The Unpublished Rulebook: How Much of the High Street's Mortgage Criteria Is Actually Published
Updated: 6 days ago
See how much of the high street rulebook actually reaches the published criteria page, and how much stays behind it.
Quick Answer
Manor Mortgages Direct put 618 standard broker criteria questions to 18 published high street criteria sets in September 2026, producing 11,124 question-and-lender pairs. On 36.0 per cent of those pairs the published page gives no answer at all, and only 11.4 per cent are a published refusal.
The gaps are not random. Of the 618 questions, 279 are answered by four or fewer of the nine residential lenders tracked, and 69 are answered by none of them. A further 14.4 per cent of all pairs carry a note that mentions the subject without deciding the case.
"Will consider" here means the published criteria do not exclude the case. It is not an offer, and every application is assessed individually on its own facts. These counts describe the lenders whose published criteria Manor Mortgages Direct tracks rather than the whole of the lending market, and no lender is named anywhere in this study.
Reviewed by Ben Stephenson, FCA-authorised mortgage adviser (FRN 496907) · 25+ years' experience · 4.9★ on Google. Updated: 9 September 2026.
Key Points
279 of 618 questions answered by four lenders or fewer
36% of question-and-lender pairs get no published answer
Only 11.4% of pairs are a published refusal
Table of Contents
Nearly half the question set is answered by four lenders or fewer
Of the nine high street residential lenders whose published criteria Manor Mortgages Direct tracks, 279 of the 618 standard broker criteria questions are answered by four lenders or fewer, checked September 2026. That is a little over 45 per cent of a question set assembled from the things brokers routinely have to ask before they can place a case. The reasonable expectation is that a core question set is answered by most lenders most of the time. The tracker does not show that.
At the two ends of the distribution the picture is sharper still. 69 of the 618 questions are answered by none of the nine residential criteria pages, and 124 are answered by all nine. Everything else sits in between, covered thinly and unevenly, with no clean majority position to read off the page.
The middle of the distribution is where the practical difficulty lives. 46 questions are answered by exactly one of the nine, 52 by two, 51 by three and 61 by four. Above the halfway mark the counts thin out again before rising at the top: 61 questions are answered by five, 43 by six, 46 by seven and 65 by eight.
Number of the nine residential lenders answering | Number of the 618 questions |
None | 69 |
One | 46 |
Two | 52 |
Three | 51 |
Four | 61 |
Five | 61 |
Six | 43 |
Seven | 46 |
Eight | 65 |
All nine | 124 |
The consequence for anyone comparing lenders is direct. On roughly one question in five you can line up all nine published positions and compare them properly. On the rest you are comparing a handful of published positions with a larger number of blanks, and a blank is not a position. That is a different exercise from the one most comparison work assumes it is doing.
A published no is the rarest thing on the page
Across the 11,124 question-and-lender pairs in the study, Manor Mortgages Direct recorded no answer on the published page for 36.0 per cent and a published refusal for only 11.4 per cent, checked September 2026. A further 14.4 per cent of pairs carry something on the page that mentions the subject but does not decide the case, a referral note, a "subject to underwriting" line or a pointer to a helpdesk. Put together, half of all pairs leave the question open once you have finished reading.
The positive side of the ledger is smaller than most people assume. 26.0 per cent of pairs are a published position that the case will be considered, and 12.1 per cent give an actual figure or a hard rule, a maximum, a minimum, a percentage or a time period. Those two categories together are the part of the rulebook that can be relied on without asking anyone.
This matters for how the phrase "the lender's criteria say" is used in practice. In the tracker, that phrase describes a published page, and on the evidence here the published page is a partial document rather than a complete one. A published refusal is a rule. Silence is not a rule, and the study treats it as silence throughout.
It also means that a broker who reports back that a case cannot be placed with a particular lender is very often reporting the outcome of a conversation rather than the content of a page. Roughly one pair in nine is a published no. The other refusals that happen in the market happen somewhere the study cannot see, which is the honest limit of what this data can tell anyone.
Silence clusters by topic, and it clusters where cases are hardest
Of the 14 criteria topics Manor Mortgages Direct tracks, retirement and later life is the least published, with 70.7 per cent of its 846 question-and-lender pairs getting no answer on the page, against 17.5 per cent for deposit and source of funds, checked September 2026. Once notes that do not decide the case are added in, 77.5 per cent of later life pairs leave the question open, and only 5.3 per cent give a figure. Just 2.0 per cent are a published refusal, which is the lowest refusal rate of any topic in the study.
