The Best Advertising Is Invisible (Elasticity Zero, Explained)

Here is a number that should end a thousand budget meetings: the typical short-term sales elasticity of broad-reach advertising is 0.0008.

Which, as Byron Sharp cheerfully rounds it, is zero. You move the spend up, you move it down - and weekly sales do nothing you can see.

Sharp dropped this in his Uncensored CMO conversation with Mark Ritson, recorded over champagne right after their joint Cannes session (we wrote up the five things they agreed on stage - this podcast is where the numbers behind the truce live). And before you cancel your brand budget: the zero is not a verdict. It's a measurement lesson most of the industry keeps failing.

Source: Uncensored CMO - Byron Sharp vs Mark Ritson, July 2026

Why zero doesn't mean broken

The logic runs through the 95/5 rule - coined, incidentally, as a "flippant" B2B shorthand by Ehrenberg-Bass's John Dawes ("a bullshit number, but a good bullshit number," says Ritson). At any moment, roughly 95% of your category's buyers are not buying. Run a big reach campaign this week and even among the people who do buy this week, most never saw it - Sharp points out a decent campaign reaches maybe 5% of the market in a week.

So the sales signal of your best advertising is mathematically invisible. It's working immediately - planting memories - but the cash register only rings as each new 5% rotates into the market, week after week, month after month. The effect isn't delayed. It's dispersed.

Meanwhile search and performance elasticities run 10 to 20 times higher. Not because performance marketing is 10-20x better at creating demand, but because it stands at the exit collecting the 5% who were leaving with a purchase anyway. Someone typing "new washing machine" is, definitionally, in the five.

Stop saying "long-term"

This is why both men - who agree on little else so enthusiastically - want the phrase "long-term effects" banned from CFO conversations. To a finance director, "long-term" means "I can't ever show you," or worse, "nothing will happen for years." Both readings are wrong.

The honest framing: brand advertising has lasting, spread-out effects. It works now; it just pays out across the year as buyers arrive in the market. Sharp's practical instruction is the sentence to steal: you can't see it in your own sales, so go to the 95 who aren't in the market and look at their memories. Awareness, mental availability, category entry points - the leading indicators that predict what the register will do when those people finally show up.

And the direction of causality only runs one way. Per System1's data, discussed in the episode: the best brand-building campaigns also correlate with short-term sales spikes - but heavily targeted promotional campaigns that spike sales do nothing for the brand. Long drives short. Short never drives long.

The Lucozade scar

Ritson tells the story on himself. Managing Lucozade through the UK sugar-tax reformulation ("we made the mistake of telling people we're changing it" - people didn't drop out because of the taste, they dropped out because they heard it had changed), he asked his team for penetration data. Twelve months: 26% of the market had bought.

Then he changed one word in the question. Three-year penetration: 46%.

Nearly half the brand's actual buyers don't show up in a one-year window - which is typical, penetration roughly doubles over three years for consumer brands. The team had been handed an eight-week relaunch objective. Ritson's reaction: "This needs to be eight years. Are you freaking kidding?"

And here is the industry punchline, from fresh Ipsos data Ritson cites across the UK, US, Canada and Australia: only around 45% of marketers can define penetration at all. "We can have a very big argument about penetration. But first we're going to have to explain to half the people in the room what we mean."

What to do with this on Monday

There is something almost liberating in the zero. It means the weekly dashboard was never going to vindicate your best work, no matter how good it was - so you can stop torturing it for proof it cannot give. The proof lives in other people's heads, months before it lives in your revenue. Measure it there, or watch someone with a better-researched view of the category quietly out-wait you.