The change log
Every market change we observe, with its date and source, and every material correction to our own work, stating what we said before and what the source shows. Append-only: entries are added, never edited away.
The uncomfortable half
90 of the 118 entries are corrections to our own records. The rest are 12 first records, 6 unsuccessful research attempts and 10 genuine market changes. That ratio is the point of the page: a research site with no visible corrections is not one that has made none.
What is recorded here
Two things in one list, on purpose. Market movement: a provider moving a rate, withdrawing a product, rewriting a criterion. And changes in us: a figure we had wrong, a document we had not opened, a claim we withdrew. A separate, quieter page for our corrections would make the market log look cleaner and tell you less.
The four kinds of entry
| Kind | Count | What it means |
|---|---|---|
| Market change | 10 | A provider or official source genuinely changed something after our previous observation. A rate moving, a product withdrawn, a criterion rewritten. |
| Correction | 90 | A record or interpretation of ours was wrong. What we had, what the source actually shows, and what changed. |
| First record | 12 | The first time we captured a state, so later movement has something to compare against. Our first capture, not the date the provider launched anything. |
| Attempt | 6 | A check we ran that produced no usable answer, logged with the obstacle rather than dropped. |
Entries by week
What earns a correction entry, and why most predate launch
A correction, by our definition, is a factual, source, regulatory, data or methodological claim we recorded or published that was wrong or materially unsupported. Copy-editing, clearer phrasing, layout changes and additions that do not reverse an earlier claim are not logged. Without that line the promise would be unkeepable.
Most early corrections predate publication. The site went live on 20 August 2026. Corrections logged before then are to records and draft pages no reader had relied on, logged the same way because the habit has to exist before it is tested.
The log
The 21 entries from the 30 days to . The 97 earlier entries are folded below, unchanged.
-
Gatehouse Bank. Four Gatehouse Bank records and statements of ours were wrong. Our finance-to-value records gave British expats and international residents 80% with no finance limit, and applied the 75% (to £3m) and 70% (to £5m) tiers to UK residents only; Gatehouse's personal home finance page and its up-to-80% criteria guide put the 80% tier at up to £2m of finance and the 75% and 70% tiers on all three customer types, so the eligibility checker could pass an expat at 80% on more than £2m. Our profile said the energy-efficient product is exactly 0.10 points lower than the standard one in every matched pair; that holds for the 16 pairs open to a purchase, while in the 4 pairs of its refinance-only range the two are priced the same, which our matching had masked. The profile dated its rate grid to the 5 September 2026 card, when most of the grid comes from the 6 August 2026 card. Two fact notes cited its above-80% criteria guide as HPPCG120826V1; the guide we read is HPPCG120826V2, and the wording is the same. Our terms hub also left Gatehouse out of diminishing musharaka and ijara, because its register entry gives only its own name for the structure, Acquisition and Rent; its Sharia certificate names a Diminishing Musharakah Agreement and a lease.
Source -
Six statements in our tools and guides had stopped matching our own records. The eligibility checker, the calculator and the rates page said the step between deposit tiers is larger than the gap between providers inside one; since Offa repriced on 28 September 2026 no tier step we hold is larger than the provider gap, so the claim is gone. The checker gave the minimum deposit as 5 to 20% depending on provider and property type; Gatehouse Bank caps new-build flats at 75% finance-to-value, so it is 5 to 25%. The calculator said it priced 64 products from 3 UK providers, but Nomo's products are for GCC residents, whom its residency options do not cover; it prices 56 from 2 and names Nomo as excluded. The deposit optimiser said it priced 29 products from 3 providers when it only ever uses UK-resident products, 16 from 2. Our comparison of Islamic and conventional mortgages, and our guide to how they work, said we had not verified portability for the providers we track, when StrideUp's intermediary criteria say its products are not portable and Offa's criteria guide lists portability among its exclusions. And the comparison said a home purchase plan has nothing but the provider's ownership interest behind it, when both published Sharia certificates we hold list a legal charge among the plan documents.
Source -
Six statements of ours described our own process as more than it is. Our methodology said pages mark a figure past its review window as due for review, and that software flags such figures; nothing on the site does that yet, so it now says the date of the last check sits beside every figure instead. It said the site does not publish if any build check fails, when one check, on published findings, takes a finding that stops holding off its page rather than stopping the build; it named 14 checks when 26 run; and it said a figure we could not verify stays out, when a published figure a later check cannot read stays up with the date we last read it. Our ask an adviser page said you pay nothing if you take advice as a result of asking, which we cannot promise: any adviser's fee is set and disclosed by that firm. Our Wayhome entry called its Deposit Bridge protection question one "we have raised", when it has not yet been put to Wayhome. Each now says what we do.
Source -
Our privacy notice's flow box said the figures typed into the tools, the term among them, are held in the browser tab's session storage. The calculator keeps the price, deposit, purchase or refinance, customer type and property details for the eligibility checker, and never the term. The box now names what is kept. Found on a read-through of the notice against the code on 9 October 2026. Nothing was ever sent; the box overstated what the tab holds.
Source -
StrideUp. StrideUp no longer shows the rates-page banner that read "Home finance rates updated" and quoted a 5.99% headline figure, which we last read on 5 September 2026. It was absent on 22 September and again on 9 October 2026, when the rates page and the homepage were read in full, as served and in a browser. The rates page heading now reads "Explore current rates for StrideUp’s halal home finance products", with no headline figure, and its calculator shows a dash where a rate would be. We never treat a missing page as a withdrawal, but these pages were not missing: they loaded complete. The record stays in our data as no longer listed and is no longer shown. It named no product and never entered our comparisons.
Source -
Nomo by BLME. Nomo no longer lists its high-net-worth ranges. Its intermediaries products page, read in full on 9 October 2026, has no "Nomo HNW up to £2m" or "Nomo HNW £2m to £5m" products; we last read them on 22 September 2026 at two-year 5.49% and five-year 5.79% (up to £2m) and two-year 5.46% and five-year 5.79% (£2m to £5m). In their place the page lists a "Nomo Enhanced Range" for £750,000 to £5,000,000 at 75% finance-to-value, two-year fixed 5.79% and five-year fixed 6.09%, with the application fee "Agreed at application (typically 1%)". We never treat a missing page as a withdrawal, but this page was not missing: it loaded complete, twice. The four records stay in our data as no longer listed and are no longer shown or priced. We have not added the Enhanced Range, because the page does not say whether it is for purchase, refinance or both.
