[themafia]

It's wild to me that advertisers are willing to use first party metrics. In any other media business you'd have a certified third party ratings agency to give "audience size" metrics some legitimacy.

Youtube has no incentive to accurately report this data and no apparent accreditation in their methodology.

[kulahan]

Google in general have been resistant to letting anyone see how effective their ads truly are - and most studies that get close tend to show extremely questionable efficacy results.

If Google shows everyone how ineffective ads actually are, they’d crumble.

[jefftk]

This is very much not true: Google has a bunch of options for measuring ad effectiveness, and when I was there (until three years ago) it was very hard to get advertisers to use them.

The two main options advertisers have are:

* Brand Lift Studies: split audience into treatment and control, use surveys on a small fraction of participants to measure impact

* Conversion Lift: again split audience into treatment and control, compare downstream actions like purchases ("conversions")

These both work on YouTube IIRC.

[kulahan]

I am not surprised google has many tools to tell you how great google is doing at using google's data for you.

Anyways, I was mostly referring to sales at physical locations; I assume it's pretty viable to build a system to figure out if someone who previously bought a lot of shein is now buying a lot of temu.

[jefftk]

If Google didn't want people to see "ineffective ads actually are" why would they build and push Brand Lift Studies and Conversion Lift?

Offline purchases are harder, of course, but pretty sure you can still do this: https://support.google.com/google-ads/answer/9994849

[kulahan]

For the same reason I would show someone a “totally real and honest” screenshot of my company’s finances for some quick cash. Doubly true if there’s effectively no way to be caught.

[jefftk]

Is your claim that these studies are incapable of measuring what advertisers actually care about? That Google is reporting fraudulently positive results? Something else?

(Once I understand your claim I'll be better able to respond: I think "there's effectively no way to be caught" is not true under both of the interpretations I see)

[lurk2]

> and most studies

Such as?

[girdi]

>It's wild to me that advertisers are willing to use first party metrics.

I agree, and find it even wilder that first party metrics from Meta and Google are trusted by most major advertisers (including ad agencies). I'm talking about six-seven figure budgets spent without any third party validation.

I've seen some studies on click fraud[0], but when advertisers are effectively choosing from a duopoly that has limited incentives not to lie in their metrics, I find it strange that there are no popular, widespread and accessible independent validation tools.

0 – https://www.mdpi.com/2073-431X/10/12/164

[Macha]

There's a whole industry of independent validation tools - DoubleVerify, IAS, Human, etc.

[kelnos]

I feel like it's more "second party" in this case. The first party is the creator, and the tracker/keeper of the view counts is Google. Google certainly isn't a disinterested, certified third party, but they're also not a creator who might make up inflated numbers to get a more lucrative sponsorship.

[rchaud]

Advertisers have 2 options for who to place ads with: Google and Facebook. When you have a monopoly, the customer has to take what it can get. Facebook has overstated its views and clicks for years to charge advertisers more, and faced no consequences for doing so.

[themafia]

This is largely true because this is where the largest volume of traffic comes from; however, it's not exclusive to these two by any means. There are some pretty big Supply Side Platforms and aggregators out there for advertisers to use. This comes into play a lot more often on podcasts and streaming audio, in particular, consider the fleet of Amazon Alexa devices out there.

Many of the advertisers that sell on these platforms are quite familiar with buying ads directly from "old school" media companies. So they have the competence and familiarity to be put off by the metrics but are apparently not in a position to force Google and Facebook to match standards used in other contexts.

[avbanks]

I've noticed this with TikTok and I'm almost certain YouTube 1P metrics are wildly inaccurate in particular views and non-bot comments.

[zahlman]

To me the wild thing is that this ad revenue model could ever have been profitable in the first place.

[roboror]

Click campaigns/conversions and user codes are more important than pure impressions.