Latest / Sleeping Barber - A Marketing Podcast / SBP 239: The Sharp Cut - Who Killed the Click?

Transcript

SBP 239: The Sharp Cut - Who Killed the Click? · Sleeping Barber - A Marketing Podcast

• 0:00
  SPEAKER_01: Welcome to the Sleeping Baba Podcast.
• 0:03
  SPEAKER_00: All right.
• 0:03
  SPEAKER_00: Let's do this.
• 0:04
  SPEAKER_00: Let's do this.
• 0:05
  SPEAKER_00: V this is the sharp cut.
• 0:07
  SPEAKER_00: You and I talked about this a fair bit, that marketing runs on comfort blankets, and the things we all repeat often is are things we repeat because they're they're warm and they and everybody else is saying them.
• 0:18
  SPEAKER_00: And so they feel good.
• 0:20
  SPEAKER_00: And what we do here is try and cut up those blankets and look what's actually inside.
• 0:25
  SPEAKER_01: Yeah, we're looking for real-world evidence, experiments where they exist, and tell you when they don't.
• 0:30
  SPEAKER_00: These are usually short, one idea, and there's nothing in them that you couldn't afford to a CEO or CEO or your CMO or any of the O's, if you wanted to.
• 0:40
  SPEAKER_01: All the C's.
• 0:41
  SPEAKER_01: Make sure you're sending to all the C's, every single one of them.
• 0:44
  SPEAKER_01: And today it's bit Yeah, and today it's built a little bit different than how we would normally do these.
• 0:50
  SPEAKER_00: Yes.
• 0:50
  SPEAKER_00: Today, in in the spirit of all the murder mysteries that are out there, the murder mystery podcast, we're creating a murder mystery.
• 0:59
  SPEAKER_00: Somebody killed the clique.
• 1:00
  SPEAKER_00: And we're gonna find out you who.
• 1:05
  SPEAKER_01: We've got a victim, a cause of death, a weapon, and six suspects.
• 1:10
  SPEAKER_01: Nobody, however, has confessed.
• 1:12
  SPEAKER_00: And for the next little while, we're not Mark and V.
• 1:15
  SPEAKER_00: We're actually detectives on this one.
• 1:17
  SPEAKER_01: The scissor boys, like the Hardy Boys, but we've got our scissors a wash out.
• 1:22
  SPEAKER_00: And we'll take you through how it was done.
• 1:24
  SPEAKER_00: We won't hand you a verdict because we don't fully agree.
• 1:27
  SPEAKER_00: And ultimately you're gonna end up being the jury, dear listener.
• 1:30
  SPEAKER_01: And here are your six suspects up front.
• 1:32
  SPEAKER_01: So you can start ruling them as we go through this episode.
• 1:36
  SPEAKER_01: The auction, the platforms, and that means all of them, not just Google, last click attribution, the CFO, the agency, and of course, us as marketers.
• 1:45
  SPEAKER_00: So hold on to that list because we'll come back to it at the end.
• 1:48
  SPEAKER_00: But let's start V as every good.
• 1:51
  SPEAKER_00: I watched Reacher on the weekend, and exactly this is how it starts.
• 1:55
  SPEAKER_00: Yeah, there's a there's a body.
• 1:58
  SPEAKER_00: San Jose, California.
• 1:59
  SPEAKER_00: Let me set the scene.
• 2:00
  SPEAKER_00: April 2012.
• 2:01
  SPEAKER_00: eBay is spending somewhere around$51 million a year on paid search in the United States.
• 2:07
  SPEAKER_00: And somebody inside the company decides to find out whether it's actually doing it.
• 2:11
  SPEAKER_01: So they turn it off.
• 2:12
  SPEAKER_01: Not everywhere, across roughly a third of the country.
• 2:16
  SPEAKER_01: 68 markets go dark on all branded keywords for 60 days, and 65 matched markets keep running as the control.
• 2:24
  SPEAKER_01: A real experiment with a real control group on live money.
• 2:29
  SPEAKER_00: Yeah, and they'd done a smaller version of this a month earlier.
• 2:32
  SPEAKER_00: In March, they stopped bidding on the word eBay itself on Yahoo and MSN.
• 2:37
  SPEAKER_00: Remember that.
• 2:38
  SPEAKER_00: And kept paying for it on Google.
• 2:40
  SPEAKER_00: So they'd have something to compare against.
• 2:42
  SPEAKER_00: So 99.5% of the clicks they stopped paying for came straight through organic search.
• 2:47
  SPEAKER_00: People still typed in eBay.
• 2:49
  SPEAKER_00: They didn't see an ad and they clicked on the blue link underneath where the ad would have been.
• 2:53
  SPEAKER_01: They'd been buying traffic that was already theirs, essentially.
• 2:56
  SPEAKER_00: Yeah.
• 2:57
  SPEAKER_00: And then the big one.