Source -
Our privacy notice said that when someone finishes the eligibility checker, the only thing recorded is which route they took next. The checker sends no analytics event at all, and no analytics runs on this site, so nothing is recorded. The sentence now says what the analytics wrapper would accept and that the checker does not send even that. Found on a read-through of the notice against the code on 9 October 2026. No reader data was affected; the notice overstated what we record.
Source -
Gatehouse Bank. Two generated sentences had gone stale. Our process and first-time buyer guides listed Gatehouse Bank's non-refundable application fee as "£149 and £0", because the refinance-only products, which publish no application fee, were read as charging £0; the fee is £149. And our home purchase plan guide said follow-on rates for two mainstream providers were not published, which has been true only of StrideUp since we first recorded Offa's 7.25% follow-on rate on 5 September 2026.
Source -
Our guide to riba said we hold published Sharia certificates for the three mainstream home purchase plans, and that a named certifier had assessed each against a published standard. We hold two published certificates, Gatehouse Bank's and Offa's; StrideUp states its certification but we have not located the document. Neither certificate names a standard it was assessed against. The guide now says what the certificates do say.
Source -
StrideUp. Three StrideUp statements of ours went further than its pages. Our visa guide put two paraphrases of StrideUp's residency wording inside quotation marks; they now quote the intermediary criteria and help-centre article word for word. Our StrideUp profile said StrideUp "advertises" a "Now from 5.99%" banner, which we last read on 5 September 2026 and did not find on 22 September or 9 October; the profile now dates it. And our process and first-time buyer guides said StrideUp takes its product fee at the application stage, as its first-time buyer page says, when its tariff list of 1 July 2026 says the product fee is paid on accepting the offer; both guides now show the disagreement.
Source -
Offa. Our remortgage guide said we had not established what Offa allows a refinance to raise money for, and had not verified its position on porting. Offa's home purchase plan criteria guide, which we cite, publishes both: capital raising is acceptable, excluding tax and gambling, and portability is listed among its exclusions. The guide now says so.
Source -
Gatehouse Bank. Our visa guide and remortgage guide said Gatehouse Bank's two home purchase plan criteria guides were both effective 5 August 2026. Since 5 September our sources have been the replacements: the guide for finance up to 80% finance-to-value is effective 4 September 2026 and the guide above 80% is effective 12 August 2026. Every criterion the two pages take from them reads the same; only the dates were wrong.
Source -
Gatehouse Bank. Our remortgage guide said no UK-resident range publishes separate refinance pricing, in its heading, its answer and its FAQ. That stopped being true on 4 September 2026, when Gatehouse Bank launched limited-edition refinance-only products for UK residents with no product or application fee, which our own records have held since 5 September. The page went on printing it, and its counter-example sentence named Gatehouse among providers "open to GCC residents rather than UK ones". The guide now says Gatehouse publishes separate refinance products, at the same rate as the matching purchase product or up to 0.10 points higher, with lower fees, and that refinance is the same product at the same price everywhere else.
Source -
The Bank of England's quoted mortgage rate series, read 9 October 2026, now give slightly lower figures for four observations we hold: April 2026 two-year fixed at 95% loan-to-value 5.80% (we held 5.81%) and five-year fixed at 95% 5.61% (5.63%); July 2026 two-year at 95% 5.48% (5.49%) and five-year at 95% 5.28% (5.29%). The two-year 90% figures for April (5.46%) and July (5.07%) are unchanged. The Bank's methodology says mortgage series are reweighted once a quarter's new-business balances are available. We have also recorded August and September 2026: two-year at 90% 5.16% and 5.37%, two-year at 95% 5.52% and 5.70%, five-year at 95% 5.36% and 5.54%.
Source -
Nomo by BLME. Our two Nomo high-net-worth records for £2,000,000 to £5,000,000 carried a note about a finance-to-value conflict in their early settlement field, so Nomo's profile listed that note as their early settlement terms. Nomo publishes one early repayment rule for its property finance: repaying in full inside the fixed period means paying the profit for the rest of it. The field now says so, as on every other Nomo record. Their rates were not re-read and are unchanged.
Source -
Gatehouse Bank. Our records gave Gatehouse Bank's four five-year refinance products (GHRU1772, GHRU1773, HRU1776 and HRU1777) the two-year early redemption charge: 2% in year 1 and 1% in year 2. Gatehouse publishes 3% in years 1 and 2, 2% in years 3 and 4 and 1% in year 5 for them, the same as its other five-year products, on both its intermediary and personal product pages. The error came from building the eight refinance records on 5 September from a two-year product and not replacing the charge on the five-year ones. A reader repaying early in years 1 to 5 was shown a smaller charge than applies, or none at all in years 3 to 5. Corrected; the rates themselves were right.
Source -
Nomo by BLME. Nomo has repriced its intermediary products page (read 9 October 2026). Refinance, £100,000 to £749,999: two-year 5.75% (was 5.46%), five-year 6.05% (was 5.80%); £750,000 to £5,000,000: two-year 5.69% (was 5.25%), five-year 5.99% (was 5.79%). Nomo Standard, £100,000 to £749,999: two-year 5.89% (was 5.51%) with APRC 6.9% (was 7.0%), five-year 6.19% (was 5.89%) with APRC 6.7% (was 6.8%). The fee on these products is now £1,999, labelled an application fee (was £1,499). The two high-net-worth ranges we hold, £750,000 to £1,999,999 and £2,000,000 to £5,000,000, no longer appear; the page lists a Nomo Enhanced Range for £750,000 to £5,000,000 instead, at 5.79% two-year and 6.09% five-year with the fee agreed at application. We have not added it, because the page does not say whether it is for purchase, refinance or both, and we keep the four high-net-worth records at their 22 September figures as not re-read rather than treat their absence as a withdrawal. Every range now states 75% maximum finance-to-value, so the 70% heading we recorded as a dispute is no longer published. Nomo's August 2026 residential factsheet, linked from the same page, still carries the old rates.
Source -
Offa. First recorded: Offa's six British expat home purchase plans, priced separately from its UK-resident range since the rate card updated 28 September 2026 (card and products page read 9 October 2026). Discounted variable 6.99% at 65% finance-to-value and 7.19% at 80%; two-year fixed 6.99% and 7.19%; five-year fixed 6.89% and 7.09%. Each carries a £499 product fee, the £149 application fee, the 7.25% follow-on rate and the same early settlement charges as the UK-resident range, for purchase or refinance, with finance up to £1.5m.
