Building the sign-off gauntlet: legal, compliance and actuarial review

"Fifteen minutes could save you fifteen percent or more on car insurance." Geico, which sells the car insurance that sentence is about, has run some version of it for the better part of two decades, and every word in it is carrying weight. Now assume your team wants the same shape of claim: one headline savings percentage, running as a paid social headline, a radio read and a landing pagelanding pageA standalone web page built for a single campaign goal, designed to maximise conversions by removing distractions and focusing visitors on one action.View full definition → hero. Pricing sends you the number on a Tuesday. It does not ship until legal, compliance and actuarial have each signed something, and none of them is signing the same thing. This lesson follows that one number through all three desks.

What the number actually asserts

Break the Geico line apart and it makes four claims at once: a time claim (fifteen minutes to get a quote), a savings claim (fifteen percent), an implied population claim (people like the person watching), and a direction claim. That last one is the interesting part. Most savings advertising says "up to", which sets a ceiling: the Advertising Standards Authority expects a "up to" figure to be achievable by a meaningful proportion of the audience, not by a lucky handful. "Or more" sets a floor instead. A floor is a harder thing to evidence, because the average outcome in your data now has to sit at or above the number rather than at the top of the range.

"Could" is what makes the floor survivable. It converts a promise into a possibility, and it is the single word your legal reviewer will refuse to let creative drop.

The three desks, on this one claim

Legal reads the sentence as a document, not a message. Does it create anything a court or a regulator could treat as a promise beyond the policy? A conditional savings figure with an audible qualifier does not. "Save fifteen percent" in the imperative, with no qualifier, edges towards one. Legal also polices the geography: US motor rates are filed and approved state by state (in the US via System for Electronic Rates & Forms Filing, SERFF), so a nationwide savings average is an aggregate across dozens of separately approved rate structures, and a state where the number does not hold is a state where the ad should not run.

Compliance is not re-litigating whether a paid social headline counts as a promotion; the foundations lesson settles that. Its work here is matching the approved wording to each channel, checking the claim against the fair-treatment and fair-value tests the brief-writing lesson turns into a method, and keeping the evidence retrievable. In the US, state unfair trade practices acts modelled on the National Association of Insurance Commissioners, NAIC model law prohibit misleading comparisons; in the UK, the ASA can ask for the substantiation of a savings claim and the FCA expects the firm to be able to show its promotions were clear, fair and not misleading at the time they ran.

Actuarial owns the population and the arithmetic, and this is where the number is actually made or lost. The actuary's first question is never "is fifteen percent right?" It is "fifteen percent of what, measured on whom, over which window?"

The arithmetic that decides the wording

Work an illustrative extract, one rolling twelve months of new business. Of every 100 people who complete a quote, suppose 20 buy. Of those 20, 13 tell you they are paying less than they were with their previous insurer, and their average reduction is 22 percent, with a median of 16. The other seven pay the same or more.

Three different honest claims come out of that one table:

• Based on the 13, the average saving is 22 percent. The claim is true only of "customers who switched and saved", which is close to circular, and the qualifier has to be visible for the claim to stand.
• Based on all 20 buyers, the average drops to roughly 11 percent. Now "fifteen percent or more" is dead and "up to" is your only option.
• Based on all 100 quoters, most of whom walked away because you were more expensive, the average saving is close to nothing.

That is the arbitration a marketer actually faces: the number does not change, the base does. If pricing hands you 11 percent and the brief wants 15, the legitimate move is to narrow the population and say so in the claim, not to widen the number. A second-order cost follows. Every narrowing you add ("customers who switched and saved", "new customers in the last twelve months") is a qualifier that must survive every cut-down, which is the thing creative will fight hardest.

One more thing the actuary will flag: the prior premium is usually self-reported by the customer at quote. You are substantiating a savings claim with a number the customer typed in. Ask how the sample is cleaned of implausible entries before you cite it, because "the customer told us" is a thin answer to a regulator who asks how the comparison was verified.

How a substantiated claim goes bad anyway

It ages. A savings figure signed off in Q1 assumes the rate book that existed in Q1. Take a portfolio rate increase mid-year and the same claim, still running on the same landing page, is now unsupported by data that has not been re-run. Sign-offs on numeric claims need an expiry date and a named owner for the re-run, or the ad outlives its evidence.

The qualifier gets amputated. What actuarial approved was "15 percent or more for customers who switched and saved". What ships in a 40-character headline is "Save 15%". The substantiated claim and the published claim are now different claims, and only one of them has a file behind it.

It crosses a border. A US extract cannot substantiate a UK ad; the ASA wants evidence drawn from the market and the period the advertising runs in. The switching logic itself also differs. Since 1 January 2022, FCA rules require home and motor renewal prices to be no higher than the equivalent new-business price, which cuts into the automatic "switch and save" premise that a rate-walking market used to guarantee. A claim built on British data has to reflect that; a claim imported from elsewhere will not.

Rebuilding the claim so it clears first pass

• Choose the population before you ask for the number. "Average saving for new motor customers who reported a lower prior premium, twelve months to 30 June" is a request an actuary can answer in a day. "What's our savings stat?" starts a week.
• Write the qualifier as part of the claim, then design to it. If the qualifier will not fit the format, the format cannot carry the claim; use a non-numeric line there instead.
• Agree the fallback wording at brief stage. If the re-run comes back at 11 percent, you already know whether the campaign says "up to 15" , switches to a different metric, or drops the number.
• Give the claim an expiry date and put it in the same place as the media end date, so nobody has to remember.

🎬 [VIDEO: "How Insurance Advertising Compliance Works" - youtube.com - search for NAIC or FCA explainer content on insurance marketing compliance and consumer protection rules - illustrates the filing-to-launch pipelinepipelineAll active sales opportunities across the stages of the sales process, together with their combined potential value and probability of closing.View full definition → in practical terms]

The substantiation file, item by item

If the ASA or a state regulator asks about your savings claim, you are answering in days rather than weeks, from whatever is already in the folder. For this one claim, that means:

1. The data extract, with the exact date range, the definition of the base population, and the count in it.
2. A methodology note in plain language: how prior premium was captured, how implausible values were handled, mean versus median, what was excluded and why.
3. The approved wording, per channel, including the qualifier as it appears in each cut-down and the voiceover script where the qualifier is spoken rather than shown.
4. The sign-off trail: who in legal, compliance and actuarial approved what, on which date, against which version of the extract.
5. The expiry date and the name of the person who re-runs the number, plus the rate changes that would force an early re-run.

The file is the deliverable, not a byproduct of one. An accurate claim with no retrievable evidence behind it is, in practice, an unsubstantiated claim, because the only version of the truth a regulator can examine after the fact is the one you wrote down.

Key Takeaways

• A headline savings claim is three claims stacked: a number, a population and a modal verb. "Could", "up to" and "or more" are the parts that decide whether your data can support it, and "or more" is the hardest to evidence because it sets a floor.
• Actuarial review argues about the base, not the number. The same table yields 22 percent, 11 percent or nothing depending on who counts as a customer, so pick the population before you request the figure.
• The honest way to reachreachThe number of unique people exposed to your message in a given period. Unlike impressions, reach counts each person once, no matter how often they see it.View full definition → a bigger percentage is a narrower population plus a visible qualifier, and every qualifier is something creative will try to cut in the short formats.
• Numeric claims decay. Give each one an expiry date, a re-run owner and a pre-agreed fallback line, or a mid-year rate change will leave live advertising standing on stale evidence.
• Evidence that cannot be produced quickly is not evidence: keep the extract, the methodology note, the per-channel wording and the dated sign-off trail together in one place.