• 2:58
  SPEAKER_00: So this is the test that they did.
• 2:59
  SPEAKER_00: So the non-brand 60 days, and this part matters because it is a bit that usually gets dropped when people tell the story.
• 3:06
  SPEAKER_00: The clicks did fall.
• 3:07
  SPEAKER_00: Paid search clicks dropped by 41%, and they didn't come back.
• 3:12
  SPEAKER_01: No, but the sales barely moved when we look at total sales volume.
• 3:16
  SPEAKER_01: New and occasional buyers responded to the ads.
• 3:19
  SPEAKER_01: The regulars didn't, and the regulars were soaking up most of the money.
• 3:22
  SPEAKER_00: So take the same campaign and we scored two ways.
• 3:25
  SPEAKER_00: So measure it the way a dashboard measures it, and people who clicked against people who didn't.
• 3:31
  SPEAKER_00: And the return on that spend is over 4,000%.
• 3:35
  SPEAKER_01: Scored against the control markets, though, which is the only way that the that answers the question, and it's minus 63%.
• 3:43
  SPEAKER_00: Yeah, 4,000% in one hand, or minus 63% in the other hand.
• 3:47
  SPEAKER_00: Same company, same year, same 51 billion dollars.
• 3:51
  SPEAKER_00: And and one of those is a lab result ultimately, and the other one is what the industry has been telling itself for 14 years.
• 3:57
  SPEAKER_01: That's our body.
• 3:58
  SPEAKER_01: The click had been taking credit for sales it didn't cause, and nobody could tell until someone actually just simply turned it all off.
• 4:05
  SPEAKER_01: Before we work out who killed it, we should say what a click you actually was, not what we use it for today, but what used to happen when somebody clicked.
• 4:14
  SPEAKER_00: Yeah, so in 1994, in a converted warehouse in San Francisco V at a site called Hot Wired, which was Wired Magazine's first go-up being online, they ran a banner ad for AT ⁇ T, a rectangle 468 pixels by 60 tall.
• 4:30
  SPEAKER_00: And then somebody clicked it.
• 4:32
  SPEAKER_00: And what happened was that a browser asked a server file.
• 4:36
  SPEAKER_00: The server file sends the file.
• 4:38
  SPEAKER_00: The server writes one line of in a log saying that it did that.
• 4:42
  SPEAKER_00: And you download this click, this file.
• 4:45
  SPEAKER_01: And essentially that was a whole event.
• 4:47
  SPEAKER_01: That is what a click was.
• 4:49
  SPEAKER_00: Yeah, there was no person attached to it, no reason attached to it, no opinion about the ad.
• 4:54
  SPEAKER_00: A machine noted that a file left the building, and it was true, it was accurate.
• 4:58
  SPEAKER_00: Nobody ever lied about that part.
• 5:00
  SPEAKER_01: No, and there was nothing to count them for you.
• 5:03
  SPEAKER_01: Today, a platform will show you your click rate the second you log in.
• 5:07
  SPEAKER_01: In 1994, there was no software that really did that.
• 5:10
  SPEAKER_01: Somebody at the agency sat down with the website's records and counted them one by one, by hand, essentially.
• 5:16
  SPEAKER_01: Digiday, the advertising trade publication, later described the first web analytics tool as a highlighter pen.
• 5:23
  SPEAKER_00: And then there's the number that comes out of the banner.
• 5:26
  SPEAKER_00: So the story goes that 44% of the people who saw that ad clicked on it.
• 5:32
  SPEAKER_00: Can you imagine a 44% clicked through ad.
• 5:36
  SPEAKER_00: And you'll see that presentation uh or see that figure in some presentations today, usually as proof of how good things were before people like us wrote them.
• 5:46
  SPEAKER_00: Same thing with emails, the same numbers, like crazy high numbers like that.
• 5:49
  SPEAKER_00: But nobody's ever actually been able to check it.
• 5:51
  SPEAKER_00: There's no report behind it.
• 5:53
  SPEAKER_00: It comes from people remembering it years later.
• 5:55
  SPEAKER_00: So I don't think anybody made it up.
• 5:57
  SPEAKER_00: Um, but it was in 1994, like a long time ago, and nobody was auditing anything.
• 6:02
  SPEAKER_00: So in the 32 years since, no one's really been able to show where that number actually came from.
• 6:07
  SPEAKER_01: Exactly.
• 6:07
  SPEAKER_01: So that that's our victim.
• 6:09
  SPEAKER_01: And it isn't that 44% because we can't even prove that.
• 6:13
  SPEAKER_01: It is the one thing that was true.
• 6:15
  SPEAKER_01: A file left a building, like you said earlier, small, honest, carrying exactly one fact.
• 6:20
  SPEAKER_01: The coroner needs something for the toe tag.
• 6:23
  SPEAKER_01: So let's call our victim clicky.