Source -
Offa. Offa has repriced every home purchase plan. Its rate card updated 28 September 2026 and its products page, both read 9 October 2026, give UK residents: discounted variable 6.44% at 65% finance-to-value (was 5.50%) and 6.55% at 80% (was 5.60%); two-year fixed 6.59% at 65% (was 5.80%), 6.65% at 80% (was 5.80%), 7.35% at 90% (was 6.60%) and 7.45% at 95% (was 6.90%); five-year fixed 6.65% at 65% (was 5.75%), 6.75% at 80% (was 5.88%), 7.45% at 90% (was 6.58%) and 7.55% at 95% (was 6.88%). British expats, who paid the UK-resident rate at 65% and 80%, are now priced separately and higher, so our six shared records now cover UK residents only. Product fees, the £149 application fee, the 7.25% follow-on rate, early settlement charges and lending limits are unchanged, which puts the follow-on rate below the initial rate on the 90% and 95% products. The April rate card is still online but no longer linked from Offa's rate cards page.
Source -
Gatehouse Bank. Our record let new-build flats reach 90% finance-to-value at Gatehouse Bank. Gatehouse caps new-build flats at 75%: its personal home finance page says so, its up-to-80% criteria guide says so, and its above-80% guide says they are not accepted above 80%, which our own first record of that guide (CL-088) already stated. The error came from the 90% and 80% tiers carrying no property condition, so the calculator, deposit optimiser and eligibility checker read them as open to new-build flats, and our visa guide and Gatehouse profile said the cap was 90%. A reader with a 10% or 20% deposit on a new-build flat could have been shown Gatehouse products they cannot take. The tiers are now scoped by property type and the 75% cap is recorded as its own fact. New-build houses are also corrected upward to the 95% purchase tier Gatehouse publishes for them.
Source -
Gatehouse Bank. Gatehouse Bank reissued its home purchase plan criteria guide for finance above 80% finance-to-value as version 2, HPPCG120826V2, still effective from 12 August 2026. The version 1 address now returns 404 and the guide sits behind a permanent download link. Every criterion we take from it reads the same in version 2. Its August 2025 international broker brochure also returns 404 and is no longer linked.
Source
Earlier entries: 97, 6 August 2026 to 6 September 2026
-
After the 5 September re-verification pass, several pages carried freshness claims that were wrong or inconsistent. The home page and the calculator still said rates were checked on 6 August 2026 while the rates and comparison pages said 5 September, because each page holds its own checked-date constant and two were not moved. The comparison page said we had checked all 13 propositions on 5 September when 11 were re-read and 2 (UBL UK and ADIB) stand at 6 August. The rates page printed "Checked 5 September 2026" on every row, including the 14 Gatehouse expatriate and international rows that could not be re-read that day; those rows now show their last verified date and say they were not re-read. The change log's own header said no market change had been recorded while listing four. Caught by an external review on 6 September 2026.
Source -
Our privacy notice still described the site as it was before the enquiry form went live on 20 August 2026. It said the parts that collect anything were not switched on, that there was no form, that we do not collect email addresses anywhere, that we hold no personal data, and that we collect nothing to transfer outside the UK. Each of those was false from 20 August: the enquiry form collects a name, email address, optional phone number and question, which reach our mailbox through Web3Forms and Google Workspace, both outside the UK. The sentences now read from the same configuration switch as the form itself, so they cannot drift again. Caught by an external review on 6 September 2026. No reader data was mishandled; the notice understated what we collect.
Source -
QIB (UK) plc. The QIB (UK) privacy policy page we cite returned 404 on 5 September 2026. The private banking page is live and unchanged: Murabaha home finance for London property, a £1,000,000 qualifying relationship balance, pricing not published.
Source -
Offa. First recorded from Offa's HPP rate card (updated 14 April 2026): a follow-on rate of 7.25% on every product, an application fee of £149 on every product, early-settlement charges of 2% then 1% on two-year fixes, 5%, 4%, 3%, 2% on five-year fixes and none on the discounted variable products, finance £60,000 to £1.5m per property, minimum property value £80,000, maximum 95% finance-to-value up to £750,000 and 80% up to £1.5m, expats to 80%. All ten Offa rates we hold match the card and the products page to the decimal.
Source -
Gatehouse Bank. We could not re-verify Gatehouse Bank's eight UK Expat and eight International Resident products on 5 September 2026. Both product pages return only the UK-resident section to our reader, cutting off after two expat rows, and the International Broker Brochure V5 (22 August 2025) is a criteria document with no product codes. Those 14 rate records stand as verified on 6 August 2026 and are marked as not re-retrievable today.
Source -
StrideUp. StrideUp's /rates/ page now redirects to /view-all-rates/, and its /faq/ and /how-strideup-home-finance-works/ pages redirect to the homepage, so that content is no longer published at those addresses. The rates page still renders its numeric rates as 0% placeholders to automated readers, so we continue to publish no StrideUp initial rate; the £1,249 product fee is confirmed. Every eligibility figure we hold on both channels is unchanged, including the consumer page's 18 and £75,000 against the intermediary criteria's 21 and £85,000, and the intermediary page still asks for six months of business bank statements in one place and three in another.
Source -
Wayhome Deposit Bridge. Two of the three Wayhome pages we cite now return 404: the Deposit Bridge eligibility page and the Sharia compliance page. Wayhome's statement on Sharia compliance now lives in a Help Center article dated 28 May 2025, which names only the retired Gradual Homeownership product as certified, by Amanah Advisors, from a 2023 document, and does not mention Deposit Bridge. The Deposit Bridge FAQ says a buyer may use Shariah-compliant finance "provided your chosen provider is comfortable with the Deposit Bridge structure". We have re-pointed both sources and left the Deposit Bridge Sharia question open.
Source -
Nomo by BLME. Nomo's intermediaries products page now publishes a £1,499 product fee on its four refinance products (2 and 5 year fixed, up to £749,999 and £750,000 to £5m), which we had recorded as not stated. All ten Nomo rates are unchanged. The page still states 70% maximum finance-to-value in the £2m to £5m section heading while its product rows state 75%, and still says England, Scotland and Wales while the criteria page says England and Wales only.
Source -
Kuwait Finance House PLC. Our single Kuwait Finance House arrangement-fee fact (0.75% of the finance amount) was cited to the Commodity Murabaha (Tawarruq) home finance page, which states 1.0%. The 0.75% figure belongs to the Ijarah page. The fact is now cited to the Ijarah page and a second fact records 1.0% for Murabaha; both pages are dated 7 April 2026. All twelve KFH rates we hold are unchanged.
