
A website can contain the correct broker disclosure and still tell customers the wrong story. This practical inspection follows twelve ways that live pages, results and follow-up messages can misrepresent the service, with concrete checks and evidence of repair.
A green tick appears. “You're approved.” The customer has supplied an email address, a postcode and a preferred loan amount. No lender has made a lending decision. Somewhere below the fold, a footer explains that the business is a credit broker. The footer may be accurate. The journey is not.
That hypothetical failure is the starting point for a useful website inspection. The question is not whether the compliance sentence exists. It is whether the customer can leave each screen with an accurate understanding of who is acting, what has happened and what happens next. A disclosure can survive every internal approval meeting and still disappear behind a cookie banner on a phone. Credit broker website red flags are often failures of assembly: approved words, a third-party widget, a marketing button and an automated email combining to say something nobody expressly authorised.
These twelve tests are Authorised Compliance Ltd's practical quality-assurance controls, not an FCA-issued checklist or prescribed wording. The scenarios are illustrative. Their purpose is to make defects reproducible and repairs demonstrable. The regulatory starting point, checked on 21 October 2026, is the current Handbook. CP26/15 proposed changes to consumer credit promotions rules; a consultation proposal is not permission to remove an existing disclosure. This inspection applies the current rules, not the consultation's proposed amendments.
The focus is consumer credit broking, principally journeys for credit not secured on land. Check each rule's application provisions before using it for other products. Consumer Duty applies to firms' retail market business and relevant promotions to retail customers, including prospective customers. It is not a universal regime for every corporate borrowing transaction or every activity of an authorised business. Nor does calling a journey “business finance” settle its regulatory treatment. PRIN 2A.1 explains the Duty's application.
Start with actual entry routes, not just the homepage: paid-search landing pages, partner links, returning-customer links and results pages. Use agreed test arrangements. Do not create real borrowing applications, trigger hard searches or pass invented customer records to lenders without authorisation. Record the URL, date, device, entry route, test inputs and version served. A screenshot without the route that produced it may be impossible to reproduce. Include unsuccessful journeys: no match, unavailable lender, interrupted connection and abandoned application.
Imagine a page headed “Our loans. Our instant decisions.” Its only correction sits beneath several screens of testimonials. The problem is the service being claimed in the main sales message, not merely the location of the disclaimer. Inspect the headline, supporting copy, buttons and loading messages together. Ask who “we” means in each sentence. Check whether the business actually makes the decision being described, and whether the wording changes between broking and lending journeys where the firm performs both roles. Where CONC 3.7.7R applies, financial promotions by a broker that is not a lender must prominently state that status. It also addresses firms that do both but promote only their broking service. Repair the claim itself. Keep the original and replacement screens, the business owner's confirmation of the actual service, and a recording of the corrected route. Moving a contradictory sentence closer to the button does not make the contradiction disappear.
Desktop approval can conceal a mobile failure. A sticky application button remains visible while the broker disclosure moves below a long image. A chat bubble covers it. Enlarged text clips the final words. Open the live page on the phone sizes your customers actually use. Test it with the cookie banner open and closed, with text enlarged, and after the keyboard appears. Check contrast, wrapping and whether important wording becomes available only through an inconspicuous accordion. CONC 3.7.8G explains prominence by reference to whether the intended average person's attention is likely to be drawn to the statement in context. It does not prescribe a universal pixel height or “above the fold” formula. For the repair record, capture the same journey before and after at the same viewport size. Include the obstructing overlays. A clean desktop screenshot proves remarkably little about a customer trying to read underneath three floating buttons.
A recognisable trading name fills the page. The legal identity belongs to another business in the footer. The embedded form names a third. The customer has no reliable way to connect them. Compare the page, form, privacy notice and confirmation message with the relevant Financial Services Register entries. CONC 3.7.5R requires the firm's legal name as recorded on the Register, not merely its trading name. Where an appointed representative operates the site, check that its status and principal relationship are described accurately. Do not transplant the principal's authorisation statement in a way that implies the representative holds that authorisation itself. An introducer appointed representative's narrower appointment must not be presented as unrestricted authority to operate the whole broking service. SUP 12 sets out the appointment framework. Evidence of repair should include a dated identity check, corrected screens and a short map showing which entity performs each activity. For a white-label journey, ask a reviewer unfamiliar with the arrangement to identify the responsible broker without being coached through the group structure.