• 6:25
  SPEAKER_00: Clicky it is.
• 6:26
  SPEAKER_00: And so the cause of death.
• 6:28
  SPEAKER_00: So now we're at the corner and we want to find out what the cause of death was.
• 6:32
  SPEAKER_00: So it wasn't really one blow.
• 6:34
  SPEAKER_00: Clicky was poisoned.
• 6:36
  SPEAKER_00: Slowly.
• 6:37
  SPEAKER_00: Over 11 years.
• 6:38
  SPEAKER_00: And none of those doses would have killed it on its own, and that's why nobody noticed.
• 6:42
  SPEAKER_00: That's right, the lab results are back.
• 6:44
  SPEAKER_01: So the first dose happened back in 1998.
• 6:46
  SPEAKER_01: Clicky becomes a price.
• 6:48
  SPEAKER_01: A search engine called go2.com starts selling ads by the click instead of by the impression.
• 6:53
  SPEAKER_01: Google moves to the same model in 2002 as well.
• 6:56
  SPEAKER_01: That sounds like an accounting change, but it isn't.
• 6:59
  SPEAKER_01: The moment you pay for something, you assume it's worth something because nobody pays for nothing.
• 7:04
  SPEAKER_01: And that's where the meaning starts to get added to the click.
• 7:08
  SPEAKER_01: Not by research, but by an invoice.
• 7:10
  SPEAKER_01: Exactly.
• 7:10
  SPEAKER_01: So now, second dose.
• 7:12
  SPEAKER_01: 2002, clicky becomes the bouncer.
• 7:14
  SPEAKER_01: Before that, whoever paid the most got the top spot.
• 7:17
  SPEAKER_01: Google changed the rule.
• 7:19
  SPEAKER_01: Now, where your ads showed up depend on two things how much you'd pay and how likely Google thought people were going to click.
• 7:26
  SPEAKER_01: So if Google guessed nobody would click, you got pushed down the page or you didn't even get served at all, no matter how much money you decide to throw at it.
• 7:34
  SPEAKER_00: And that's the one that kind of is troubling.
• 7:36
  SPEAKER_00: Up until then, a click just described what happened.
• 7:39
  SPEAKER_00: And then after that, it decided who got in.
• 7:42
  SPEAKER_00: So whether your ad was even seen at all came down to a number that only ever meant use previously that a file left the building.
• 7:49
  SPEAKER_01: Exactly.
• 7:49
  SPEAKER_01: So now the third dose, 2005.
• 7:52
  SPEAKER_01: Clicky gets the credit.
• 7:53
  SPEAKER_01: Free analytics arrives.
• 7:54
  SPEAKER_01: Last click attribution switched on by default.
• 7:57
  SPEAKER_01: And whichever click happened last gets credited with the sale.
• 8:00
  SPEAKER_00: So in 11 years, the click goes from machine noting a file went out the door to this caused the purchase.
• 8:07
  SPEAKER_00: And there's no study underneath any of that.
• 8:09
  SPEAKER_00: The billing decision, the ranking decision, and a default setting and a free tool.
• 8:14
  SPEAKER_01: And the whole time the dashboard said that the patient was fine.
• 8:17
  SPEAKER_01: Clicks up, cost per click down, return on ads been looking great.
• 8:22
  SPEAKER_01: Not one of those doses was an argument about how advertising actually works.
• 8:26
  SPEAKER_01: And it took eBay switching the ads off to run the blood test.
• 8:30
  SPEAKER_00: And when we look at it, what was in the bottle and the poison, it actually was meaning.
• 8:35
  SPEAKER_00: Every dose was somebody adding something to the click that it was never able to carry.
• 8:40
  SPEAKER_00: A price, then power over who gets seen, and then credit for the sale.
• 8:43
  SPEAKER_01: And it kept working because an auction needs one number to sort bids by.
• 8:47
  SPEAKER_01: Whatever the number becomes, the whole thing is what the market then starts chasing.
• 8:52
  SPEAKER_01: And once everybody's chasing it, it stops describing really anything.
• 8:57
  SPEAKER_01: There's a name for that.
• 8:58
  SPEAKER_01: This is not the first time we've talked about Goodhart's Law.
• 9:01
  SPEAKER_01: When a measure becomes a target, it stops becoming a good measure.
• 9:04
  SPEAKER_00: Yeah, that's a great, great line.
• 9:06
  SPEAKER_00: And that's the whole case ultimately in one sentence.
• 9:09
  SPEAKER_01: All right.
• 9:09
  SPEAKER_01: Now it's time to get into who did it.
• 9:12
  SPEAKER_01: Who done it?
• 9:13
  SPEAKER_01: We've got who done it.
• 9:14
  SPEAKER_01: That's right.
• 9:14
  SPEAKER_01: Who done it?
• 9:16
  SPEAKER_01: We've got six suspects.