Source -
Gatehouse Bank. First recorded from Gatehouse Bank's two new intermediary criteria guides, HPPCG040926V1 (effective 4 September 2026, finance up to 80% finance-to-value) and HPPCG120826V1 (effective 12 August 2026, finance above 80% finance-to-value): maximum age 70 at application and 75 at end of term; minimum property value £94,000 at 80% finance-to-value and £115,000 at 65% (£79,000 at 95% purchase-only, £84,000 at 90%); leasehold at least 70 years remaining at application and 40 at end of term; minimum main-applicant income £15,000 (UK residents up to 80% finance-to-value), £25,000 (UK expats and international residents) and £20,000 above 80%; new-build flats not accepted above 80% finance-to-value; income multiples up to 6.00x at or below 80% and 5.50x, 5.00x or 4.49x above it depending on income; England and Wales only. Every criterion we already held, including minimum age 18 for UK residents and 21 for expats and international residents, matches both guides.
Source -
Gatehouse Bank. Gatehouse Bank has added eight UK-resident refinance products with no product fee and no application fee: GHRU1770 to GHRU1773 (Green, EPC A or B) and HRU1774 to HRU1777, fixed for 2 or 5 years at 65% or 80% finance-to-value, rental rates 5.63% to 5.76%, £75,000 to £5m. The intermediaries page lists them without a label; the personal home-finance page groups them under "Refinance Products". We record them as refinance-only until Gatehouse states otherwise. The 16 UK-resident products we already held are unchanged to the decimal, and the SVR is still 7.25%.
Source -
Our homepage hero said every Islamic mortgage UK buyers can currently apply for is in our register and that we check each one, a completeness claim our methodology explicitly refuses to make. The hero now describes the propositions we track rather than the market entire. Two unmeasured claims went with it: that eligibility stops most applications, which we cannot know without application data we do not hold, and that other comparison sites never admit what they could not find out, which we have not measured either.
-
Gatehouse Bank. Our homepage showed Gatehouse’s deposit as "from 5%, up to 95% finance-to-value" without the conditions that make 5% possible: purchase only, finance to £600,000. The same condition had already been corrected on the rates page, the comparison, the profile and the register; the homepage was the stale copy of the fix. The qualifier now travels with the figure.
-
Our homepage ranked the mainstream providers by each one’s lowest published rate of any shape, setting a 65% discounted variable against a five-year fixed green product and calling the result an order. On the site’s own like-for-like benchmark, a UK resident buying with a 20% deposit on a two-year fix, the order reverses, so the homepage and the comparison page showed opposite rankings generated from the same dataset. The homepage now orders by the same benchmark as the comparison, names each provider’s lowest-anywhere figure as a secondary line with its shape stated, and scopes its hero figure to what it is: the lowest mainstream UK-resident rate of any product shape, not the register-wide minimum, which belongs to a Gulf-resident refinance product.
-
Offa. A review stated that Offa publishes no follow-on rate, which would have made the 7.25% follow-on on our Offa rate rows an import contamination, since Gatehouse’s standard variable rate is also 7.25%. We retrieved Offa’s rate card PDF on 11 August 2026. It publishes a "Follow on Rate" column of 7.25% against every product, so our record is correct and the identical figure across the two providers is a genuine coincidence. No change made; recorded because the check that confirms a suspicious-looking record is worth as much as one that overturns it.
-
Our homepage cost strip priced products its reader could not have. Its product pool had no customer-scope filter, so at plausible high-value inputs a Gulf-resident-only product was shown as the lowest figure to a UK visitor. Worse, at the default inputs the strip’s first paint showed a green product’s rate with its EPC condition silently dropped: the defect corrected on four other surfaces, live in the first figure a first-time visitor sees. The pool is now scoped to products open to UK residents and the winning product’s conditions travel into the result line. The hero specimen, which promises the conditions that earn its figure, now consults the purpose and EPC conditions too, so the promise cannot break silently if a conditional product ever takes the top slot.
-
Our calculator presented its output as "lowest monthly payment" and headed its column "monthly", while the methodology six screens below admitted the figure is standard amortisation applied uniformly, not any provider’s contractual schedule. The label now carries the semantics at the number: modelled monthly payment, modelled payments plus provider fees. The arithmetic itself was checked independently and is correct; what was wrong was the name.
-
Nomo by BLME. Our calculator described everything it prices as a Home Purchase Plan, stating "the regulated term for what you are pricing is a Home Purchase Plan". Its own priceable universe includes Nomo’s commodity murabaha products, which repay a deferred sale price rather than combining rent with share acquisition. The same page-universe error we corrected on the rates page, restated on the tool. The copy now scopes the structural explanation to the structures actually present.
-
Our calculator said three times that nothing a reader types is stored. Our own code stores the entered scenario in the browser’s session storage on every input, deliberately, so the eligibility checker can carry it over; a bug we fixed days ago existed only because of that stored handoff. Not sending figures to us and not storing them anywhere are different claims, and we were making the wrong one. Every instance now distinguishes the two, describes the session-only store and its purpose, and points at the privacy page. A homepage card describing the calculator carried the same false claim and is corrected with it.
-
Our register stated that two specialist propositions publish lower headline figures than any mainstream product. One does. Kuwait Finance House’s lowest figure equals the lowest mainstream figure rather than beating it. The sentence was also a rate-ranking claim on the one page whose method is not to rank, and it has been replaced with the actual method: specialist figures can look competitive but sit behind materially different criteria, so they are listed and not ranked.
-
Our register FAQ described all thirteen entries as Sharia-compliant home finance propositions. The register’s own taxonomy says otherwise: one alternative model is not presented by its own provider as Islamic home finance, and four entries are names where our finding was that no current route exists. The answer now describes propositions and provider routes relevant to the market, which is what the register actually contains.
-
Gatehouse Bank. Our provider register summarised Gatehouse’s ceiling as "up to 95% UK resident". Our own records scope that ceiling to a UK-resident house purchase with finance to £600,000, with refinance topping out at 90%. The register card now carries the scope with the maximum, matching the correction already made on the rates page, the comparison page and the profile.
-
The build check that should have caught the register defect contained a regular expression that had never matched anything. A shell-to-script escaping step had turned a word-boundary marker into a literal backspace character, so the pattern for "from X%" price claims was unmatchable from the day it was written, and every rate introduced with the word "from" escaped the condition checks. The character is invisible in an editor, which is how it survived review. Repaired, and verified by watching the check fail on the register before the card was fixed.