The pattern repeats down the table. Schemes and affordable housing sits at 52.9 per cent with no answer on the page, process, documents, valuation and fees at 43.9 per cent, and property type and construction at 41.1 per cent across its 2,232 pairs. These are the areas where a case either works or does not, and they are the areas where the page is quietest.
Adverse credit behaves differently and is worth separating out. Only 31.7 per cent of its 972 pairs get no answer at all, but 61.0 per cent leave the question undecided once notes are counted, and a mere 1.4 per cent give a figure. In other words the subject is usually mentioned and rarely settled, which is consistent with what the firm found when it looked at how the published adverse credit ladder is graded, and it is why a joint application where one applicant has the credit history is so often a question about how each lender treats a partner's record rather than a question about the page. Checking your own file before any of this is the cheap first step, and a full multi-agency check is the usual way to do it.
At the published end of the range the topics look completely different. Loan size, LTV, term and affordability has only 18.2 per cent with no answer and 50.8 per cent giving a hard figure, easily the most concrete topic in the set. Deposit and source of funds is next at 17.5 per cent with no answer, and self-employed and company income is better published than its reputation suggests, at 23.6 per cent with no answer, 38.7 per cent published as considered and 18.2 per cent giving a figure.
One topic publishes refusals more than any other. Expat, foreign national and visa questions get no answer on 24.3 per cent of pairs, but 38.5 per cent are a published no, which is more than three times the study average. Where lenders have decided to exclude something, they do tend to say so, and this topic is the clearest example of that in the data.
Topic | Share of pairs with no answer on the page |
Retirement and later life | 70.7% |
Schemes and affordable housing | 52.9% |
Process, documents, valuation and fees | 43.9% |
Property type and construction | 41.1% |
Buy-to-let specific | 34.9% |
Interest only and repayment | 34.3% |
Employed and other income | 33.5% |
Adverse credit | 31.7% |
Benefits and maintenance income | 30.9% |
Remortgage, capital raising and transactions | 25.5% |
Expat, foreign national and visa | 24.3% |
Self-employed and company income | 23.6% |
Loan size, LTV, term and affordability | 18.2% |
Deposit and source of funds | 17.5% |
The same lender publishes more on residential than on buy-to-let
Of the seven brands in the study that publish both a residential and a buy-to-let criteria page, Manor Mortgages Direct found 705 of 4,326 matched questions answered only on the residential page and 267 answered only on the buy-to-let page, checked September 2026. On 1,640 of those matched questions both pages give an answer, and on 1,714 neither does. The asymmetry is not marginal: the residential page is the fuller document by a factor of more than two and a half.
One buy-to-let page in the set states that anything it does not cover follows the lender's residential policy, which mechanically inflates its silence. Removing that brand from the comparison leaves 3,708 matched questions, of which 479 are answered only on the residential side. The direction of travel survives the adjustment even though the size of the gap shrinks.
At sector level the same thing shows up in the headline percentages. The nine residential sets return no answer on 31.7 per cent of their 5,562 pairs; the nine buy-to-let sets return no answer on 40.4 per cent of theirs. Buy-to-let sets are also slightly more likely to publish a refusal, at 12.1 per cent against 10.7 per cent, and less likely to publish that a case will be considered, at 21.4 per cent against 30.7 per cent.
There is no need to read anything sinister into this. Buy-to-let pages tend to be shorter documents aimed at a narrower set of transactions, and a good deal of the underlying policy genuinely is shared with the residential book. The practical point is simply that a buy-to-let question is more likely to need asking than a residential one, and that the answer is more likely to come from a person than from a page.
The gap between the fullest and the thinnest criteria set is more than double
Excluding one buy-to-let page that defers to its lender's residential policy, the share of questions with no answer on the page ranges from 19.6 per cent to 46.8 per cent across the criteria sets Manor Mortgages Direct tracks, checked September 2026. The deferring page sits well outside that range at 65.4 per cent and is reported separately for that reason. Even setting it aside, the fullest set answers more than twice as high a proportion of the question list as the thinnest.
Four of the 18 sets come in below 25 per cent, and five come in above 44 per cent. That is a real difference in how much of the rulebook a lender has chosen to commit to writing, and it does not correlate with how easy that lender is to place a case with. A thin page can belong to a flexible lender and a full page to a strict one.
The spread has an odd practical effect. A lender that publishes a lot gives brokers and comparison tools more to quote, which means it appears in more comparisons, including the unfavourable ones. A lender that publishes little is quoted less often and is harder to rule out on paper, which can look like flexibility when it is really just absence of information.
