Account audit template for Fintech

The account audit with the checks a regulated financial product needs: whether every live ad's claim is covered by a current review, whether delivery has drifted outside authorised markets, whether disclosures still render where they should, and how approval rates differ by campaign.

Everything else is the general audit, unchanged. One row per finding, evidence attached, honest severity, one action, one owner, one date.

Run it quarterly and immediately after any change to a rate, a fee or a term. The second occasion is the important one: a change makes a set of live creative stale at the same moment, and an audit is how the set is found.

Run it also before an external review of any kind, because finding your own problems first is considerably cheaper.

In most categories an audit is a performance exercise. Here it is partly a record: a dated statement of what was running, in which markets, under which reviews, checked by a named person.

That changes how it should be written. The evidence column carries more weight than usual, and the evidence itself should be durable and shared — lists of live ads with their references, screenshots of delivery settings, the rendered ad on each placement, the landing page as it appeared.

It also changes the cadence. A quarterly audit is the baseline, and any change to a rate, a fee or a term triggers one immediately, because that is the moment a set of correct creative becomes incorrect without anybody touching it.

The first check is mechanical: list every ad that is live, with the review reference it carries, and compare each against the review record and its date.

Three findings recur. Ads with no reference at all, usually duplicated from older campaigns in a hurry. Ads whose reference is genuine but predates the last change to a published term. And references that do not exist in the review record, which is usually a typo and occasionally something worse.

The recommendation for stale creative is replacement rather than editing. Editing an ad in place attaches new wording to an old reference and corrupts exactly the record this audit exists to maintain.

Check the delivery settings of every campaign against the markets the product is authorised in, and keep screenshots.

Market restrictions usually live in a single setting, and settings drift: a campaign is duplicated and the exclusion is not carried over, an audience expansion is enabled, a platform default re-applies on save, somebody widens a radius that crosses a border.

None of that announces itself. The audit is the only routine that looks, and the finding is either clean or urgent — there is rarely a middle case.

Open the live ads on each placement and the landing page on a phone, following a real ad link with its parameters.

Disclosures fail in rendering. Text is truncated by a placement's limit. A summary that sat above the fold in a desktop layout sits below it on a phone. A template drops a footer. An overlay is covered by the platform's own interface in one format.

The design files will look correct in every one of those cases, which is why the check has to be of the rendered thing. Screenshots are the evidence, and they are the only record of how it actually appeared.

The performance finding that matters most here comes from joining advertising data to underwriting data.

Compare, by campaign and where possible by creative, the proportion of applications that were approved and the proportion that were funded. The common finding is a wide spread, with the cheapest-per-application campaign at the bottom. That campaign has been rewarded by every cost-per-application report the business produces.

State the finding as the spread, with the window. The recommendation is usually to change what the account is optimised and reported on rather than to turn a campaign off, because the campaign may be fine at a different weight with different creative.

Check that every landing page a live campaign points at states the current rate, fee and terms, in the cleared wording.

Pages drift independently of ads, and the drift is invisible from the ad account. A rate updated in the product system and on the main site but not on a campaign page produces a correct ad and an incorrect journey, which is the same failure from the customer's side.

Follow the actual ad links rather than opening the pages directly, because redirects and parameters can change which version is served.

A specific check: list every retargeting audience in use and what it is built from.

Audiences assembled from application starts are defined by a financial circumstance, and the creative reaching them should be a reminder rather than a reference to the reader's situation. Audiences built years ago, by somebody who has left, are the ones most likely to be wrong, and they are rarely re-examined because they perform well.

The dull check that matters most when something has gone wrong: who can administer each ad account, page, payment method and integration, and whether any of them depends on one person or on a supplier no longer engaged.

In a regulated context there is a second question attached to it: who can edit a live ad. Where a large number of people can change creative in place, the record of what ran under which review is only as reliable as everybody's discipline, and the audit should say so plainly.

Record the list with the date. It is the finding that costs nothing to produce, is never urgent, and is the first thing anybody wishes they had on the day an account is locked or an ad is changed without explanation.

Seven checks. Every finding separates what was found from what should be done, with durable evidence. Every live ad has been checked against a current review by reference. Delivery settings have been checked against authorised markets, with screenshots. Disclosures have been checked as rendered on each placement and on a phone. Approval and funding rates have been compared across campaigns. Landing pages have been opened through real ad links. And every row has an owner and a date agreed with the person named.