-
Nomo by BLME. Our provider register showed Nomo’s lowest figure as "From 5.25%, 2 years fixed, 75% finance-to-value, Gulf residents" without stating that the product is refinance-only. A buyer was shown a price not available for a purchase at all. The register card carried every other condition and dropped this one; it now renders transaction purpose with the rest.
-
StrideUp. Two stale statements survived the StrideUp profile rebuild. A footnote said "minimum age 18 for all applicants" ten lines above a table saying 18 or 21, sources differ: the same raw-value access the eligibility table had been cured of, surviving in a summary line. And an open question still said we could not retrieve the visa-specific residency article, on a page that quotes that article two sections up; it was retrieved and registered during our visa research. The footnote now renders through the display module, and the question is narrowed to what the article genuinely does not publish: which occupations receive the 12-month treatment.
-
StrideUp. While repointing the StrideUp residence record we cited a source id that was not registered, and the build passed because the field was outside the citation check’s list. The same class of slip as the Offa tariff citation earlier the same day, caught this time by re-checking rather than by the guard. The citation now points at the four registered documents, and the check covers every criterion field the profiles render.
-
The three failures above shared one cause: the profile template read raw fact values through a single accessor that took the first record and its bare value, dropping scope, dispute and absence. Criteria now render through a display module with the same contract as the finance-to-value display layer: pages receive finished text, disputes always show both figures, and no page touches a raw value. The templated sentence claiming each provider "publishes more applicant criteria than most providers we track", which appeared on all three profiles and was measured against nothing, has been replaced with a provider-specific description.
-
StrideUp. Our StrideUp profile correctly displayed the 85-or-90 finance-to-value and 18-or-21 age disagreements in its research table, then flattened both a screen later: the deposit section said "10% deposit, all applicants" and the applicant table said "minimum age 18". Both now render both figures with the disagreement named. Its residence row also stated a settled two years while our own visa research had recorded a banded position, 12 or 24 months depending on occupation; the fact record has caught up with the research and the row renders the bands.
-
Offa. Our Offa profile rendered "the plan must finish before every applicant’s nullth birthday". The record correctly holds that Offa states a maximum age exists without publishing it, and the template interpolated the recorded absence into prose. It also rendered a £1,000,000 maximum finance as universal when the record scopes it to new-build purchases, contradicting the £1.5m range shown at the top of the same page. A null now renders as the absence it records, and scoped figures keep their scope.
-
Gatehouse Bank. Our Gatehouse profile stated a universal minimum age of 18 in its applicant table while the paragraph above it correctly said 18 for UK residents and 21 for expat and international applicants. The table read one record and dropped its customer scope. It now renders every scoped record with its scope. The same table’s 5% deposit row also omitted that the 95% tier is purchase-only, so a remortgager was shown a tier not open to them; the transaction condition now renders on the tier row and in the deposit answer.
-
Our structured-data check could never match a question containing an apostrophe. It replaced every HTML entity with a space, so a rendered "Gatehouse Bank's" normalised to "Gatehouse Bank s" and could not match the same words in the page’s JSON-LD. It reported two pages as making unbacked structured-data claims when the questions were rendered on them. Entities are now decoded rather than blanked.
-
Our records held that two figures for a criterion disagree, but not which document published which figure, and the sort order cannot stand in for it: StrideUp’s consumer page publishes the lower figure for minimum age and minimum property value and the higher figure for finance-to-value. Any page inferring the channel from position would have printed one of the three backwards. The attribution is now recorded rather than derived.
-
StrideUp. Our provider profile described StrideUp’s channel disagreements from a hand-written table in the page rather than from the records. Every figure existed twice with nothing keeping the copies in step, which is how our comparison page came to describe one of the same disputes as "older help material" while the records held it as a live conflict. The table is now derived from the canonical facts.
-
Gatehouse Bank. Our Gatehouse profile and our homepage both stated a lowest published rate without saying the product requires an EPC rating of A or B. The profile said "the lowest published UK resident rate being 5.53%", which is a green five-year product; the reader’s figure without that rating is 5.63%. This is the third page where the same defect appeared, and the third time our build check missed it because it had been taught the phrasings used on the previous two. A superlative next to a rate is a price claim in whatever words, so the check now looks for the claim rather than the sentence.
-
Gatehouse Bank. Our comparison page described Gatehouse’s finance-to-value ceilings as a count followed by a list of bare percentages, which included 80% twice because two separately scoped rules share that figure. Stripped of their conditions the numbers told a reader nothing and looked like a data error. It now states that no single universal maximum is published and gives the range with the dimensions the ceiling varies on.
-
Our comparison page rendered the conditions attached to a deposit ceiling as raw record fragments joined by semicolons, reading "house; finance to £600k; uk resident; down to 70% on other cases". Every part was accurate and none was written for a reader. It also printed a provider’s dispute sentence twice in the same cell. A conditioned fact should not lose its conditions when summarised, and it should not arrive as a payload either.
-
Two records in our open-questions file recorded contact state in the status field, with a status reading "not yet raised with the provider". That is not a resolution state, and it meant the status vocabulary appeared to have five values when it has three. Contact state has its own field and has been moved there, and the comparison page now shows it on each question, which the page had promised and never displayed.
-
Questions we had answered went on rendering as unanswered. Our records helper exported every entry in the open-questions file, closed ones included, and five pages read it as counts, as per-provider lists and as the sentence stating how many questions are unresolved. Two questions closed earlier the same day by retrieving the source documents continued to appear as open everywhere. A register of what we do not know is worth nothing if answering something fails to remove it.
-
Wayhome Deposit Bridge. We applied a regulatory protection statement to the wrong product. Our comparison page said Wayhome’s Deposit Bridge carries no Financial Ombudsman or compensation-scheme cover, citing the company’s published regulatory information. That document scopes its Ombudsman and FSCS sentence to "your interest in an LLP", which describes Wayhome’s former co-investment product. Deposit Bridge expressly gives the customer no ownership and no equity, so there is no LLP interest for that sentence to attach to. We now state only what the general regulatory statement supports and have raised the Deposit Bridge position as an open question. Right company, right document, wrong product, which is the error our product-scope rules exist to prevent.