For anyone reading criteria at a distance, the safest habit is to treat the volume of published material as a fact about the document and not a fact about the lender. The tracker can tell you which page is fuller. It cannot tell you which lender is more likely to help, and this study makes no attempt to claim otherwise.
What the silence actually looks like
The 69 questions that none of the nine residential pages answer are not obscure, checked September 2026, and Manor Mortgages Direct records them as unanswered across all nine sets rather than as refusals. They include whether a higher lending charge applies, how non-standard utilities are treated, most retirement interest-only questions, properties near overhead power lines, income from influencer or social media work, purchases made for a student to live in, historical administrations in a company's history, local authority grant schemes and the maximum number of bedrooms a property may have.
Read as a group, these fall into two families. Some are genuinely uncommon, so a lender may reasonably prefer to handle them one at a time rather than publish a line that would need constant revision. Others, such as later life lending and newer forms of self-employment, are simply moving faster than criteria pages are being rewritten.
The 124 questions that all nine answer are just as revealing. They cover accountant certificates, payslips and bank statements, large loans, additional borrowing, acceptable deposit sources, maximum loan to value for first time buyers, the affordability calculator, the maximum number of applicants, non new build flats, porting, shared ownership, zero hours contracts, commission pay, umbrella companies, part and part repayment, share of business classed as self-employed, and the treatment of pre-settled status and non-EU foreign nationals. These are the questions that come up on ordinary cases every week.
The shape that emerges is a published rulebook built for the standard case and largely silent on the rest. That is a defensible way to write a criteria page, but it has a consequence worth naming. The further a case sits from the ordinary, the less the page will tell you, which is exactly why the firm keeps a separate specialist hub for the lenders whose criteria have to be read in full rather than skimmed.
Method, limitations and what the study does not say
The tracker is Manor Mortgages Direct's own internal record of published lender criteria, maintained so that advisers can check a position before a client is put through an application. For this study, 618 standard broker criteria questions were put to the published intermediary criteria pages of 11 high street lender brands, covering 18 criteria sets, nine residential and nine buy-to-let, and producing 11,124 question-and-lender pairs. Each pair was recorded as one of five outcomes: no answer on the page, a note that does not decide the case, a published position that the case will be considered, a specific figure or rule, or a published refusal. Pages were captured on 1 and 2 September 2026.
"Will consider" means only that the published criteria do not exclude the case. It is not an offer, not a decision in principle and not a promise of any kind, and every application is assessed individually on its own facts by the lender concerned. All counts in this study describe the lenders whose published criteria are tracked, not the whole of the lending market, and lenders are anonymised throughout.
Four limitations apply specifically to this data and should be read alongside every figure above. First, "no answer on the page" means exactly that and nothing more: a lender may well hold a clear policy in an internal criteria tool, on a broker helpdesk or in guidance given by a business development manager, so the study measures what is published rather than what is decided. Second, the 12 specialist and building society lenders in the same tracker are recorded from full criteria guides answered question by question, which means silence cannot be measured for them in a comparable way, and they are excluded from every figure in this study.
Third, criteria change constantly, and a page captured on 1 September may read differently a fortnight later, so these figures are a snapshot rather than a standing description. Fourth, one buy-to-let page states that anything it does not cover follows the lender's residential policy, which inflates its recorded silence; it is shown separately at 65.4 per cent wherever the per-set range is quoted.
The study also does not say several things it might be read as saying. It does not say that lenders are withholding information, and nothing in the data supports that reading. It does not rank lenders, and it does not measure how likely any lender is to lend on any case.
Finding | Figure and its denominator |
Questions answered by four or fewer residential lenders | 279 of 618 questions, nine residential sets |
Questions no residential lender answers | 69 of 618 questions, nine residential sets |
Questions every residential lender answers | 124 of 618 questions, nine residential sets |
No answer on the published page | 36.0% of 11,124 question-and-lender pairs |
A note that does not decide the case | 14.4% of 11,124 question-and-lender pairs |
Published refusal | 11.4% of 11,124 question-and-lender pairs |
Highest topic silence, retirement and later life | 70.7% of 846 pairs in that topic |
Lowest topic silence, deposit and source of funds | 17.5% of 342 pairs in that topic |
Answered only on the residential page | 705 of 4,326 matched questions, seven brands |
Answered only on the buy-to-let page | 267 of 4,326 matched questions, seven brands |
Per-set silence range, excluding the deferring page | 19.6% to 46.8% across 17 criteria sets |
Reading your own position off this data
If your circumstances are ordinary by the standards of the question set, a full time employed applicant, a standard construction house, a mainstream loan to value, then the published pages are likely to answer most of what applies to you. That is the part of the rulebook that is written down, and 124 of the 618 questions are answered by every residential lender tracked. Comparison on paper is reasonably reliable in that territory.
