Facebook to Pay $40M Under Proposed Settlement in Video Metrics Suit • Hacker News Board

Re: Facebook to Pay $40M Under Proposed Settlement in Video Metrics Suit

#31

Post by wastedhours » Mon, Oct 14, 2019, 1:03 PM UTC

disgruntledphd2 wrote: post #29

xiphias2 wrote: post #26

Earlier quoted context omitted.

Even ROI can't be counted separately, as both companies can count the same conversion as theirs (up to 30 or 90 days layer if I recall correctly). You really need to use 3rd party for validation.

And then you can't do viewthrough (especially on mobile). It's a mess, to be honest.

There's nothing I hate more than a viewthrough - media agency telling me they're responsible for hundreds of conversions when all they've done is put a banner on a webpage (usually combined with retargeting) that's in the viewport for a second.

Basically saying a non-guaranteed second of eyeball time in a banner-blind spot from someone they're retargeting who's already visited our site who converted 60 days later is their acquisition. [0]

I'm a marketer and I despise some of the crap these ad companies are selling.

[0] Conversely, this makes my personal KPIs look amazing, since I can take credit for other people's work and attribute it as my own. Obviously not doing that has caused me twice as much work to get proper attribution set up. If you want an easy life, push the BS metrics they give you.

Re: Facebook to Pay $40M Under Proposed Settlement in Video Metrics Suit

#32

Post by AndrewThrowaway » Mon, Oct 14, 2019, 1:34 PM UTC

cwkoss wrote: post #3

Impression fraud is going to be a huge scandal. So many view counts are inflated for profit - either by platforms themselves or by third-party bots interacting with content.

Once upon a time we have launched a rich media banner with three videos in it. What was interesting that all three videos generated around the same viewability stats. What was event more interesting that one video in banner was hidden with display:none.

I wonder if anybody does test like this - e.g. checks if hidden video generates views, button with visibility: hidden generates clicks and etc. and in this way evaluates traffic.

Re: Facebook to Pay $40M Under Proposed Settlement in Video Metrics Suit

#33

Post by wdr1 » Mon, Oct 14, 2019, 3:32 PM UTC

buboard wrote: post #15

siruncledrew wrote: post #4

Watching acquaintances raise funding around metrics like “viewership”, “engagement”, “reach”, and “accounts” has made me especially weary of companies espousing some kind of Golden Metric (big red flag if they don’t mention earnings/money) as a means to justify they are an awesome business. It’s just too easy to “fudge the numbers” and give the very-best-case estimates that don’t incorporate uncertainty or sensitivit…

Ideally you want to correlate the metric with the dollar ROI. I have yet to see extensive reports on that about google & FB. It's crazy to think the amount of money being thrown at them willy-nilly

Willy-nilly is a bit of stretch. There's quite a bit happening in the space. It's just a hard problem.

Re: Facebook to Pay $40M Under Proposed Settlement in Video Metrics Suit

#34

Post by disgruntledphd2 » Mon, Oct 14, 2019, 4:13 PM UTC

wastedhours wrote: post #31

disgruntledphd2 wrote: post #29

Earlier quoted context omitted.

And then you can't do viewthrough (especially on mobile). It's a mess, to be honest.

There's nothing I hate more than a viewthrough - media agency telling me they're responsible for hundreds of conversions when all they've done is put a banner on a webpage (usually combined with retargeting) that's in the viewport for a second. Basically saying a non-guaranteed second of eyeball time in a banner-blind spot from someone they're retargeting who's already visited our site who converted 60 days later is…

I dunno, I like view-throughs. I can definitely remember viewing an ad and then converting many, many times so I regard it as pretty normal. I totally get where you're coming from with providers and attributions though :)

Re: Facebook to Pay $40M Under Proposed Settlement in Video Metrics Suit

#35

Post by buboard » Tue, Oct 15, 2019, 12:11 AM UTC

disgruntledphd2 wrote: post #28

buboard wrote: post #27

Earlier quoted context omitted.

That's a great excuse for 1830 but today i would expect to see tens of precise , large scale studies from an industry that spends billions every year. The fact that I don't, makes me suspicious

These studies exist, but tend not to be published. Essentially there's little to no commercial benefit from publishing, so people tend not to. Like, I would argue that 80-90% of Goog/FB revenue is correlated with sales. This may be done badly (often), but it is done.

So how do advertisers find out who has the most effective ad system if they re not getting published? They choose in the dark?

Re: Facebook to Pay $40M Under Proposed Settlement in Video Metrics Suit

#36

Post by wastedhours » Tue, Oct 15, 2019, 8:16 AM UTC

disgruntledphd2 wrote: post #34

wastedhours wrote: post #31

Earlier quoted context omitted.

There's nothing I hate more than a viewthrough - media agency telling me they're responsible for hundreds of conversions when all they've done is put a banner on a webpage (usually combined with retargeting) that's in the viewport for a second. Basically saying a non-guaranteed second of eyeball time in a banner-blind spot from someone they're retargeting who's already visited our site who converted 60 days later is…

I dunno, I like view-throughs. I can definitely remember viewing an ad and then converting many, many times so I regard it as pretty normal. I totally get where you're coming from with providers and attributions though :)

Yeah, on a more technical level, the issue I put to the agency was that they had the same weighting in their conversion reporting for a retargeted viewthrough as they did for a click through conversion from a source that was the only channel in a linear attribution.

The former is some nice data on steps to convert, the latter is a conversion driver - the problem is, the latter is hard and the former is easy, so ad tech agencies try and spin them as being equal since they want to push the easy "wins".

Re: Facebook to Pay $40M Under Proposed Settlement in Video Metrics Suit

#37

Post by disgruntledphd2 » Tue, Oct 15, 2019, 3:03 PM UTC

wastedhours wrote: post #36

disgruntledphd2 wrote: post #34

Earlier quoted context omitted.

I dunno, I like view-throughs. I can definitely remember viewing an ad and then converting many, many times so I regard it as pretty normal. I totally get where you're coming from with providers and attributions though :)

Yeah, on a more technical level, the issue I put to the agency was that they had the same weighting in their conversion reporting for a retargeted viewthrough as they did for a click through conversion from a source that was the only channel in a linear attribution. The former is some nice data on steps to convert, the latter is a conversion driver - the problem is, the latter is hard and the former is easy, so ad te…

Yeah, that agency was talking bullshit. Sadly, all too common. I really wish Goog/ FB would get their shit together and allow proper measurement of them both, but it's not really in their interests to do so, sadly.