-
Nomo by BLME. A reviewer reported that Nomo currently publishes a 4.99% limited refinance offer and that 5.25% is standard high-net-worth pricing rather than a refinance rate. We retrieved the live intermediary products page on 11 August 2026. It matches our 6 August capture in every figure: 5.25% is the two-year fixed refinance product for £750,000 to £5m, standard high-net-worth pricing at £750,000 to £1,999,999 is 5.49%, and no 4.99% rate appears anywhere on the page. No change made. Recorded because a check that confirms the existing record is worth the same as one that overturns it.
-
Our comparison page described a provider’s lowest rate anywhere in its range as differing from the benchmark only by finance-to-value tier. For two of the three that was wrong: the cheaper figure was a five-year fixed product in one case and a discounted variable in the other. The note now names the rate type and period, so a reader cannot read the lower figure as the same product on a bigger deposit.
-
Our comparison page said it covered halal mortgage providers UK-wide. None of the three mainstream providers lends on property in Scotland or Northern Ireland, and one lends in England only. Corrected to describe the UK market rather than UK coverage, and each provider row now shows the nations it lends in.
-
The column explaining what the non-Home-Purchase-Plan alternatives actually are rendered empty for both Wayhome and Pfida. The section exists specifically to explain why those propositions are not comparable to a Home Purchase Plan and what consumer protection differs, and it asked the question without answering it. Both descriptions were retrieved from the providers’ own live pages on 11 August 2026, including their statements that the products are outside FCA regulation and carry no Financial Ombudsman or FSCS cover.
-
StrideUp. Our comparison page described StrideUp’s minimum age and minimum property value disagreements as a current eligibility page against "older help material". That understated our own evidence. Both are disagreements between two current first-party channels: the consumer eligibility page states 18 and £75,000, and the intermediary criteria guide states 21 and £85,000. The summaries now describe the dispute as our records hold it.
-
The deposit column on our comparison page rendered no figure at all. When we rewrote the finance-to-value display module to return finished text rather than raw numbers, so that no page could flatten a disputed ceiling into a single value, the comparison page was not migrated with the others. It went on reading a numeric field that no longer existed, so every row showed a bare percent sign where the minimum deposit should be, on a page that repeatedly tells readers the deposit tier matters more than the choice of provider. Nothing caught it because the arithmetic produced an empty value rather than an error.
-
Both our rates page and our comparison page listed the lowest recorded rate for specialist providers as a bare figure. Nomo’s lowest is a refinance-only product, so a reader buying a home was shown a price not available for that transaction. Listing a provider rather than ranking it does not exempt its price from carrying what it depends on.
-
Our comparison page published a benchmark range of 5.58% to 5.80% and ranked providers by it without stating that the cheapest figure is a green product requiring an EPC rating of A or B. The 95% figure quoted alongside it is both EPC-gated and purchase-only. This is the same defect we corrected on the rates page earlier the same day, and the build check written for it did not catch this one: it identified a priced product by its product code, and the comparison page names no product codes at all. A page can price a product without ever naming it. The check now also treats a quoted rate as a handle, and both figures now carry their conditions.
-
While adding Offa’s newly retrieved criteria we cited them to a source id that was already in use for Offa’s tariff of charges, which cannot publish an age or property-value minimum. The build passed, because the id resolved. The page we had actually read was already registered under a different id, the third time in this project that a document we needed turned out to be one we already held. Both facts were repointed, and a build check now rejects a criterion cited to a kind of document that cannot carry it.
-
Gatehouse Bank. We have re-sourced our two Gatehouse minimum-age records after flagging their citation earlier today. The Personal Home Finance page publishes minimum age 18 for UK residents and 21 for UK expat and international customers, which matches the scoped values we already held. The document was already registered among our sources, so the values were never wrong and no new research was needed: the citation had simply pointed at the wrong one of our own records.
-
Offa. We retrieved Offa’s Home Purchase Plan overview page on 11 August 2026 and recorded two criteria we had never held: a minimum age of 18 and a minimum property value of £80,000. The page also publishes minimum finance of £60,000, a minimum 5% deposit, and availability to UK residents and British expats, all of which we already held. Our Offa research had only ever read the products page and the rate card, so criteria published on the parent page had never been captured.
-
Gatehouse Bank. Rebuilding the eligibility checker’s source trail so that it lists only the documents behind the rules actually tested revealed that both our Gatehouse minimum-age records cite "Our Shariah Approach", a page covering Sharia governance and supervisory-board membership that publishes no eligibility criterion. The figures may well be correct; the citation cannot support them. We have flagged the citation and shown it as under review on the page rather than repointing it at a likelier-looking source, because inventing provenance is worse than admitting we cannot show it.
-
Our eligibility checker stated that where a provider’s channels publish different figures we "check against the consumer figure". The engine has never done that, and it should not: a difference between a consumer page and an intermediary one can be a deliberate policy distinction, stale documentation or a genuine contradiction, and choosing between them on the basis of which audience a page addresses is not evidence. The description now matches the behaviour, which is to check against every published figure.
-
Our eligibility checker headed provider cards "no conflicts found" while displaying rows reading "the provider’s own channels publish 18 and 21". Both statements were true of different things: no criterion ruled the reader out, and the evidence behind one criterion disagreed with itself. Whether the evidence disagrees and whether that disagreement changes a particular reader’s result are independent questions, and the card reported only the second. Each card now reports both.
-
Our eligibility checker rendered no row at all for a criterion we hold no rule for, so an absence looked identical to a pass. Offa cards silently omitted minimum age and minimum property value while both appeared on other providers’ cards and the page stated it tests them. Those rows now say the criterion is untested rather than met, and count as unresolved.
-
Our eligibility checker resolved a disputed minimum property value by taking the lower of the published figures. On a settled rule that is simply the value; on a disputed one it silently decided the disagreement in the provider’s favour. The same criterion is now evaluated as an interval, matching how minimum age was already handled: below every published figure it fails, above every figure it passes, and between them it is unresolved.
-
Nomo by BLME. The same defect on Nomo’s maximum finance-to-value. Our note recorded that the product rows publish 75% while the section heading covering £2m to £5m states 70%, and the record carried only 75%. Both figures are now held and the ten affected rate rows are flagged as mirroring a disputed fact.
-
StrideUp. We recorded StrideUp’s minimum property value as disputed on 7 August and wrote both figures, £75,000 from the consumer eligibility page and £85,000 from the intermediary criteria, into the note. Only £75,000 went into a field. Every page reading that record therefore rendered £75,000 as settled, and our eligibility checker told a reader with an £80,000 property that the criterion was met when one of the provider’s own two published thresholds says it is not. Both figures are now in the data, the criterion reads as unresolved between them, and a build check rejects any disputed fact whose disagreement exists only in prose.