The button promises an eligibility check. Pressing it submits a full application, starts a different credit search or sends the customer into a process they have not been told about. Follow the submission into the systems behind it. Establish what information is sent, to whom, what checks occur and what the customer has actually requested. A credit application can legitimately be submitted through a broker. The defect is describing one action while performing another. Pay particular attention to “no impact on your credit score”. Does that claim describe this initial check, or does the design make it appear to cover the subsequent lender application too? Test retries and returning visits as well as the first attempt. Repair the relationship between label, explanation and action. Retain the approved process description, redacted submission logs and the revised screen sequence. Ask test users what they believe pressing the button will do before showing them the next screen. That is more revealing than asking whether the wording looks clear.
A lender returns “potential match”. The broker displays “approved”. An application awaiting further checks receives a celebratory animation and an invitation to collect funds. Inspect the translation between the lender's actual response and every customer-facing result. Include pending, referred, declined, expired and technical-error states. A failed connection must not inherit yesterday's positive message. CONC 3.3.3R and its accompanying guidance address claims that credit is available regardless of financial circumstances or status, with a specific exception for certain pawnbroking agreements. The guidance warns that guaranteed or pre-approved claims may contravene the rule. Separately, an eligibility indication must not be misrepresented as a lending decision. The repair evidence is a response-to-message table supported by test runs, not a promise that someone changed the success page. Where a lender has genuinely communicated a conditional approval, identify whose decision it is and explain the outstanding conditions accurately. Do not replace every useful result with vagueness; replace unjustified certainty with the actual position.
The visitor moves a slider and sees “Your monthly payment”. The number was calculated from an assumed rate, not a lender's offer to that customer. Test the assumptions behind the display: amount, term, interest, fees and any initial or final payment. Change the inputs to awkward values. Check what happens when the advertised product is unavailable and whether rounding creates inconsistent totals. Make the distinction between an illustration, an eligibility result and an actual offer visible where the figure is used. A small “illustrative only” label cannot sensibly repair a large promise of a personal quote. Also assess the applicable cost-disclosure triggers. Under CONC 3.5, interest or cost figures can trigger a representative example, while certain other claims can trigger representative APR disclosure, subject to the relevant exceptions. Retain independent calculation checks, input-and-output captures and the reasoning for the disclosures used. The repair should establish both that the arithmetic works and that the number is being presented as the right kind of information.
“We compare 40 lenders” appears above results generated from three available integrations. Some panel members do not offer this product; others are unavailable to this customer. The top result is labelled “best” because it occupies the first position. Inspect three different things: the contractual panel, the lenders actually considered and the offers returned. They are not interchangeable. Reconcile the claim with live routing rules, product restrictions, outages and eligibility filters. Then inspect how results are ordered, including any commercial influence. CONC 3.7.3R requires firms within its scope to explain the extent of their powers. Guidance in CONC 3.7.4G addresses independence claims. Neither makes an impressive logo wall proof of market coverage. Repair the coverage and ranking descriptions to match the service delivered. Preserve a dated panel record, sample routing logs and the ranking explanation. Test an excluded lender and an unavailable integration. If the headline remains true only when every partner is functioning and every customer qualifies, it needs more work.
The landing page says the service is free. A later screen introduces a processing fee, a subscription or a charge described as verification. Trace every amount the customer can become liable to pay, including charges imposed through another entity. Identify who charges it, what triggers it, when it is collected and what cancellation or refund terms actually apply. Distinguish a broker service with no customer fee from borrowing that carries no interest or charges. Where the information-notice and customer-confirmation requirements in CONC 4.4 apply, inspect those controls before fee-taking or requesting payment details for that purpose. A generic acceptance of website terms is not evidence, by itself, that the relevant requirements have been met. A repaired journey should be supported by the charging schedule, customer-facing explanations and test records showing that payment cannot proceed prematurely. Recheck abandoned journeys too. The customer who leaves before receiving an offer may still be the person who discovers what “free” meant.