• 9:17
  SPEAKER_01: And we're asking the same two questions about each one.
• 9:21
  SPEAKER_01: Did they have a reason to?
• 9:22
  SPEAKER_01: And did they have the access to do it?
• 9:25
  SPEAKER_00: Right.
• 9:26
  SPEAKER_00: So the suspect number one is the auction.
• 9:28
  SPEAKER_00: The auction needs one countable number to run it all.
• 9:32
  SPEAKER_00: And it's had its hands on the click since 1998.
• 9:36
  SPEAKER_00: And not just Google's auction.
• 9:38
  SPEAKER_00: Every platform that could build one did.
• 9:40
  SPEAKER_00: And they all build the same way.
• 9:42
  SPEAKER_00: They report the same way and they train you the same way.
• 9:45
  SPEAKER_00: So motive and opportunity both.
• 9:48
  SPEAKER_00: Check, check.
• 9:48
  SPEAKER_01: The platforms they sell the inventory that scores well on the number they happen to be promoting.
• 9:54
  SPEAKER_01: Google published research saying about 89% of paid search clicks were incremental.
• 9:59
  SPEAKER_01: And Google's auction runs on clicks.
• 10:02
  SPEAKER_01: Facebook sent 2016 telling advertisers the click is only a proxy and lift is the truth.
• 10:07
  SPEAKER_01: And feed inventory tests badly on clicks and well on lift.
• 10:11
  SPEAKER_00: Yeah, everybody argues for the scoreboard that flatters their own shelf.
• 10:14
  SPEAKER_00: And that's not a conspiracy.
• 10:16
  SPEAKER_00: That's just selling.
• 10:17
  SPEAKER_00: Exactly.
• 10:18
  SPEAKER_01: So now last click attribution, no motive at all.
• 10:21
  SPEAKER_01: Because it never claimed to be anything other than a model.
• 10:24
  SPEAKER_01: But the opportunity was enormous because it was free and it was on by default, and almost nobody turns it off by default.
• 10:32
  SPEAKER_00: Yeah, then there's the CFO, 40 years of motive because they've always wanted a number that fits in the spreadsheet.
• 10:38
  SPEAKER_00: There's no opportunity.
• 10:40
  SPEAKER_00: The CFO never touched any of this.
• 10:42
  SPEAKER_00: They only asked for a number and we handed them one.
• 10:44
  SPEAKER_00: And that makes them an accomplice, not the poisoner.
• 10:48
  SPEAKER_00: Okay.
• 10:48
  SPEAKER_01: Now the media agency.
• 10:50
  SPEAKER_01: The reason is simple.
• 10:51
  SPEAKER_01: Agencies get judged every month, and a click report is the easiest thing to show a client that makes it look like everything's working.
• 10:58
  SPEAKER_01: And they had the access because in a lot of cases, they were the ones running the accounts and building the reports.
• 11:05
  SPEAKER_01: Internalization was not a thing yet.
• 11:07
  SPEAKER_00: And us marketers, we had the reason because a good click number made us look good in the mes in the meeting.
• 11:12
  SPEAKER_00: And we had more access than anybody else in this.
• 11:15
  SPEAKER_00: We were in the accounts every single day.
• 11:17
  SPEAKER_00: In a poisoning, that's usually who you look to first.
• 11:20
  SPEAKER_00: And we'll come back to that.
• 11:21
  SPEAKER_01: Now, one last piece of evidence before we get out of the costume.
• 11:24
  SPEAKER_01: And it changes how you feel about all of it.
• 11:27
  SPEAKER_01: Somebody new.
• 11:28
  SPEAKER_00: Yeah.
• 11:29
  SPEAKER_00: And there's a Facebook sales deck from 2016.
• 11:32
  SPEAKER_00: It's not a leak.
• 11:33
  SPEAKER_00: It's a deck that they showed to advertisers.
• 11:35
  SPEAKER_00: And inside it, there's a diagram with clicks filed under proxy metric and then lift filed under under true value.
• 11:41
  SPEAKER_01: There's also a Nielsen study in there across four hundred and seventy-eight campaigns finding no meaningful link between click rate and any brand measure, not recall, not awareness, not purchase intent.
• 11:51
  SPEAKER_01: And a second one, a supermarket scanner panel across more than two hundred packaged good campaigns, finding no relationship between click rate and return on investment.
• 12:03
  SPEAKER_00: So in twenty sixteen, a platform was showing clients a seven-year-old study saying that the click doesn't predict return.
• 12:25
  SPEAKER_00: Yeah, and there's a footnote deeper in the same deck in a six-point type saying that post-engagement is a rec isn't a recommended driver of advertiser return.
• 12:35
  SPEAKER_00: It's sitting underneath the chart telling the product that increases post-engagement.
• 12:40
  SPEAKER_01: Exactly.
• 12:40
  SPEAKER_00: So nobody hit it.
