7 checks for referral attribution work with Leeds financial services clients

What to take away

Check one: attribution model choice for referral programmes

Pick the model before you brief an agency or configure a platform. In Leeds financial services marketing, the choice decides which partner gets credited and which budget line gets defended at quarter end.

Last click is the default in most referral marketing software. It is easy to explain and easy to game: a cashback site or a comparison page can sit at the end of a chain it did nothing to start.

First click suits introducer and professional-connection programmes, where the solicitor or accountant who opened the door deserves the credit. It undervalues the nurture that follows.

Linear and position-based models spread credit across the chain. They are harder to reconcile with partner agreements that promise a fixed fee per approved application.

Data-driven attribution needs volume. A Leeds intermediary referral scheme with forty conversions a month will not produce stable weights, and the model will drift between reporting periods.

Write the model into the programme rules, not just the platform settings. If a partner asks why their referral was not paid, the answer should be a published rule, not an analyst's judgement. Document the model, the attribution window and the tie-break rule in the same place partners read their payment terms.

For a wider comparison of how these models behave in loyalty and referral settings, see attribution methods compared.

Check two: FCA record-keeping for Leeds financial services

If your firm is authorised or appointed representative, referral records are part of your books. The FCA's general rules on record-keeping in SYSC 9.1 require a firm to arrange for orderly records of its business and internal organisation, including all services and transactions, sufficient for the FCA to monitor compliance and check that the firm met its obligations to clients. For a referral programme, that means records that show what was communicated, to whom, when and on what basis.

A referral is a communication. A click log alone rarely satisfies an examiner who wants to see the introduction, the consent and the outcome.

Keep the chain: who referred, what they were told, what consent they gave, which firm received the introduction and what happened next. Store it where it can be retrieved without the agency's help.

Retention periods should follow your firm's own schedule and the FCA's record-keeping requirements, not the platform's default. SYSC 9.1 itself requires a common platform firm to retain records relating to its MiFID business for at least five years, and other firms should take the period from the rules that apply to their own business. Some referral marketing software deletes event data after twelve or twenty-four months unless you change the setting.

Leeds Building Society, First Direct and the high street banks' city offices recruit from the same pool, and staff move between them. Records that live only in one account manager's inbox will not survive that turnover.

Test retrieval twice a year. Ask a colleague who did not build the programme to pull a sample referral from eighteen months ago and explain the payment decision.

Check three: agency contracts and data ownership

Agency contracts referral clauses are where most Leeds programmes quietly fail. The agency runs the platform, the client assumes it owns the data, and nobody checks until the contract ends.

Name the controller and the processor in writing. The ICO's detailed guidance, Controllers and processors, helps decide whether each party acts as a controller, processor or joint controller, and it matters when a consumer asks what happened to their referral data.

Decide who owns the raw event data, the derived attribution records and the reporting outputs. These are three different assets and they can have three different owners.

Set exit terms. On termination, the client should receive a complete export in a documented format, plus the model configuration and the consent records. Without that, you cannot reproduce last year's numbers.

Check sub-processors. Many referral platforms use hosting, email and analytics suppliers outside the UK. The contract should list them and require notice of changes.

Agree audit rights. You need the right to inspect how referrals were counted, not just the dashboard totals.

Add that last clause to the contract checklist as well: referral payouts and data errors both scale with spend, so the liability cap should too.

Check four: consent evidence across referral chains

Referral chains involve at least two people: the existing customer and the friend. Consent for one is not consent for the other.

The referrer needs a lawful basis for sharing the friend's details. In practice that means the referrer confirms they have permission, and you record that confirmation.

The friend needs clear information before you track them. The ICO's Guidance on the use of storage and access technologies covers cookies, pixels and similar technologies used to follow a referral link, and it applies whether the tracking sits on your site or a partner's.

PECR rules on electronic marketing sit alongside UK GDPR. A referral email to a friend is still marketing unless a genuine personal recommendation exemption applies, and that exemption is narrow. The rule for email is regulation 22 of the Privacy and Electronic Communications (EC Directive) Regulations 2003: unsolicited direct marketing by electronic mail needs the recipient's prior consent, and the soft opt-in only covers people whose details were obtained in the course of a sale or negotiations for a sale.

Record what was shown, when, and which version of the privacy notice. A consent record without the notice version is hard to defend two years later.

Avoid pre-ticked boxes and bundled consent. They weaken the whole chain, and Leeds compliance teams will flag them in a review.

Check five: reporting definitions Leeds teams can audit

Every number in a referral report should have a written definition that a finance analyst can test. Otherwise the same dashboard tells two different stories.

Define a referral, a qualified referral, an approved application and a paid reward. These are not synonyms and mixing them inflates conversion rates.

Agree the attribution window. Thirty days, ninety days or the life of the cookie produce very different partner league tables.

State the treatment of returns, cancellations and clawbacks. Financial services products have long cancellation rights, so a referral counted in month one may reverse in month four.

Version the definitions. When a definition changes, publish the change date and restate prior periods where possible. The case for reporting dashboard needs fixed definitions is worth reading before you rewrite a partner league table.

Common errors in this area are covered in measurement mistakes explained.

Check six: platform integration and versioning

Referral marketing software rarely stands alone. It feeds a CRM, a data warehouse and often a finance system, and each connection is a place where attribution can break.

Map every integration and record its direction, frequency and owner. A nightly batch job that fails silently for a week will distort a monthly partner payout.

Use versioned APIs where the provider offers them. HubSpot's API reference shows how versioned endpoints and change logs are documented, which is the standard to expect from any CRM integration carrying referral attribution data.

Pin versions and test upgrades in a sandbox before they reach production. A minor platform release can change how a referral link is tagged.

Log the payload, not just the success message. When a dispute arises, the payload shows what was actually sent.

Check seven: escalation routes when attribution is disputed

Disputes are normal in referral programmes. What matters is that the route from query to decision is short, documented and fair.

Set a first-line owner, usually the programme manager, with a response target measured in working days. Partners should not have to escalate to a director to get a straight answer.

Define the evidence pack: click records, consent records, model configuration, integration logs and the published rules. Without the pack, every dispute becomes a negotiation.

Set a second-line reviewer who was not involved in the original decision. In Leeds that role usually sits in compliance or finance, and the city's accountancy and legal firms can supply an outside reviewer for partner disputes.

Publish the outcome and the reason. Partners accept adverse decisions far better when the reasoning is visible and consistent.

Review disputes quarterly for patterns. Three disputes about the same partner or the same model usually point to a rules problem, not a partner problem.

Before you commit to a rebuild, read this guide to measurement and reporting.

Common questions