Most active commenters
  • jariel(4)
  • tomnipotent(4)
  • cm2012(4)

←back to thread

Facebook reported fake numbers to advertisers

(news.yahoo.com)
618 points elorant | 11 comments | 19 Feb 21 14:34 UTC | HN request time: 0.037s | source | bottom
Show context
sputr ◴[19 Feb 21 15:18 UTC] No.26194057[source]▶
I keep warning small time (ie most) FB page owners who advertise on FB to be very very careful as they are being subjected to a beefed up version of the psychological manipulation that regular users face as they, not the regular users, are the main customers.

Facebooks corporate incentive is to get you to FEEL like your getting good value out of advertising on Facebook and to get you addicted to doing it.

Not to actually deliver results.

So don't trust any metric they show you, because even if its not a total fabrication it's still presented in a way to deceive you to think its better than it is.

Always monitor your ROI and always calculate it using your truly end goal (sales, or in the case of civil society some sort engagement off Facebook that's tightly bound to you mission). Likes, shares, comments and reach should NEVER be the goal. Even if FBs interface is trying to convince you otherwise.

replies(10): >>26194191 #>>26194413 #>>26194461 #>>26194504 #>>26194560 #>>26194714 #>>26195371 #>>26195775 #>>26196787 #>>26198495 #
spideymans ◴[19 Feb 21 15:54 UTC] No.26194413[source]▶
>Facebooks corporate incentive is to get you to FEEL like your getting good value out of advertising on Facebook and to get you addicted to doing it.

Even more reason for us to be doubtful about FB's claims that small businesses would be decimated without FB's invasive tracking.

replies(1): >>26194708 #
cm2012 ◴[19 Feb 21 16:22 UTC] No.26194708[source]▶
If FB was actually completely forbidden from tracking, I'd estimate 85% of small shopify stores would die with it. The winners would be giant marketplaces like Amazon, who would be the only reliable sources left of customer acquisition.
replies(7): >>26194740 #>>26194771 #>>26194928 #>>26195096 #>>26195599 #>>26196285 #>>26198728 #
Red_Leaves_Flyy ◴[19 Feb 21 16:57 UTC] No.26195096[source]▶
What's your basis for this thesis? Likewise, how many of those shops are dropshippers that never touch product?
replies(1): >>26195826 #
jariel ◴[19 Feb 21 17:57 UTC] No.26195826[source]▶
Facebook is one of the only means to do reasonably targeted advertising with a broad reach.

Google is keyword only, and that's limited. Banner network display ads are useless.

The privacy debate is woefully lopsided by people who have never spent a dime marketing. I suggest all the startupy people on HN spend some time trying to get the word out and then they'll realize what the 'hard part' of the business is because it's not code.

Efficient advertising, which is to say getting in front of people who have a legit curiosity for your product with ads that are not distracting, is possible and ideal for everyone, but can only be done with at least some data.

The economy would grow literally by 1% more if we could get people connected with the things they need, when they need them and we'd all be better off.

replies(5): >>26195913 #>>26195994 #>>26196121 #>>26198548 #>>26200026 #
bart_spoon ◴[19 Feb 21 18:20 UTC] No.26196121[source]▶
This entire thesis hinges on targeted advertising being effective. There is a growing group of people who are increasingly doubtful of this [0].

I personally have worked as a data scientist trying to assess the value generated by various advertising campaigns, and I personally found that the field is rife with egregious statistical misuse, usually because it was necessary to prove significant ROI on advertising.

[0] https://thecorrespondent.com/100/the-new-dot-com-bubble-is-h...

replies(3): >>26196781 #>>26198038 #>>26199485 #
1. jariel ◴[19 Feb 21 21:07 UTC] No.26198038{6}[source]▶
"This entire thesis hinges on targeted advertising being effective. "

Targeted ads are unequivocally more effective than non targeted ads, on the aggregate - there is no dispute other than at the margins.

Do you think that advertising makeup to the general population has the same effectiveness than advertising it to women? Or women who have shown an interest in makeup?