• 12:41
  SPEAKER_00: The warning label was on the bottle the whole time.
• 12:44
  SPEAKER_00: And so that's a mystery.
• 12:46
  SPEAKER_00: Let's get out of the costume, V, because you and I have lived this one.
• 12:49
  SPEAKER_00: Um, and the doses weren't only in 1998 and 2002 and 2005, but we were at our own every day.
• 12:55
  SPEAKER_00: Oh man.
• 12:55
  SPEAKER_01: We absolutely did.
• 12:56
  SPEAKER_00: Yeah, we did.
• 12:57
  SPEAKER_00: And none of us really stopped to ask what a click was.
• 13:01
  SPEAKER_00: We didn't have to because we just decided what it meant.
• 13:04
  SPEAKER_00: I'd look at a click rate and read it as how good the creative was.
• 13:08
  SPEAKER_00: This ad works and that one doesn't.
• 13:10
  SPEAKER_00: Then it meant that a good click rate started meaning people like our brand.
• 13:15
  SPEAKER_00: Then it's a good click rate started meaning that people are into us.
• 13:19
  SPEAKER_00: And then it meant we're really different from the competition.
• 13:23
  SPEAKER_00: And a file leaving a server can't tell you any of that.
• 13:26
  SPEAKER_01: And it's not that anybody decided to believe that, it just filled a gap.
• 13:30
  SPEAKER_01: There was a number, and then there was no other number.
• 13:33
  SPEAKER_00: Yeah, and the part that I still can't defend is that you know, like a 3% click rate is good, is a good day today.
• 13:39
  SPEAKER_00: We'd all be thrilled with that.
• 13:41
  SPEAKER_00: The part that's troubling is like the 97% out of every hundred people we paid to reach did nothing we could see, and we never talk about them at all.
• 13:50
  SPEAKER_01: Yeah, and the other thing, and I don't think our industry has ever properly reckoned with it, that we let the platforms train us on our discipline.
• 13:58
  SPEAKER_01: Certifications, best practices, a quarterly review with a rep who tells you what good really looks like.
• 14:04
  SPEAKER_01: Yeah, and we assume that that training was accurate.
• 14:07
  SPEAKER_01: It's worth remembering that the auction didn't start as a scam, it started because there wasn't enough inventory.
• 14:13
  SPEAKER_01: A handful of sites, a lot of interested advertisers, so you put it into an auction.
• 14:18
  SPEAKER_00: Yeah, I used to sell banner ads V at$200 CPMs.
• 14:22
  SPEAKER_00: Whew.
• 14:23
  SPEAKER_00: And it's just crazy.
• 14:25
  SPEAKER_00: I think people bought them.
• 14:28
  SPEAKER_00: $200 per thousand impressions fixed price around the site.
• 14:31
  SPEAKER_00: Uh and you really probably just get laughed out of the room for that now.
• 14:34
  SPEAKER_01: And you could charge it because the inventory was so limited and the advertiser got the whole site to themselves.
• 14:40
  SPEAKER_01: Then we all got excited about it.
• 14:41
  SPEAKER_01: The platforms manufactured as much inventory as possible so they could meet that demand.
• 14:46
  SPEAKER_01: Targeting later arrived, so two people could use different ads on the same page, and from there the whole case moved to efficiency.
• 14:54
  SPEAKER_00: Yeah, and then budgets get cut, and then you're told to do more with less.
• 14:57
  SPEAKER_00: So you switch on audience expansion, which um will have an amazing impact at dropping your cost per click.
• 15:06
  SPEAKER_00: And you feel really clever about it.
• 15:08
  SPEAKER_00: And nobody anywhere in that chain asks what the inventory actually was.
• 15:12
  SPEAKER_00: Some of it's bots and some of it never actually renders on a screen, which is my favorite line when they use the non-viewable impressions.
• 15:20
  SPEAKER_01: Those are the most important impressions.
• 15:22
  SPEAKER_01: Platforms started competing with each other on exactly that number.
• 15:25
  SPEAKER_01: Cheaper clicks over here than over there.
• 15:28
  SPEAKER_00: Which is why I don't think we get to stand outside the on this one.
• 15:33
  SPEAKER_00: We're complicit.
• 15:35
  SPEAKER_00: So Black Friday, Cyber Monday.
• 15:37
  SPEAKER_00: We've talked about the story a couple times before, but you and I were working at Sports Jack, this retailer, Canadian retailer, sporting goods retailer.
• 15:44
  SPEAKER_00: The whole catalog was online, pretty much.
• 15:47
  SPEAKER_00: Hundreds, thousands of SKUs.
• 15:49
  SPEAKER_00: We're in a war room, and it isn't a dashboard on a wall.
• 15:52
  SPEAKER_00: It looks a lot more like the New York Stock Exchange ticker.
