Nov 18, 2026
A side lesson from The Art of the Revisit. Five minutes, more if you experiment.

Arnaud Blanchard
Vice President of Product
Here is a question I like to ask in interviews. You have two ads and two people. Do you show one ad to each of them, or both ads to the same person? People who think in reach answer fast: two people, once each. People who think in frequency answer just as fast: one person, twice. Before reading on, pick one.
Sam and Alex both watch TV every evening. Your campaign runs Monday to Friday, so that's when your two ads can air. What counts is how much they still remember at a checkpoint after the campaign: the day of the sale, the vote, the launch.
Drag the two TVs onto either timeline, and pick a checkpoint. Each person's percentage at the checkpoint is their memory score: the chance the ad would come to mind if you asked them that day. Your score is the two added together, the total memory: how many people you can expect to remember, from 0 to 2.
Memory works exactly as in The Art of the Revisit: it fades by half every day after a first ad, and a second ad slows the fading down, most of all when it lands in the sweet spot.
Check on Saturday, and two people once wins. Put both ads on Friday: a day later, each of them still has half the memory. That's 0.50 + 0.50 = 1.00. Sam twice, on Wednesday and Friday, only gets 0.75.
Now move the checkpoint to Sunday. The order flips: Sam twice keeps 0.57, two people once keeps 0.50. By the following Wednesday it's 0.25 against 0.06, four times as much.
Two things are going on. A first ad gives someone a memory that fades fast. A second ad, after a gap, makes the memory sturdier. Two fresh memories are worth more right away. One sturdy memory is worth more later.
Try putting both of Sam's ads on the same evening. That loses at every checkpoint: the second ad lands while Sam still remembers the first, so it adds almost nothing.
Adding memory scores can feel odd. Isn't one person who really remembers better than two who might? Picture 100 people like Sam and 100 like Alex, and count heads.
Two people at 30% each beat one person at 50%: out of 200 people, 60 remember instead of 50. A strong memory in one head doesn't make up for an empty one next to it. It's the same total memory The Art of the Revisit uses to compare whole campaigns.
So one person twice only wins when the second ad adds more than the first one did, which means the repeated memory has to be more than twice as strong as each single one. A badly timed second ad never gets there. A well-spaced one can, once the single memories have had time to fade. That's the flip you saw in Fig. 1.
Here is the same race with every possible checkpoint at once. Each line is the best total memory you can get with that strategy, placing the ads as well as the campaign allows.
For a typical ad, one person twice takes the lead about 1.7 days after your last ad. That's the moment Sam's spaced memory becomes more than twice as strong as each fresh one. Move the memorability slider. The better the ad, the later the flip: each single memory lasts longer on its own, so reaching two people stays ahead for longer. Better creative is a reason to spread, not to repeat.
So far, Sam and Alex watch every evening. Real viewers don't. Say Alex only watches on Monday and Tuesday. If you want to reach Alex at all, his ad has to air early in the week.
Now one person twice wins even on Saturday, 0.75 against 0.56. Alex's ad airs on Tuesday and has mostly faded by the weekend. A new person is only worth as much as the timing you can get on them.
That doesn't mean you should give up on light viewers. It means a bidder has to know when each household watches, and weigh a first ad that lands early against a second ad that lands well.
So far, neither of them had heard of you. Now say Alex already knows your brand, and both watch every evening again.
The timelines still track the same thing: whether this campaign's ad comes to mind. That stays at zero until he sees it. But a new ad from a brand he knows has something to hook onto, so it sticks longer. Ads for familiar brands hold up better against competing ads (Kent & Allen, 1994). Here, a half-life of four days instead of one.
Two people once still wins on Saturday, 1.34 against 0.93, and now holds through Sunday. The flip moves to Monday.
When you do repeat, repeat on Alex, and further apart: Monday and Friday. His memory takes longer to fade into the sweet spot. By Wednesday, Alex twice keeps 0.68. Sam twice keeps 0.25. A second ad pays more on someone who already knows you.
This counts memory, not customers. If you need new buyers, Sam is still the one you have to win.
Two people once, when the checkpoint comes soon after your last ad. One person twice, well spaced, when people need to remember for days.
Research on advertising shows the same split, depending on what it measures. Field experiments that count a response right after the ad, like a click or a sign-up, find that the first ad tends to do the most, which favors reaching more people (Lewis, 2015). Memory research finds that spacing repetitions out makes memories last much longer (Cepeda et al., 2008), and the same holds for ads (Appleton-Knapp, Bjork & Wickens, 2005). Both are right. They look at different moments.
Two Google researchers recently ran the same test as Fig. 2 on brand lift data. A second ad was worth 1.5 to 1.8 times the first, short of the 2 it takes to beat reaching someone new (Perez and Martín Calvo, 2026). In this model, a well-spaced second ad is worth 1.5 times the first the day after the campaign, right in their range. Two days after, it's worth 2.3 times.
With two people, you can work it out by hand. With a few hundred thousand households, a budget and a deadline, it's the same trade made millions of times: reach someone new, or come back to someone at the right moment. A bidder makes that call for every household, every evening.
Reach or repeat is a timing question.
The full course is The Art of the Revisit, and this side lesson sits alongside lesson six.
Same teaching model as The Art of the Revisit, not Vibe's production bidder, and its numbers are illustrations rather than forecasts. A first ad gives a memory with a half-life of one day for a typical ad. Each later ad resets memory to full and multiplies the half-life by up to 2.5, depending on how much was left when it landed: almost nothing if the memory was still fresh, the most around one third, about a third more if it was already gone. Ads air in the evening. Total memory is the sum of the two memories at the checkpoint evening.
For someone who already knows the brand, the first half-life is four days instead of one. That's an assumption for illustration: forgetting curves were measured on new material, and a new message from a familiar brand has something to attach to. The score is still memory of the campaign, not of the brand, so it starts at zero.