-
Our eligibility checker stated criteria for three providers without naming a single source document, on a page whose whole claim is that those criteria are published. Every other page on the site carries a source trail. It now lists the documents behind each provider’s rules, built from the same records the matching engine reads so the two cannot drift apart.
-
Our calculator stored the transaction type as "refinance" while our eligibility checker’s form used "remortgage". A reader who chose Remortgaging in the calculator and continued to the checker was silently switched back to Buying, while a banner told them the transaction had been carried over. The values are now mapped, and the EPC rating carries across as well.
-
Our matching engine compared EPC answers in the wrong vocabulary. Products record what they require as ratings, such as A or B, while a reader who does not know their exact certificate can only answer in a band. The comparison was a direct list membership test, so the band form would have told someone with an A-rated property that they failed a requirement their property meets. Nothing had reached the code, because no page asked the question. It now compares the set of ratings an answer allows against the set a product accepts, and a band that straddles a requirement stays unresolved rather than resolving either way.
-
Our eligibility checker never asked for a property’s EPC rating, so a criterion the matching engine has always been able to evaluate returned "unresolved" for every reader. Sixteen priced products in our record carry an EPC requirement. The question is now asked, and the reader’s answer resolves the criterion instead of permanently deferring it.
-
Offa. A reviewer stated that Offa’s current HPP finance range runs to £1,000,000. Our record holds £1,500,000 at the 65% and 80% tiers and £750,000 at 90% and 95%, captured 6 August 2026. We have not re-retrieved the source, so we are not changing a stored figure on the strength of an assertion, and we are not treating our own figure as confirmed against it either. Logged as an open discrepancy for the next scheduled rate check.
-
Kuwait Finance House PLC. We explained our specialist section by saying those providers serve a different customer. That is true of Nomo, which lends to GCC residents and not UK residents, and of QIB (UK), which is private banking. It is not true of Kuwait Finance House, which our own record shows serves UK and non-UK residents. It sits outside the ranked tables because its published criteria start at £250,000 of finance, cap finance-to-value at 70% and restrict the property to named regions, not because a UK resident cannot use it. The section now gives the actual reason per provider.
-
Three statements on our rates page described the market in the present tense when our evidence was dated. A variable rate was "what is published today", a rate card was "still on the provider’s live page today", and we stated that no provider had changed a published rate since we began recording. The first two outran the observation date shown elsewhere on the same page. The third could not be established at all: with one verified reading per product we can say no movement has been recorded, not that none occurred, because a rate could change and change back between checks. All three are now scoped to what we observed and when.
-
We published a fallback rate that was not always available either. Having added the winning product’s EPC and transaction conditions to our rates page, we then told readers who failed those conditions what "the lowest available here" was. Finance limits differ between providers: our record has one mainstream provider starting at £75,000 and another at £60,000, so a reader financing £65,000 at 80% finance-to-value could have neither the winner nor the fallback we named. A page that does not know a reader’s finance amount cannot identify the lowest rate available to them. It now reports the next-lowest published row, prints every product’s finance range, and a build check rejects the stronger claim.
-
Our rates page summary stated that three mainstream providers were the set the ranked tables compare against each other. Three are open to UK residents, but only two have a verified product-specific rate and enter those tables; the third is shown separately under advertised claims because we hold no rate we can attribute to one of its products. The summary now says which number is which.
-
We described the rates page dataset as Home Purchase Plan rates. It is not: of the 64 priced products we hold, 18 use commodity murabaha or murabaha rather than a co-ownership or lease plan, including every Nomo product and half of the Kuwait Finance House range. The page now describes its universe as Sharia-compliant home finance and reserves Home Purchase Plan for the mainstream tables where it is accurate.
-
Gatehouse Bank. We published lowest-rate claims that omitted the winning product’s eligibility conditions. Our rates page tagged Gatehouse Green products as the lowest in their band without stating that they require an EPC rating of A or B, and tagged the 95% winner without stating that it is purchase only. A reader with an EPC C property was shown 5.58% as their lowest two-year fixed at 80% finance-to-value when theirs was 5.68%; a reader refinancing at 95% was shown a product not open to them at any price. The conditions were already recorded against every affected product and were simply not rendered. Every page that prices a product now prints its conditions, and a build check refuses any page that does not.
-
The rates page selected its "what has changed" entries with a text search for the word "rate" or a percent sign, written before this log declared an entry kind. It matched 14 of our entries and displayed four, all of which were our own data-import corrections rather than any movement in a published rate. A page whose stated value is accumulated rate history was therefore leading that history with our spreadsheet fixes. It now selects on kind, and states plainly that no provider has changed a published rate since we began recording.
-
We recorded two defects in our own rendering that do not exist. Our open-exposures register stated that the rates page and the cost-comparison page displayed a disputed StrideUp finance-to-value ceiling as though it were settled. Neither page does: StrideUp holds no verified rate, both pages filter to products with a numeric rate before grouping by finance-to-value, and the tier headings in question contain only Gatehouse and Offa products whose ceilings are confirmed. The rows were written by reasoning from the code rather than reading the output. Both have been withdrawn, with the reasoning error recorded alongside them.
-
Our about page stated that Islamic Mortgage Guide is not FCA-authorised "and it does not need to be for what it currently does". The first half describes our own conduct and we can state it. The second half is a conclusion about where the regulatory perimeter sits, which is not ours to reach on a web page. The claim was withdrawn and replaced with a description of what we do and do not do.
-
Our how-we-make-money page cited the FCA guidance at PERG 4.5.14G as the basis for its introduction policy. That guidance sits in the chapter on regulated mortgage contracts. This site concerns home purchase plans, which have their own chapter: PERG 14.4 Q34 for the article 33A introduction exclusion and PERG 14.8 for financial promotions. The sources were corrected and linked.
Source -
Our site-wide footer stated that we may introduce readers to an FCA-authorised specialist. No such route exists and none is configured. The footer described a planned service in the present tense, on every page. Corrected across the site to say we do not currently provide advice or adviser referrals.
-
Gatehouse Bank. We recorded the Gatehouse Bank product-switch timing as a single figure disputed between three and four months, attributed to a difference between its consumer and intermediary channels. That was wrong twice: the statements measure four different things, and the consumer page carries both figures itself, so channel cannot be the explanation. Now recorded as five scoped observations with the relationship unresolved.