Ads are complicated and nuanced, but everyone in the industry already knows this.

There will always be science at the margins as we discover the means by which people truly engage, but otherwise, there is no arguing with core demographic targeting. It would be like completely non-technical people saying "Javascript is completely ineffective because of null ambiguity" whereas it's universally used, and the limitations of JS are recognized to all but the most junior developers.

replies(1): >>26198793 #
2. tomnipotent ◴[19 Feb 21 22:22 UTC] No.26198793[source]▶
The general consensus is that personalized advertising, specifically, is no better than contextual advertising that came before it. Everyone was fine and happy with contextual advertising.
replies(2): >>26200877 #>>26200925 #
3. jariel ◴[20 Feb 21 02:21 UTC] No.26200877[source]▶
Yes, this is true, but I suggest it's because despite the privacy invasion, we don't know that much about individuals. We don't need to though.

Literally just basic demographics, plus a few interest points ... that's all I think we need to do well of those could be reliable.

replies(1): >>26200936 #
4. cm2012 ◴[20 Feb 21 02:31 UTC] No.26200925[source]▶
Wow, the "general consensus". That's why FBs advertising revenue is so low compared to Reddit! If only personalized advertising were effective, FB could be worth something.
replies(1): >>26201091 #
5. cm2012 ◴[20 Feb 21 02:32 UTC] No.26200936{3}[source]▶
This is incredibly not true. For one, the best data point FB has on predicting personal likelihood to click an ad is which ads that person has clicked in the past.
replies(2): >>26201047 #>>26201107 #
6. jariel ◴[20 Feb 21 02:51 UTC] No.26201047{4}[source]▶
I don't know what you are objecting to i.e. Social Media surprisingly not knowing a lot about us - or - the need only for relatively small amount of solid demo data.

On the first point - they just don't know that much about us. That's what the whole post is about (!) They were lying advertisers ad to the probabilistic estimates of who we are.

Go here: https://adssettings.google.com/authenticated?hl=en

Google will show what they know about you if you have that on. It's not very accurate at all. It's very fuzzy. For me they have wrong age, a lot of wrong interests.

So they surprisingly don't know a lot, just because someone clicks on an add ... that is not a lot of information - there's so many reasons we click on ads. Google thinks I'm into cars and I don't even drive, I hate cars!

As for basic targeting - it's mostly what advertisers need. So age, education, income, zip code (or zip code classification), rough location would be great. Plus just a few, solid interests (real ones, not estimated) - and that would be a big step forward for ads.

So yes - our privacy is fairly aggressively invaded at the same time they can only glean so much from it. Seems like a Paradox.

7. tomnipotent ◴[20 Feb 21 03:00 UTC] No.26201091{3}[source]▶
That's why Google average revenue per user is so much higher than Facebook.
replies(1): >>26202124 #
8. tomnipotent ◴[20 Feb 21 03:04 UTC] No.26201107{4}[source]▶
> the best data point FB

Do you have any evidence to back this up? I can't find any published literature where Facebook has hinted this, including this paper [0] saying:

> As expected the most important thing is to have the right features: those capturing historical information about the user or ad dominate other types of features.

0. https://research.fb.com/wp-content/uploads/2016/11/practical...

9. cm2012 ◴[20 Feb 21 06:31 UTC] No.26202124{4}[source]▶
Google display network (contextual ads) is tiny. Google search is direct intent, very different.
replies(2): >>26206692 #>>26207844 #
10. tomnipotent ◴[20 Feb 21 18:49 UTC] No.26206692{5}[source]▶
It's still 15-16% of Google ad revenues, or about 40% of Facebook.
11. Closi ◴[20 Feb 21 20:52 UTC] No.26207844{5}[source]▶
Google search also uses personalised advertising. If you search for restaurants you will get ads for restaurants nearby, if you search for clothes it will match the ad results to your gender if google knows it, you can also do the same remarketing techniques with other ads (eg if someone has visited your website and they search for a related topic you can make sure your ad is displayed).