• 15:56
  SPEAKER_00: There's people shouting, there's numbers moving, there's people, there's money going where the numbers are.
• 16:02
  SPEAKER_00: So the return on ad spent ad by ad by ad, updating live, and we're moving budget in real time out of the ads with lower ROAS and into the ones with higher.
• 16:11
  SPEAKER_01: I was right in the room with you, man.
• 16:12
  SPEAKER_01: I remember that.
• 16:14
  SPEAKER_00: Sell, sell, sell.
• 16:15
  SPEAKER_00: Bye, bye, bye.
• 16:16
  SPEAKER_00: So it's exciting.
• 16:17
  SPEAKER_00: It feels really great.
• 16:18
  SPEAKER_00: Right.
• 16:18
  SPEAKER_00: Because it feels like you're really doing something.
• 16:20
  SPEAKER_00: Absolutely.
• 16:22
  SPEAKER_00: But it's a good thing to know that we didn't stop the reach spend, which is worth saying, because the easy version of this story is that we cut brand and we didn't.
• 16:31
  SPEAKER_00: Well, I shouldn't say that exactly.
• 16:33
  SPEAKER_00: It's so nuanced.
• 16:34
  SPEAKER_00: We had these inspirational kinds of images that were still sales messages, but trying to be inspirational, so somewhat brand-like.
• 16:42
  SPEAKER_00: But anyway, that's a whole other story.
• 16:44
  SPEAKER_00: The money really kept going out the door for that, and it just wasn't on the board that we were looking at that week.
• 16:51
  SPEAKER_00: That just was a separate dashboard.
• 16:53
  SPEAKER_00: That week the board was all about ROAS and nothing else was on it.
• 16:57
  SPEAKER_00: So Reach was still running, it just couldn't win in an argument because it had no number in the room.
• 17:02
  SPEAKER_01: And we thought Reach was doing something really specific.
• 17:05
  SPEAKER_01: We thought it was building the audiences we'd retarget later.
• 17:08
  SPEAKER_00: Yeah, and that was the actual plan.
• 17:09
  SPEAKER_00: So the campaigns with no number were filling a pool, and the campaigns fishing out of that pool had the best number in the building.
• 17:15
  SPEAKER_00: So we separated them, but we just kept really focusing on the ROAS numbers and moving money toward the best performers.
• 17:24
  SPEAKER_00: And that ultimately was, you know, we were great.
• 17:26
  SPEAKER_00: We were smart.
• 17:27
  SPEAKER_00: We were doing really good, useful, productive things in those days.
• 17:31
  SPEAKER_01: Absolutely.
• 17:32
  SPEAKER_01: And you still can't say whether the retargeting even worked, though.
• 17:35
  SPEAKER_00: No, like for that filling the pool part or or or fishing in that pool that we were creating out of the reach ads.
• 17:41
  SPEAKER_00: No, we couldn't.
• 17:42
  SPEAKER_00: It might have been making it easier for people to get to the store, the retargeting ads.
• 17:46
  SPEAKER_00: It might have been filling us for people who are already coming.
• 17:50
  SPEAKER_00: Nobody in that room could tell you which, including me.
• 17:53
  SPEAKER_00: We also had partners that were there from the different big platforms sitting in the same room.
• 17:57
  SPEAKER_00: And it actually was never a conversation whether any of this was incremental or not.
• 18:01
  SPEAKER_00: Ultimately, it just the highest ROAS number was always the winner.
• 18:06
  SPEAKER_00: And it's the number that people want to hear.
• 18:07
  SPEAKER_00: And almost nobody in the in that room or that we're reporting to ever cared whether anything else was true.
• 18:13
  SPEAKER_00: Because it sounded like we were doing such a great job.
• 18:16
  SPEAKER_01: And that's the part that should worry everyone.
• 18:18
  SPEAKER_01: It didn't win the argument because it was right.
• 18:20
  SPEAKER_01: It won because nobody wanted the other answer.
• 18:23
  SPEAKER_01: And then I left and built a machine to do it faster.
• 18:27
  SPEAKER_01: Ooh, do tell, V.
• 18:28
  SPEAKER_01: Do tell.
• 18:29
  SPEAKER_01: Yeah.
• 18:29
  SPEAKER_01: So circa 2017, I joined the sporting life group.
• 18:33
  SPEAKER_01: I built the in-house system and let us move money near real time, you know, from one tactic to another within the platforms, within the ecosystems.
• 18:41
  SPEAKER_01: And the whole point of this was speed.
• 18:44
  SPEAKER_01: The signal says reallocate so you didn't wait.
• 18:47
  SPEAKER_01: Leadership loves that because it looks like you've got a pulse on the business every day.
• 18:52
  SPEAKER_01: You talked about the stock market, it was exactly that, but from a day-to-day perspective.
• 18:56
  SPEAKER_01: And the money came out of YouTube mostly, last click showed no conversions there.