If your circumstances sit outside that, the published pages will answer less, and the study shows where the thinning happens. Later life borrowing, scheme purchases, unusual property types and newer forms of income are the topics where the page is quietest, at 70.7 per cent, 52.9 per cent, 41.1 per cent and 33.5 per cent with no answer respectively. On those subjects, an absence of any published position is the most likely thing you will find.
The single most useful conclusion for a reader is about interpretation rather than lending. When you read that a lender's criteria do not mention your situation, that is not a refusal and should not be treated as one. Only 11.4 per cent of the pairs in this study are a published no, and silence was excluded from that count deliberately.
The corollary is that a case can rarely be ruled in or out from published criteria alone once it stops being standard. Half of all pairs in this study end in something that does not decide the case, which is the practical measure of how much of the rulebook lives in underwriting rather than on the page.
To cite this study: Manor Mortgages Direct, "The Unpublished Rulebook: How Much of the High Street's Mortgage Criteria Is Actually Published", September 2026, manormortgagesdirect.com/unpublished-rulebook. Figures may be quoted with that attribution; lenders are anonymised and counts describe the lenders tracked.
FAQs
How many criteria questions did the study cover?
618 standard broker criteria questions were put to 18 published criteria sets, nine residential and nine buy-to-let, from 11 high street lender brands. That produces 11,124 question-and-lender pairs in total. Every figure in the study is drawn from those pairs, checked September 2026.
What does "no answer on the page" mean?
It means the published intermediary criteria page does not address the question at all, in either direction. It does not mean the lender has no policy, because a position may exist in an internal criteria tool, on a broker helpdesk or in guidance from a business development manager. Across the study, 36.0 per cent of pairs fall into this category.
Is silence the same as a refusal?
No, and the study deliberately keeps the two apart. A published refusal accounts for only 11.4 per cent of the 11,124 pairs, while 36.0 per cent get no answer at all and a further 14.4 per cent carry a note that does not decide the case. Silence is recorded as silence and never counted as a no.
Which topics are the least well published?
Retirement and later life is the least published, with 70.7 per cent of its 846 pairs getting no answer on the page, followed by schemes and affordable housing at 52.9 per cent. Adverse credit is mentioned more often but settled less: 61.0 per cent of its 972 pairs leave the question undecided and only 1.4 per cent give a figure. Deposit and source of funds is the best published at 17.5 per cent with no answer.
Do lenders publish less about buy-to-let than residential?
On this data, yes. Among the seven brands publishing both, 705 of 4,326 matched questions are answered only on the residential page against 267 only on the buy-to-let page. Excluding one buy-to-let page that defers to its lender's residential policy, the residential-only figure is 479 of 3,708 matched questions.
Why are specialist and building society lenders excluded?
The 12 specialist and building society lenders in the same tracker are recorded from full criteria guides, answered question by question rather than read off a summary page. That means silence cannot be measured for them in a comparable way, so they are excluded from every figure in this study. The study therefore covers 11 high street brands only, checked September 2026.
Summary
Putting 618 broker questions to 18 published high street criteria sets produced 11,124 pairs, and just over a third returned nothing at all. Only about one pair in nine was an outright published no, while 279 of the 618 questions drew answers from four or fewer of the nine residential lenders. Coverage thins fastest on later life, schemes and unusual property. The published page is a partial record of how lending decisions get made.
Reviewed by Ben Stephenson, FCA-authorised mortgage adviser, CeMAP-qualified.
Manor Mortgages Direct is FCA authorised, FRN 496907, has traded for 25 years, is highly positively reviewed, 4.9 rated on Google, and has helped thousands secure the right mortgage. Bristol-based mortgage brokers, assisting clients nationwide.
Sources
Manor Mortgages Direct (2026), criteria tracker, 618 standard broker criteria questions across 18 published high street criteria sets, checked September 2026 - https://www.manormortgagesdirect.com/unpublished-rulebook - accessed 9 September 2026
Financial Conduct Authority (2026), Mortgages and Home Finance: Conduct of Business sourcebook - https://www.handbook.fca.org.uk/handbook/MCOB/ - accessed 9 September 2026
Related Guides