PensionBee runs a high volume of marketing across the UK and the US. Like most regulated teams, they used to put all of it through the same compliance review, which meant a routine blog post sat in the same queue as a technically complex piece about pension law. The change that freed up their team was simple to describe. They sorted their content by risk, and let the low-risk side approve itself once the system had cleared it. Here is how they do it, and why it stands up to a regulator.
Why one queue for everything slows a team down
A missing risk warning on a paid ad is a serious problem. A typo in an evergreen blog is a small one. When both go through the same review, the senior reviewer becomes the rate limit on the whole marketing function, and most of their time goes on content that was always going to pass. PensionBee's insight was to decide, up front, which content genuinely needs human judgement and which only needs a reliable check.
How PensionBee splits its content
PensionBee's team draws the line between direct-to-consumer content and technical business content, and their own data backs the line up. Across their consumer-facing material the error rate is very low. Out of 105 blogs, three needed changes. Out of 107 website pieces, two did. That is the profile of content that suits self-approval, because the system clears it accurately and the exceptions are rare and easy to catch.
The technical end is a different animal. Content dealing with pension legislation, employer obligations and the finer points of US retirement rules is where the genuine risk sits, because the mistakes are subtle and expensive. One piece described an employer as being "required" to do something the law actually makes optional. On a quick read it looks fine. It is wrong in a way only someone who knows the rule would catch, which is why PensionBee keeps a human on that category.
How the model runs day to day
Each asset is classified as consumer or technical when it's uploaded, with a tag. Consumer content becomes eligible for self-approval the moment the system clears it, and the compliance lead runs a monthly spot-check across everything that was auto-approved. Technical content, anything law-specific, and anything citing a case or a rule stays with a human reviewer every time. The marketing team moves fast on the bulk of the work, and compliance spends its attention where attention actually changes the outcome.
The spot-check is the part that makes the whole thing hold. It keeps the compliance lead seeing a sample, accountable, and able to pull a content type back into full review if the error rate moves.
The safety net for technical content
The real risk in technical content is definitive language, words like required, mandatory, obligated and guaranteed. Those are the words that turn a helpful explanation into a misstatement of the law. On the high-risk category, the system flags absolute language and points to the rule that governs it, so the reviewer's eye goes straight to the sentence most likely to be wrong. It is the same idea as checking a performance claim against the SEC marketing rule, applied to legal and regulatory assertions. The low-risk content approves itself, and the high-risk content gets a sharper human review than it had before.
Why it stands up to a regulator
Self-approval sounds like reduced control until you look at what sits underneath PensionBee's version of it. A content type earns a self-approval path only after a track record of clean clearances. A human still samples it every month. Every decision, by the system or by a person, lands in an audit trail. That combination, a documented threshold, ongoing sampling and a complete record, is the practical shape of taking reasonable steps, and it is a stronger position than a single overloaded reviewer waving things through under time pressure.
Running it on one platform
PensionBee runs this on Adclear across both its UK and US content. Teams tag content by type on upload, low-risk content clears for self-approval once it passes, and technical and law-specific content routes to a human every time, with the whole thing recorded for evidencing. The fact-checking on definitive language and regulatory claims runs across the high-risk category, and the monthly spot-check arrives as a report rather than a manual trawl.
FAQ
Is self-approval allowed under the FCA regime? There is no rule against marketers approving low-risk content, provided the firm can show a controlled, evidenced process with appropriate oversight. The risk lies in doing it without a track record, a sampling check or a record. The principle itself is fine.
Which content is safe to self-approve? Repeatable, consumer-facing formats with a low historical error rate, like standard blogs and website copy. Technical, law-specific or claim-heavy content should keep a human reviewer, which is the line PensionBee draws.
Does self-approval just move the risk? Only if you apply it to the wrong content. Done the way PensionBee does it, it concentrates human attention on the high-risk material and lifts it off the low-risk material, which lowers overall risk rather than shuffling it.
How do you prove it to a regulator? A documented threshold before a content type is cleared for self-approval, a monthly spot-check, and a complete audit trail of every decision.
Once you're booked in, we'll send you a free playbook on Financial promotions compliance for FinTechs.