• 19:00
  SPEAKER_01: So I'd pull it out into search and then retargeting.
• 19:04
  SPEAKER_01: And I need to give the system a chance to convert.
• 19:06
  SPEAKER_01: That's what I'd say in the room.
• 19:08
  SPEAKER_01: And we moved it without thinking twice.
• 19:10
  SPEAKER_01: Did you ever get go back and check on it?
• 19:12
  SPEAKER_01: No.
• 19:13
  SPEAKER_01: I ran the paid side.
• 19:14
  SPEAKER_01: So the paid story was the only story I was interested in telling.
• 19:18
  SPEAKER_01: I never looked at what happened to query volume or brand impressions or organic demand.
• 19:22
  SPEAKER_01: And what made it safe is that there aren't many large golf retailers in Canada.
• 19:28
  SPEAKER_01: There was enough demand in the category that showing up was usually enough.
• 19:32
  SPEAKER_01: So it worked in then in that case.
• 19:35
  SPEAKER_01: It worked.
• 19:36
  SPEAKER_01: Whether it worked because of what I did is a completely different question.
• 19:40
  SPEAKER_01: And I never asked it.
• 19:41
  SPEAKER_00: I love it.
• 19:46
  SPEAKER_00: There's one piece of evidence that I've used in client rooms for years, and I I want to throw it out on air because it it doesn't meet our own standard.
• 19:54
  SPEAKER_00: In 2017, Procter and Gamble cut around$200 million.
• 19:58
  SPEAKER_00: It's not that's$200 million, by the way, of digital ad spend and said that REACH went up 10%.
• 20:05
  SPEAKER_00: And the line that everybody quotes is that the first hundred million came out with no appreciable impact on the business.
• 20:12
  SPEAKER_00: The biggest advertiser in the world pulls the money and nothing happened.
• 20:16
  SPEAKER_01: Except there's no control group anywhere in that.
• 20:19
  SPEAKER_01: They cut, then looked at the business.
• 20:22
  SPEAKER_01: And the person telling you it worked is the person who made the call.
• 20:25
  SPEAKER_01: It also happened in the middle of a proxy fight with an activist investor who was publicly arguing the company used and withheld money to flatter its numbers.
• 20:35
  SPEAKER_01: And that's PNG's own filing with its SE.
• 20:38
  SPEAKER_01: And they changed the scoreboard.
• 20:39
  SPEAKER_01: They stopped reporting on that thing, they cut and started reporting REACH instead.
• 20:44
  SPEAKER_00: Yeah, and we we held that up as proof we'd be doing exactly what we spent this episode describing.
• 20:49
  SPEAKER_01: And there's an alibi that holds, which we're not going to hide.
• 20:53
  SPEAKER_01: Economists ran the same shutoff design at a car buying site on the theory that eBay might be a special case because everybody on earth already knows eBay.
• 21:02
  SPEAKER_01: Same substitution effect, far smaller.
• 21:05
  SPEAKER_01: Less than half the paid traffic came back on its own.
• 21:11
  SPEAKER_00: So there the ads were doing real work.
• 21:13
  SPEAKER_00: The same test, the opposite answer.
• 21:14
  SPEAKER_00: The same dose, different patient.
• 21:17
  SPEAKER_00: And that's the honest middle of this.
• 21:19
  SPEAKER_00: If you're huge and famous, the click mostly bills you for demand you already had.
• 21:23
  SPEAKER_00: If nobody's heard of you, it might be the only thing working.
• 21:27
  SPEAKER_01: Which is why you're getting the case and not a verdict.
• 21:30
  SPEAKER_01: Now, one last thing before we hand it over, because the victim is already being buried.
• 21:34
  SPEAKER_01: More and more, it's not people doing shopping, it's AI assistants shopping on their behalf.
• 21:39
  SPEAKER_01: And there's research, yeah, there's research this year where they built pretend online stores and watch how these AI shoppers actually buy.
• 21:48
  SPEAKER_01: Three things came out of it.
• 21:49
  SPEAKER_01: They trust ads less, so anything marked as an ad gets pushed down their list.
• 21:54
  SPEAKER_01: They still favor whatever is listed first, even though they aren't looking at the screen at all.
• 21:59
  SPEAKER_01: And they Keep picking the same handful of products so when the AI company updates its software, the winners can change overnight.
• 22:06
  SPEAKER_01: Nothing about your product changed, the software did.
• 22:09
  SPEAKER_00: Yeah, and if we think about what that means for the click, we always said a click showed attention that somebody saw the ad and cared.
• 22:16
  SPEAKER_00: And we baked all this meaning into that.
• 22:19
  SPEAKER_00: But these AI shoppers pick whatever's at the top without looking at anything.
• 22:23
  SPEAKER_00: So the top spot was never really proof of attention in the first place.