Source -
Gatehouse Bank. We recorded that Gatehouse Bank does not publish pricing for its existing-customer product switch range. It does: the existing-customer product page publishes rates, tiers, product fees and settlement charges. The correct statement is that the range is not yet modelled in our rate record, which is a different thing from unpriced.
Source -
Kuwait Finance House PLC. We recorded that Kuwait Finance House does not publish how early settlement of the deferred sale price is treated on its commodity murabaha. Its intermediary sheet states there are no early payment charges for the variable rate, and separately that there will be no discount for early settlement of the deferred sale price. Both are now recorded as separate facts rather than collapsed into one.
Source -
Kuwait Finance House PLC. We described the Kuwait Finance House commodity murabaha as a property buy-and-resell arrangement falling under the alternative property finance route at FA03/S73. It is not. The bank publishes it as first-charge finance: it takes a first legal charge while the client retains registered ownership, and does not purchase the property. The tax mapping was removed.
Source -
StrideUp. We recorded StrideUp as publishing a settled two-year UK residency baseline. Its help centre article for applicants on a visa publishes 12 or 24 months depending on occupation. Four of its live home purchase plan sources now sit side by side and none is treated as governing. At 24 months or more every published statement is satisfied; the unresolved band is 12 to 23 months.
Source -
StrideUp. We rendered the StrideUp maximum finance-to-value as a settled 90% on a comparison page. Our own canonical record marks it disputed at 85 or 90: the consumer page states 90% and the intermediary criteria state 85%. The figure had been read from the rate rows, which mirrored it without the dispute. A build check now fails if a disputed fact is copied elsewhere as settled.
Corrects entry CL-008, recorded 6 August 2026.
Source -
Gatehouse Bank. We published that Gatehouse Bank two-year self-employed trading requirement applies only above 80% finance-to-value, inferring it from the up-to-80% guide not stating a minimum. Its general HPP eligibility page states self-employed for a minimum of 2 years, with no finance-to-value condition attached. The claim was withdrawn. The finance-to-value difference we can evidence is minimum income: 15,000 pounds up to 80% against 20,000 pounds above it.
Source -
Offa. We recorded that Offa publishes no immigration criteria for its home purchase plan. It does. Page 7 of its HPP Criteria Guide is headed Visa Requirements for Non-UK Citizens and names ten acceptable statuses and routes, with the Family Visa capped at 65% finance-to-value. The guide sits behind the intermediary criteria-documents hub; we had checked the consumer documents hub only.
Source -
Gatehouse Bank. We recorded that Gatehouse Bank does not name the members of its Shariah Supervisory Board. It does. The board page names Sheikh Dr Nizam Yaquby, Sheikh Dr Esam Khalaf Al Enezi and Sheikh Dr Abdul Aziz Al-Qassar, and the Home Purchase Plan certificate of Shariah compliance, dated 27 January 2022, is signed by all three and lists the seven documents reviewed. We had read the Shariah approach page only.
Source -
On importing the provider workbook, effective dates were unformatted spreadsheet serials. Converted: Gatehouse effective 6 August 2026, Offa 14 April 2026, Kuwait Finance House 7 April 2026. Effective date is now shown separately from the date we checked a figure.
-
StrideUp. On importing the provider workbook, StrideUp rate rows had follow-on rate prose sitting in the numeric initial rate column. Moved to the follow-on field; initial rate left null, because the live rates page renders 0% to automated readers and no rate has been captured manually.
-
Kuwait Finance House PLC. On importing the provider workbook, Kuwait Finance House rate rows had the 0.75% arrangement fee sitting in the maximum finance column. Moved to the application fee; maximum finance left null, published as at bank discretion.
-
Nester. Correction to our own record. We had listed Nester as a provider awaiting checks. It is a property investment and finance platform, not an owner-occupier Home Purchase Plan provider, and is excluded from our provider counts.
Source -
Al Rayan Bank standard retail HPP. Correction to our own record. We previously held Al Rayan as a single restricted provider. It is two distinct positions: existing-customer servicing of legacy Home Purchase Plans, and Premier Home Finance for applicants from Qatar, Saudi Arabia, Kuwait, Bahrain and Oman. We have not verified any standard new retail application route.
Source -
Wayhome Deposit Bridge. Recorded that the Sharia certification published on Wayhome relates to Gradual Homeownership, which is retired. No product-specific certificate was verified for the current Deposit Bridge product, so we do not describe it as Sharia-compliant.
Source -
StrideUp. First recorded StrideUp published criteria: up to 90% finance-to-value, minimum finance 50,000 pounds, England only, 1,249 pounds product fee. Rates remain unpublished here because the live rates page returns 0% to automated readers, which is a display artefact and not a real rate.
Superseded by entry CL-015, 10 August 2026. This entry is left exactly as it was written.
Source -
Kuwait Finance House PLC. First recorded Kuwait Finance House PLC as an active regulated Islamic home finance provider: Ijarah and Commodity Murabaha, up to 70% finance-to-value, minimum finance 250,000 pounds, published variable rates from 5.5%. Previously Ahli United Bank (UK) PLC.
Source -
Pfida OwnTogether. First recorded Pfida's published regulatory position: its products are not FCA regulated and customers are not covered by the Financial Ombudsman Service or the FSCS, per its own homepage disclosure.
Source -
Al Rayan Bank standard retail HPP. First recorded Al Rayan Bank's home finance status: the only route shown is Premier Home Finance, currently available to applicants from Qatar, Saudi Arabia, Kuwait, Bahrain and Oman. The former residential Home Purchase Plan page returns a 404.
Source -
Gatehouse Bank. First recorded Gatehouse Bank's published Home Purchase Plan rate card: 16 UK-resident and 8 expat products, rental rates 5.53% to 6.94%, product fees 499 to 999, SVR 7.25%.
Source
The rules this log runs on
- Append only. An entry is never edited to say something else. If it was wrong, a new entry says so and both stay.
- A correction names both sides. What we had recorded, and what the source actually shows, with the source linked where there is one to link.
- Our mistakes are logged like anyone else's changes. There is no separate, quieter place for them.
- Dates are the date of the observation, or of the moment we found our own error, not the date we got round to writing it up. An entry date is never inherited from a site-wide checked date, and the build fails on any entry dated in the future.
- A correction can itself be corrected. If one of these entries turns out to be wrong, a new entry says so and the original stays where it is.
Recheck windows are set out in the methodology. The date next to a figure is what actually happened, and that is the one to read.