“No broker fee” answers one question. It does not explain how the broker is paid or whether payment arrangements influence what the customer sees. Compare the actual remuneration arrangements with the wording beside recommendations, results and the route to an agreement. “We may receive commission” deserves scrutiny where payment is a normal, known feature of the arrangement. So does an assertion that commission cannot affect the customer's costs when nobody has established that. Within its scope, CONC 4.5.3R requires prominent, timely disclosure of remuneration's existence and nature where it could affect impartiality or materially affect the customer's decision. CONC 4.5.3AR also requires an explanation, at the same time and with equal prominence, of how it may affect amounts payable. CONC 4.5.4R addresses disclosure of the amount, or likely amount, on request before the agreement. These are not a universal instruction to publish an individual commission figure on every webpage. They are also not satisfied by a sentence disconnected from the commercial arrangement. Keep the remuneration-to-disclosure mapping, evidence of when the explanation appears and a tested response to a customer asking how much the broker will receive. This website inspection is not an exhaustive assessment of wider legal duties concerning commission.
The form appears to send an enquiry to a lender. Behind it, information goes first to a platform and then to other brokers. The privacy notice offers only the comforting phrase “trusted partners”. Compare the explanation at collection with the real data route. Identify the recipients or meaningful recipient categories, their roles, the purposes of sharing and the relevant lawful basis. Check whether the route changes when the first destination returns no match. UK GDPR Article 13 requires privacy information when personal data are collected from the individual, subject to its exceptions, including purposes, legal basis and recipients or categories of recipients. It does not invariably require every recipient to be individually named on the opening screen. A practical repair can use a concise explanation at the form with accessible supporting detail. It must describe the actual arrangement. Keep a redacted data-flow test and the notice version shown in that test. Treat transparency, the lawfulness of sharing and any permissions needed for subsequent contact as separate checks, including applicable FCA restrictions. A privacy link cannot do every job.
The customer presses “Continue”. A lender's form opens under almost identical branding, or another broker takes over without any visible explanation. The customer assumes the original firm is still making every decision. Inspect the outgoing screen, redirect, destination and any embedded form together. Identify the point at which another firm begins receiving information or providing the service. Explain the next firm's role and the next step accurately. Under CONC 3.3.2R, the lender must be named in relevant broking communications where it is known. Test failed redirects, expired sessions and the back button. Check whether the customer can tell whom to contact about a missing application, a broker service issue or a lender's decision. The repair evidence should combine a screen recording with a responsibility map agreed by the relevant businesses. A mandatory extra pop-up is not the universal answer. Sometimes a clear transition within the existing flow works better. The test is whether the change of firm remains understandable when something goes wrong.
The website now says “enquiry received”. Five minutes later, the automated email announces “Your loan is ready”. The website repair has lasted precisely one marketing automation. Inspect the messages generated by the actual test journey: email subjects, SMS previews, reminder sequences and chatbot responses. Check abandoned, declined and expired cases, not just successful introductions. Compare each message with the current application state and identify whose service or decision it describes. Remove invented deadlines, unsupported statements that funds are waiting and messages that treat a referral as an approval. Check separately that the contact itself is permitted; correcting the wording does not resolve every marketing-permission issue. Keep the delivered messages, the conditions that triggered them and evidence that obsolete sequences have been disabled. Rerun the journey after a status changes. An old reminder that arrives after a decline is not made accurate by having been scheduled when the application was still open.
“Copy amended” is an activity, not a demonstrated repair. A useful closure record connects the original defect, its likely customer consequence, the change made and the successful retest on the affected route. Give the action an owner and record the version actually released. For serious misleading claims, consider whether the affected promotion or route should be stopped while the issue is resolved. Assess customers already exposed, rather than assuming a corrected screen disposes of the past. The appropriate response depends on the facts and the risk.
Then check understanding. Where Consumer Duty applies, PRIN 2A.5.3R requires communications that support properly informed decisions. Screenshots can show what was displayed; they cannot establish, on their own, what someone understood. Proportionate testing might ask intended users to explain who will decide, what the result means and who will receive their details.
Authorised Compliance Ltd helps firms connect regulatory requirements with practical website reviews and customer-journey controls. The value is not another immaculate document describing a journey that customers never see. It is finding the point where the firm's intended explanation and the customer's reasonable understanding part company. The customer should not have to solve the organisation chart before pressing “Continue”.

I’m Will Hurst, and I bring 20+ years of hands-on experience across credit broking, AR/IAR oversight, lender relationships and regulated finance operations.
Learn more about my practical, FCA-focused approach