• 22:26
  SPEAKER_00: And then and when the thing doing the clicking is software, which is the bots, there's no interest, there's no intent, and nobody paying for attention at all.
• 22:35
  SPEAKER_00: What's left is a machine noting that a file went out the door, which is all the click ever was.
• 22:39
  SPEAKER_01: So where do the two of us land, Mark?
• 22:41
  SPEAKER_01: And this is just our view.
• 22:43
  SPEAKER_01: You, the listener, are still the jury.
• 22:45
  SPEAKER_00: Yeah, I don't think it's Google that killed the click.
• 22:49
  SPEAKER_01: No, I agree.
• 22:50
  SPEAKER_01: I don't I don't either.
• 22:51
  SPEAKER_01: And I'd be careful if anybody says that it is.
• 22:53
  SPEAKER_01: Every platform built the same kind of auction and sold the same number.
• 22:57
  SPEAKER_01: Picking Google is just picking the biggest one, and I wouldn't put it on the CFO either, by the way.
• 23:03
  SPEAKER_01: They asked for a number that made sense and were the ones that gave it to them.
• 23:06
  SPEAKER_00: Yeah.
• 23:06
  SPEAKER_00: I'd go a little bit further.
• 23:08
  SPEAKER_00: Who's to say that any of the platforms are guilty really?
• 23:11
  SPEAKER_00: Like they built what people were asking for, which is proof right now.
• 23:16
  SPEAKER_00: And whether it was true proof was somebody else's job and that somebody was us.
• 23:21
  SPEAKER_01: And I think that's where I land.
• 23:23
  SPEAKER_01: Us, the marketers, we're in the accounts every day.
• 23:26
  SPEAKER_01: We took the numbers at face value and we really didn't push back.
• 23:30
  SPEAKER_01: We let the people selling the ads teach us how to judge the ads.
• 23:34
  SPEAKER_01: And we never asked what a click actually meant.
• 23:37
  SPEAKER_00: And what we'd like you to take out of this isn't guilt.
• 23:40
  SPEAKER_00: It's that you're allowed to push back.
• 23:43
  SPEAKER_01: Exactly.
• 23:44
  SPEAKER_01: When a rep tells you there's headroom in your impression share and you should be at 94%, that's a sales target, not a law of nature.
• 23:52
  SPEAKER_01: You probably don't need it because the organic demand is already there and you're already paying to cover it.
• 23:57
  SPEAKER_00: And what and ask what happens if you don't ask for the hold out.
• 24:00
  SPEAKER_00: That's the same question that eBay asked back in 2012.
• 24:03
  SPEAKER_00: They had an answer in 60 days.
• 24:05
  SPEAKER_01: With one warning, because we don't want anybody hearing this as turn your ads off.
• 24:10
  SPEAKER_01: 60 days is long enough to see what a search engine is doing.
• 24:14
  SPEAKER_01: It's nowhere near long enough to see what the brand advertising is doing.
• 24:18
  SPEAKER_01: The Emberg Bass Institute, which is the marketing science group in Australia, looked at brands that stopped advertising on TV, radio, and print for a year or more.
• 24:26
  SPEAKER_01: On average, sales were down 16% after the first year, a quarter after two, and then more than a third after three.
• 24:34
  SPEAKER_01: It's slow, which is exactly why a short test really misses this.
• 24:39
  SPEAKER_00: Yeah, and that isn't an experiment.
• 24:40
  SPEAKER_00: So it's not the same kind of evidence as eBay.
• 24:43
  SPEAKER_00: Brands that stop advertising are often already in trouble, but the same pattern shows up across a few hundred brands in two countries.
• 24:50
  SPEAKER_00: And that kind of advertising is exactly what we're talking about in this episode.
• 24:53
  SPEAKER_00: It's the reach spend that had no number on the board in our war room at SportTech.
• 24:58
  SPEAKER_00: It's the YouTube money that cut polled at Golftown.
• 25:01
  SPEAKER_00: Hold out will tell you what the click is harvesting.
• 25:04
  SPEAKER_00: It won't tell you what you lose when you cut the thing that fills the pool.
• 25:08
  SPEAKER_01: Now it's over to you.
• 25:09
  SPEAKER_01: Tell us who you charge.
• 25:10
  SPEAKER_01: Reply to this episode, find us on LinkedIn, your suspect, and one line on why.
• 25:15
  SPEAKER_00: And the question I want you taking into the next one, V.
• 25:19
  SPEAKER_00: If the click was never worth what we said it was worth, what exactly are we doing right now with attention seconds?
• 25:25
  SPEAKER_00: That's a whole other episode, man.
• 25:28
  SPEAKER_01: That's why you're taking it.
• 25:30
  SPEAKER_01: What?
• 25:31
  SPEAKER_01: Anyways, thanks for keeping us in your ears.
• 25:34
  SPEAKER_01: Stay sharp, everyone.
• 25:36
  SPEAKER_01: We did it.