MARK RITSON: HERE’S A RADICAL IDEA – STOP CALLING IT REACH IF NOBODY LOOKED

Mark Ritson has spent years accepting the gospel of maximum reach and minimum frequency. New attention data has given him a reason to question one of media planning’s favorite rules.

In June 2019, Marc Pritchard went to Cannes and said out loud what most of the industry already believed. “Excess frequency is the biggest waste [in advertising].”

As chief brand officer of Procter & Gamble, he was the most important marketer on the planet. P&G was going to stop hitting the same people again and again and spend the money instead on reaching new ones. “What’s important is how many people we’re reaching.”

His was the most important confirmation of a long-running trope in marketing: that reach was more important than frequency.

John Philip Jones had shown that back in 1995, and shocked a generation of marketers brought up on the Krugman school of multiple repetitions. Jones showed that a single exposure in the week before purchase did most of the selling, and that the second and third added very little impact, despite what they managed to add to the budget.

Erwin Ephron then turned it into recency planning. And the Great Byron Sharp made reach the closest thing media planning has to a commandment, partly because the Ehrenberg-Bass data backed it, partly because reach was cheaper than frequency and partly because it gave every planner on earth a simple rule they could put on a slide and never think about again.

Thirty years on, reach of one is the default setting. Every agency deck has the slide. Every platform sells you a reach objective with a frequency cap bolted on. And every brand manager knows, or thinks they know, that the second exposure is an indulgence and the third borderline criminal. And, on the whole, the rule has served marketers well.

Or we thought it did.

Last week, Lumen, working with Amazon Ads, took 270 campaigns from six countries, each with a brand lift study attached and almost 200,000 survey respondents. It knew which ads each of those people had been served, and it ran each one through its attention models to estimate how many seconds they actually spent looking at it.

I remember eye-tracking experiments in the 1980s. In fusty old textbooks, they would show you the data from eye tracking, and you’d be impressed. Until you saw the giant fucking crash helmet with a video camera taped to the top of it that subjects had to wear while looking, “totally naturally,” at a series of print ads. We have come a long way. And Lumen is now able to track where people actually look in almost perfectly natural conditions.

Frequency, the number of times someone was served an ad, explained almost none of the movement in brand metrics. The total seconds a person spent looking at a campaign, what Lumen calls cumulative attention, explained eight times as much. Two of the campaigns in the data, for example, both reached their audience five times on average.

On every media plan in the world, those two are identical. But, they are very far from identical. One campaign works, and the other is diabolical.

One delivered nine seconds of attention per person, while the other managed 71. Apples and very large oranges.

The reason for the fruity difference is the exchange rate. A Prime Video ad on a television in someone’s front room gets looked at 92% of the time for nearly 18 seconds. Put the same ad into an open internet video player and the chance anyone looks at it falls to 42%, for about three and a half seconds.

Do the arithmetic, and one TV impression buys around 16 seconds of attention. The open internet version buys a second and a half.

Same ad.

Same person.

Same line on the plan, at an exchange rate of roughly eleven to one that few in our industry ever bother to check.

And that is where the reach-of-one theory comes unstuck. When Jones counted an exposure in 1995, it was a bloke on a sofa watching a series of 30-second spots in between Seinfeld. When a digital planner counts one today, it is a pre-roll playing in the corner of a news site while the reader hunts for the little X to shut it down. We ignored the changes in media context, kept our sacred rules and quietly debased the whole point of media planning. Call it phantom reach: the people we count as reached who never really saw anything at all. Or saw so little they might as well have not seen it.

That last line is important because Lumen also worked out how much attention it takes to move the needle. The median campaign needed about 21 seconds to shift awareness and 47 to shift purchase intent. Based on the data, on the open internet you would need around 14 impressions per person to get there, and north of 30 before anyone thinks about buying anything.

Never mind a frequency of one, how do you like a frequency of 30?

It gets worse.

Hand a platform a reach objective, and its algorithm goes shopping for the cheapest unique users it can find. And cheap impressions are cheap for a very good reason. Nobody is looking at the fucking things. Programmatic – the big black box of turds and spiders – has not just automated the hunt for phantom reach, it is positively predisposed to deliver it and then send you a lovely report congratulating you on how efficient you have been.

Grace Kite and Tom Roach have long made the case for “lots of littles,” plenty of cheap exposures that add up. Uncle Les Binet and Will Davis prefer to go big or go home. I am a card-carrying Bothist, and it turns out that the data is, too. Both approaches can work if the seconds add up. Combining them works better still: at the same level of attention, Prime Video plus the open internet delivered 3.8 times the brand lift of the open internet on its own.

And the order matters, which I did not expect. Running the big TV spots before the littles produced 30% more lift than the other way around. That’s a genuinely new insight. Write it down. It will be months, years, before another one comes along.

Sure, there are caveats, and to its great credit, Lumen lists most of them itself. Yes, Amazon co-authored the paper and, shock horror, Amazon’s TV product comes out of it smelling of roses. Yes, the data is correlational and skewed toward big brands, and eight times better than frequency still only explains 8% of brand lift, with the creative, which matters more than anything, nowhere in the model. Lumen also measured brand lift through surveys rather than sales, and it never tested whether reaching more people beats more attention per person. Plenty to pick up on in the future here.

But advertising effectiveness is a swamp of noise, and finding a plannable variable eight times more useful than the one the entire industry plans against is not nothing.

It also sits very comfortably next toKaren Nelson-Field’s decade of work on attention thresholds. Mike Follett and his team at Lumen have been making this argument patiently, often to a half-empty room, for years. I think this is the paper that proves them right. And makes them more important.

So what do you do with all this on Monday morning? Get your agency to translate the plan from impressions into seconds, and if they cannot, demand it of them. Why not? Set a seconds target for each objective, about 20 for awareness and closer to 50 if you would like someone to buy something. And I am sure you would.

And the easy, important one. Schedule your bigs up front and your littles later. Stop paying for impressions that nobody looks at, which will probably mean stopping paying for quite a lot of them. Buy less, for more, for better impact.

None of this kills reach. The Dark Lord of Penetration is still right that you cannot grow a brand among people you never reach. But reach of one was always shorthand for one proper exposure to one actual human being, and somewhere between the sofa and the smartphone and the gigantic flashing array of aimless, empty, unseen ads, we forgot the second half of that sentence.

Source: thedrum.com