The method
The Fielderz method
The symptoms
What’s happening in the account
We scaled the Winner, and it crashed.
It’s called the winner’s curse, and it’s simple statistics. Take 20 completely identical ads, with a true CPA of ₪200, and give each a budget for about 10 conversions. Through noise alone, one of them will show up in the report at ₪125. Give it more budget and it goes back to 200. Then comes the expensive part: everyone looks for a reason why that one won, and next month’s brief is written on luck. That’s why we test fewer ideas, ones that are truly different from each other, with enough data for each.
20 identical ads, with a true CPA of ₪200, scattered around a horizontal line at ₪200. One of them, the winner, lights up at ₪125.
We gave it three days. It didn’t deliver. We shut it.
A new ad doesn’t start from zero, it starts below it. At the first exposure the brain asks “what is this?”, at the second “what does it mean to me?”, and only at the third does a decision come. The researcher Herbert Krugman described it as early as 1972, and studies of repeated exposure have shown again and again that people like what they already know. A new message, and a brand message above all, needs those repetitions, and a “three days and we shut it” rule kills exactly that. With us every ad goes live with a definition set in advance: when it should start to work, and how we’ll know.
Two performance curves over 10 days. Ad A jumps on day one and fades. Ad B starts weak, and after people have seen it three times it overtakes. A vertical line on day 3: this is where most accounts shut an ad.
A frequency of 2.1. So why is the CTR falling?
An average of 2.1 sounds safe, but impressions aren’t shared out evenly. In a typical distribution, about half the audience saw the ad once, and about 4% saw it six times or more. Those 4% alone got about 15% of the impressions. And the algorithm keeps showing it to exactly the people most likely to buy. In the brain this is called habituation: a familiar stimulus stops being processed. People don’t hate the ad, they simply don’t see it any more. That’s why we don’t wait for the average to rise. We replace by what happens at the edge.
A distribution in bars: how many people saw the ad once, twice, and so on up to 10 times or more. About half saw it once. A dashed line on the average, 2.1. The bars for 6 times or more are coloured: here the ad no longer works.
A great hook rate, and no sales.
A Facebook study of the mobile feed found that people remember content they saw for as little as a quarter of a second. The brain tags things very fast, but only what it actually sees. If the product and the brand appear at second 8, most of the audience is already gone, and the great hook rate only counts how many people started watching an ad that sold them nothing. Across the 100,000 videos we’ve produced, [X]% of viewers left before the third second [to approve]. That’s why with us the brand and the product come into the first seconds, even when the story hasn’t started yet.
A video’s retention curve, second by second, falling sharply. At second 8 the product appears, with 15% of viewers left. The area under the curve up to second 8 is shaded: you paid for all of these.
The approach
How the customer decides
How involved is the customer, and what decides it for them?
Every purchase decision comes down to two things: how involved the customer is in it, and what decides it for them in the end, information or feeling.
In a high-involvement decision – a flight, insurance, a system for the whole company – the customer isn’t looking for a reason to buy. They’re looking for a reason not to get it wrong. Psychologists call it regret aversion: the fear of choosing wrong is stronger than the wish to choose well. An ad that adds one more advantage doesn’t help them; an ad that lowers the risk does. In a low-involvement decision it’s the other way round. The brain runs on shortcuts, and an explanation only slows it down.
Example: In home insurance, the number worth checking first is how many people reached the second where the comparison appears, not how many stopped in the first second.
A 2 × 2 map: high involvement against low involvement, information decides against feeling decides. For each square: what the ad must show, and which number is worth checking first. Proof: comparison and specs, and the number is how many reached the second with the proof. Identity: a world, and people like me, and the number is searches for the brand name, and LTV. Ease: easy, available, good value, and the number is CVR. Impulse: the moment itself, and the number is Hook rate.
How ready are they?
When you run “buy now” on a cold audience, you pay for everyone and actually talk only to the few who are already ready.
At any moment, only a minority of the audience is in the market at all. For B2B and expensive products, studies put it at around 5% a quarter. Every competitor fights over the same 5% with the same message. The rest are the customers of the coming quarters, and many of them will prefer whoever they got to know first.
An example from tech: A lead who got to know the problem from an ad two months earlier comes to the call more ready, and their sales cycle gets shorter.
100 dots, 5 of them marked “in the market now”, and every competitor is aiming at them. Next quarter, 5 others light up, and three of them are already in your color: the ones you talked to before.
When do they think of you?
Whoever thinks of you before they search costs less when they search. That’s next year’s CAC, and it’s set today.
The customer doesn’t compare every product on the market. They compare the three or four they remember when the need comes up. Researchers call it the consideration set. Whoever isn’t in it doesn’t lose the comparison – they simply aren’t part of it. Most brands advertise for the moment of search, where there’s already an auction and a comparison. The category also comes to mind at other moments: when the product runs out, when guests are coming, when a friend recommends it. That’s where it’s decided who gets into the set.
A week in Noa’s life, Sunday to Saturday. Four moments when the category comes to her mind: when the product runs out, at the search, when guests are coming, when a friend recommends it. A bar along the bottom shows that every competitor advertises only at the moment of search.
Data first
The readings in the console
| The two numbers | What it means | What changes in the ad |
|---|---|---|
| High Hook rate · low Hold rate | The hook promises something the body doesn’t deliver | Replace the body, not the hook |
| High CTR · low CVR | The message attracts the curious | A message that filters, even if CTR drops |
| Leads up · SQLs flat | The ad brings interest in the product, not need | Talk about the problem, not the product |
| CPM up · CTR steady | Competitors came in, and the ad isn’t good enough for the auction to prefer it | An ad people respond to is cheaper to show |
| ROAS up · MER flat | The platform takes credit for purchases that would have happened anyway | Messages for people who don’t know you yet |
| Good CAC · the budget doesn’t grow | The ready audience is small, and it’s already been bought | Messages for earlier stages |
| CAC down · LTV down | Buying cheap customers who don’t come back | A different entry message, not a discount |
The paradox.
Overall, message A is almost 3× cheaper than message B.
By campaign, message B wins in both prospecting and retargeting. The reason: message A ran mostly in retargeting, where everything is cheap.
Whoever reads by campaign will scale the wrong message. That’s why every ad we make is tagged by what it says, and the numbers are read by the tags.
Hook rate doesn’t only measure the hook. It also measures who saw it. The same hook will score far higher in retargeting than with a cold audience.
A bar chart of CPA, what a conversion costs. Overall: message A 109, message B 286. By campaign: in prospecting, message A 500 and message B 400; in retargeting, message A 100 and message B 80.
| Spend | Conversions | CPA | |
|---|---|---|---|
| Prospecting · message A | 10K | 20 | 500 |
| Prospecting · message B | 90K | 225 | 400 |
| Retargeting · message A | 90K | 900 | 100 |
| Retargeting · message B | 10K | 125 | 80 |
| Overall: | a CPA of 109 for message A, and 286 for message B. | ||
Want us to read your account like this?
Book a meetingFrom account to profit
From the account to profit
A 20% discount on a product with a 40% margin means doubling sales just to stay in place.
A message that justifies full price is worth more than any improvement in CPA.
In tech, it shows in Payback: a lead who came in from a “free demo for everyone” message closes less often, and takes longer to pay back its cost. A lead who came in from a message about the problem gets to the call already knowing why they’re there.
Discount calculator
To make the same profit you need to sell ×
From the work
All the workLet’s talk
One meeting. Your account on the screen. No deck.
You’ll leave with three things: which of the situations is happening in your account, which message you’re missing, and what we’d put up first. Even if you don’t work with us.