot be rearranged for the rate. The amortization schedule computes interest the way a servicer does — rounded to cents each month with the final payment trimmed — which differs from the formula-based total by about $4.71 over a 30-year loan, so both figures are reported rather than hiding the discrepancy. The retirement tool draws savings down against spending that rises with inflation and names the age the money runs out. The homepage is now grouped into labelled categories with a dedicated financial calculators hub.

The site menu had grown to twenty-six flat links that wrapped to roughly eight lines on a phone before any content appeared. Links are now grouped into six labelled categories and collapsed behind a menu button on small screens; every link remains in the page at all widths. Calculator results are now announced to screen readers when they change, having previously updated silently. A print stylesheet strips navigation, ads and footers so an amortization schedule or tax breakdown prints as a clean document. Animations now respect a reduced-motion preference. Five of the thinnest pages — the tip, percentage and loan comparison calculators, the currency converter, and the about page — gained substantive material on the questions people actually ask of them.

Three changes prompted by pages Google had discovered but not indexed. The seven state comparison guides shared one paragraph verbatim; it has been replaced on each page with the specific non-income-tax factors that cut against that pair — New York City’s own local income tax, the additional local income taxes paid in the Portland area, Tennessee’s sales tax, New Hampshire’s property tax. The two number-word converters were thin enough that Google never spent crawl budget on them, and now carry the hyphenation and “and” rules, cheque-writing conventions, and the rule that written words govern over figures on a negotiable instrument. A stale SearchAction block that made Google crawl a placeholder URL was removed, and a missing https-to-non-www redirect was added.

Compares renting and buying on net worth over a chosen period. Because the answer depends entirely on assumptions nobody can verify in advance, all sixteen inputs are on screen, the headline figure is the home appreciation rate at which the two come out level, and a sensitivity grid shows where the verdict changes. The comparison is symmetric: whichever option costs less per month, that side invests the difference.

The extra-payment tool shows how much time and interest a larger payment or lump sum removes from a loan, and quantifies how much of that saving is down to timing alone. The refinance tool reports the break-even month alongside the lifetime cost, and warns explicitly when a lower monthly payment would cost more overall because the term was reset — a case most refinance calculators leave unstated.

Arkansas, Delaware, Massachusetts, Missouri, Rhode Island, South Carolina, Virginia, West Virginia and the District of Columbia apply one bracket schedule to every filing status. This was verified rather than assumed, and the calculator now says so directly instead of carrying a filing-status caveat it did not need.

Delaware House Bill 13 would add 6.75% and 6.95% brackets for tax years beginning after 31 December 2025, and at least one secondary source treats it as being in force for 2026. Legislative tracking shows it introduced in January 2025 and still pending in the House Revenue & Finance Committee. It is not law, so it is not modelled.

In the eleven states above, head-of-household thresholds are not separately published in the source used here, so single-filer thresholds stand in. The calculator now discloses this in its own result whenever head of household is selected, rather than presenting the number as exact.

The site referenced a favicon, an Apple touch icon and an Open Graph share image that did not exist, so shared links rendered without a preview. All are now present, generated from the site’s own colour and type tokens.

Fourteen calculators now encode their inputs into the page URL, so a scenario can be bookmarked or sent to someone else. Values arriving in a URL are validated against each field before being applied, and anything unrecognised is ignored.

A reference page of US and Canadian tax, contribution and document-arrival deadlines, with weekend and holiday shifts applied and any date whose sources disagree flagged as unsettled rather than guessed.

Side-by-side take-home pay comparisons between states, computed with the same code the salary calculator uses. Two further pairs were researched and deliberately not published because the difference between the two states fell inside the margin of error of the model’s own simplifications.

A side-by-side loan comparison tool, five worked walkthrough guides, and balance-over-time charts on the loan, auto loan, mortgage, debt payoff and 401(k) calculators.