Hacker News Reader

Comment by runarberg | Hacker News Reader

Targeted ads don’t work any better then static ads. This is new for me. Actually now that I think about it, all evidence I remember at the moment is anecdotal. So perhaps you are right.

Regardless of its efficacy, the legality should be out of the question.

View on HN

Yes they do. It really bugs me all these HN threads state this and rely on their own personal experience not as an ad buyer but as a consumer. It's not true.

One can directly measure ROI and prove the value of this advertising. Especially FB provides for my business (political marketing) at least 10X better direct response value and that is mostly using 1:1 targeting and lookalike modeling.

Cool. Just as I thought. It seemed wrong that such an easily measurable thing would never have been tested and a whole industry (arguable the biggest industry in the western world at the moment) had never measured it (or they had and found no effect).

What I like about the HN discourse is that if someone slings out a statement which is demonstrably false, someone that knows better might respond with a correction. That is why I left this comment, as I all my knowledge with the targeted ad business was anecdotal, and I desperately needed a correction.

FB has also lied consistently about the performance of their ad products, so who knows if you are really getting that ROI.

i know... and so do probably most major and direct response marketers.

there are a lot of ways to measure outside of FBs analytics tools.

including the very simple simply using a refcode or even a specific product/url. How many purchases did you get / how much did you spend?

this is what i'm referencing. like that's so incredibly basic and it bugs me HN just either doesn't understand or lets their opinion on it go first

In my experience FBs inaccurate reporting is just a tiny amount of impressions delivered and they refund the $1 or whatever it is a couple times a year. Have they ever had a reporting anomaly on their analytics tools for tracking off FB conversions? they very well could have but I haven't heard of one.

> there are a lot of ways to measure outside of FBs analytics tools.

There is no way to validate their impression/click data that all the downstream metrics are reliant on. If you have a way, I'd love to know about it. The claim isn't that they make up conversions, the claim is that they overstate the ROI and cv% based on the denominator.

It is true and has been verified from companies who conduct proper studies internally and choose to release it such as eBay.

There is also an academic researcher who regularly publishes as well.

https://pubsonline.informs.org/doi/abs/10.1287/mksc.2019.118...

and of course there's Tim Hwang.

https://www.fsgoriginals.com/books/subprime-attention-crisis

Internet ads do not work past simple location based targeting.

There are plenty of examlpes on both sides.

I have routinely gotten over 100% immediate last-click roi from fb ads, for many clients.

I know for a fact many others have too. There are a ton of direct to consumer online-only products that easily measure this. Just make a page or item which you only use to advertise on FB. sales/spend. Game developers track this closely too.

This is anecdotal evidence. Nobody is disputing those exists. However I think FB and Google owe their existence a little better evidence, such as A/B testing, control group studies and even experiments to actually demonstrate their effectiveness. Until we have those (and perhaps we do; I seriously know nothing of the existing literature) the anecdotal evidence we do have should be taken only as indicative of effect, not proof of effect. Anecdotal evidence come with a ton of bias.

Was in adtech for awhile.

Targeted ads work really well for many scenarios, b2b software.

Like I said, all evidence I know of is anecdotal. Which—for sure—is indicative, but by no means conclusive.

I’m not sure it makes sense to speak this broadly about efficacy, since there are probably variations in returns amongst different targeting groups and value props. Execution matters.

I’m also not sure if it matters whether they are actually effective. At least for the short term. If people believe they are effective, is there any difference in the dynamic?

Reply on news.ycombinator.com