Live data from Hacker News

Ad Fraud on LinkedIn

samueljscott.com

271–280 of 286 posts

Re: Ad Fraud on LinkedIn

#271

I observed the same with Google's "Smart" display ads. I ran a campaign for a couple of weeks and watched all of the traffic with HotJar. The first fishy thing was 99.9% of the traffic came from foreign countries despite the target being US/Canada. When I saw the targeting wasn't effective, I even set up specific geographic exclusions for every country != US/Canada and still the traffic was from mostly underdeveloped…

So would this then mean that Google's Adwords program for advertising your business based on keywords and target countries or regions is broken and unusable for business owners? (at least not profitably usable).

I've read very different from other commentators right here on HN who claim that Google ads work well for them and likewise I've heard plenty of anecdote from users of Facebook's and other ad platforms that they have delivered positive ROI for ad placement of their offers.

Any opinion?

Re: Ad Fraud on LinkedIn

#272

Earlier quoted context omitted.

Every been to a big break and motor store and they asked for your zip code? That is because they have someone in the back office to wants to see if the ads - dead tree paper - they spent to your zip code worked. It isn't hard to work the numbers from there, but it is tedious and requires some statistics.

ZIP code is also used for credit card verification.

Only at gas pumps.

Re: Ad Fraud on LinkedIn

#273
post #177
post #172

Earlier quoted context omitted.

On the one hand, I agree that the capability is concerning, but if it only records the behavior within a single visit and doesn't correlate it with an account, or across multiple visits, or across websites at all, I think that that probably wouldn't be much of an issue, for me at least. However, I don't see a way to allow that without also making the aforementioned unwanted connections also possible.

Here's the documentation for passing a user identifier to Hotjar: https://help.hotjar.com/hc/en-us/articles/360033640653-Ident... This is a bit better than it used to be. At one point the documentation had a field for User Email Address, and only a footnote in some documentation suggested it be hashed. Still, for a company that chooses to use the feature, the session recording can be tied back to the user account and…

Hashing email addresses doesn't really work. Most people don't use unique addresses per service and its easy to get a large list of email addresses that you can hash and do lookups in.

Re: Ad Fraud on LinkedIn

#274
post #160

Earlier quoted context omitted.

I work at Hotjar and love it that you're getting value out of the product! We do try to not record bots that advertise known user agents, like Google, Bing and other common ones. Reasoning being that recording non-humans will usually not tell you much about how your UX is optimized for humans. It's really interesting that somehow the fact that we recorded these less known user agent bots actually made the product mor…

Sounds creepy af. It's stuff like this that makes me keep javascript disabled whenever possible.

I think it goes so far as to be unethical, because it takes advantage of a gap in understanding[1] where most users aren't even aware of the possibility of this being possible.

[1] https://quinnkeast.com/writing/the-possibility-gap-its-time-...

Re: Ad Fraud on LinkedIn

#275
post #187

Earlier quoted context omitted.

Well, at my last employer we had our own systems analyzing traffic, we had integrations with third parties (DoubleVerify being the biggest) to detect fraud, our clients often had their own third party integrations to detect fraud and would complain if we hadn't found it but they did at too high a rate, etc. Honestly after the ad images/video, all the fraud detection javascript from the advertiser, the publisher, the…

Every been to a big break and motor store and they asked for your zip code? That is because they have someone in the back office to wants to see if the ads - dead tree paper - they spent to your zip code worked. It isn't hard to work the numbers from there, but it is tedious and requires some statistics.

No actually. Post codes are a recent thing here, and are mostly not hierachical. Also the standard reaction would be wondering why they need any info to process an instore purchase.

Re: Ad Fraud on LinkedIn

#276
post #187
post #182

Earlier quoted context omitted.

It's hard to get the numbers, though. I mean, I guess we could ask....

Well, at my last employer we had our own systems analyzing traffic, we had integrations with third parties (DoubleVerify being the biggest) to detect fraud, our clients often had their own third party integrations to detect fraud and would complain if we hadn't found it but they did at too high a rate, etc. Honestly after the ad images/video, all the fraud detection javascript from the advertiser, the publisher, the…

Amazing that we have an industry whose biggest component is keeping out fraud in the industry. Imagine pitching that to a VC.

Re: Ad Fraud on LinkedIn

#277

Earlier quoted context omitted.

Every been to a big break and motor store and they asked for your zip code? That is because they have someone in the back office to wants to see if the ads - dead tree paper - they spent to your zip code worked. It isn't hard to work the numbers from there, but it is tedious and requires some statistics.

And that strategy still doesn't account for word of mouth.

actually it does. That is why you need a full time statistician on staff (likely a department), they spend a lot of effort finding and accounting for confounding factors.

Re: Ad Fraud on LinkedIn

#278

Earlier quoted context omitted.

Revolut offers virtual and disposable cards. I've been (unethically) using them for these 1GBP / month trials a certain online newspaper offers.

Revolut was featured on HN just this morning: https://thehftguy.com/2020/08/12/a-haven-for-fraud-and-stole...

Guess she should have used a disposable Revolut card then

Re: Ad Fraud on LinkedIn

#279
post #212
post #208

Earlier quoted context omitted.

Hehe yes since i installed adnauseum i love ad's...and i love that they have to pay for it, atm at 5732$ ;)

Make sure you keep all your software up to date. Malvertising is pretty common, and adnauseum does a great job at firing payloads on your machine. It is the equivalent of installing an app that silently opens every attachment that is emailed to you.

They should make it so that it talks to a VM (with no persistent storage) and sends the links to be opened to it (with the current browser's context such as cookies, etc so attribution still works).

Re: Ad Fraud on LinkedIn

#280

Earlier quoted context omitted.

Wow, I had the same issue. I'm an ad-noob in the fullest sense of the word, but I tried the Google "Smart Ad". It turned out the same way you mentioned. I looked at where my clicks were coming from, and they were all pretty much from Hindi oriented Youtube channels. I was baffled at how that happened as I targeted US/Canada as you did.

It seems like pay-per-click and similar models should just be discontinued, they don't work since they can be exploited. An independently verifiable metric needs to be used (if the price for the ad should be traffic or interaction based) - or something completely different. Fixed prices?

Conversion tracking is the much loathed "spying/stalking" of ads.
Post reply on HN