Hacker News Reader

Comment by wyc | Hacker News Reader

Bob Hoffman at the Ad Contrarian called this in January:

"P&G To Online Ad World: We've Had Enough" - https://adcontrarian.blogspot.com/2017/01/p-to-online-ad-wor...

Digital advertising is rife with (what should be considered) fraud. Advertisers feel as though they are getting fleeced, and the advertising middlemen reap massive margins. The accountability isn't where it needs to be; there are websites that hide ads under other ads, still reporting displays for all ads per page load. I suspect moves like this one will help the ecosystem change for the better.

View on HN

Yeah, but it's not like traditional advertising is free of fraud either. You can't even really pretend to measure the effectiveness of campaigns there. And there's much more money in it.

> You can't even really pretend to measure the effectiveness of campaigns there.

I disagree - for something at the scale of P&G, you can bet that they're doing a LOT of measuring for every campaign they run.

The article reads to me as if they're applying the same metrics to their online ads that they use already for traditional campaigns, and finding little impact.

Agree that they are measuring, but determining the cause and effect of each campaign is very hard. Sometimes ads have a clear impact, but those are cases that no one argues with. Mostly advertisers simply say (in effect) we got your message out, and there was an impact. If the low/high uptake was due to the ad or to the quality of the product or to the weakness of competitors - well no one can say.

>well no one can say. //

P&G must have the most and best data on the planet. By now they can probably say with a pretty high confidence interval.

That's not a real argument though.

Just assuming that they have a magical way to solve attribution for traditional ads because they're a bit corporation. If they did have that, they could just use the same tools on digital and report fake clicks and bots to Facebook and Google to get reimbursed. (Google at least refunds all impressions and clicks that were later detected as being fraudulent)

The sort of data they have would macro-, that's not telling you if any particular click is fake.

Correct. We're likely talking econometric/MMM modeling which likely tells them at the channel level what is or is not working, but they may not have enough data for that at the campaign level or lower.

a while back I saw a traditional display ad for KFC it was on a site heading out of town where the next KFC would have been 20 miles away - not sure they where getting much value out of that advert,

Well, this is the thing that needed to be said.

I worked in online ads for bigger part of my career, I certify that fraud is that industry's second nature. But the type of fraud they engage in is the one for which they can't be prosecuted (disclaiming, audits, etc)

Is there no government agency(ies) who track this to protect businesses from fraud? Otherwise with these practices in place, clearly the businesses selling ads are simply seeing what level of bullshit businesses will tolerate - how far into businesses' profits they can dig before the businesses say fuck this.

Advertisers are the most complicit in all this. They always have the option of using whitelists and buying properly but they (meaning the dozens of layers in the agency/vendor supply chain) always go for the cheap impressions and clicks to make their reports look good.

> there are websites that hide ads under other ads, still reporting displays for all ads per page load.

While people have done that, most ads seem to load load with some JS to confirm if the ad is visible, and if it's above or below the fold.

Why does it matter if the ad is visible? When those same companies post in People magazine, they pay for the placement they want and get no data about how many people saw the ad.

The agencies that buy the ad verify it was placed in the right spot at the right size and that's about all they can do.

Why should the web be any different? As long as they insist on bundling javascript with their ad, I'm going to run an ad-blocker.

> Why should the web be any different?

What? Because it can - it's the biggest reason so much money is transferring to digital and away from print, because of data and accountability. An ad that was at least scrolled into view is better than one that isnt and many campaigns are now bought on that basis.

It's not perfect but it's far better than before and we're not going back.

> it's far better than before

That's not really what P&G has found.

CPG can't be tracked is the problem, there's a major last mile / offline attribution issue and it's only getting tougher as security/privacy gets more regulated.

That being said, we work with P&G, they definitely know more now than decades before with older mediums.

How would it have any clue if it's below another div? Pretty much impossible

Because this is running on the customer's browser, CPU efficiency doesn't matter at all. They can do all sorts of stupid things, like grab the DOM from the browser, lay it out in a JS rendering engine and see if their div ends up being visible.

I haven't had the time to look through the ad javascripts on your typical fat webpage, but when there's tens of megabytes of code and it pegs my CPU at 100%, I know it's doing more than just displaying a .png in an iframe.

This kind of resource misuse is one more reason people run adblockers.

That and the security risks of running other people's code on your system.

Could you just pick a couple of pixels from the actual page display and check they're the right color?

Not sure either of these work without access at the top page level, ie you can't get the data from scripts in an iframe? Been a while since i did any JS dev.

Reply on news.ycombinator.com