Hacker News Reader

Comment by kevingadd | Hacker News Reader

Other metrics reporting ~100-150k vs Facebook reporting 600k+ clicks seems well beyond the level of error that would be produced by not tracking things in the exact same way.

Coincidentally, FB metrics reporting 2-5x what the target site's metrics show would be very easily explained by bots generating traffic to FB's servers without actually redirecting over to the target site...

View on HN

Exactly. Facebook probably sends 302 Temporary Redirect, counts it as a click, and the bot simply does not follow the Location header.

Are there any ways to avoid this?

They should put a web bug on the target page to confirm the redirect was completed.

Except that the contract prohibits doing this sort of "third party audit".

I mean Facebook should put a web bug on the target page.

But Facebook is benefiting from the problem! At least in the short term, until it harms their reputation in the advertising market.

Our testimonial sharing app tracks clicks from sharing on facebook, linkedin, twitter and google+. When we first implemented it we saw that shares on social media were getting like 80 clicks in a few minutes. We poked around and found that bots inflating clicks had been a problem for url shortners like bit.ly in their early days.

After looking carefully at user agents, IP addresses and behaviour we found that most bots don't load javascript, and could be filtered out by just loading some javascript as part of the redirect. Not sure what Facebook is doing..

FB says they track clicks. If I click an ad really fast, say 5 times before my browser can redirect and load the target site does FB count that as five and server analytical show only one because only one of the five clicks resulted in a request? I might have to do a micro experiment to find out....

It sounds like the ads were directed at Facebook pages. Then the 150k was the number who clicked then ad, then clicked through to his actual website(s). In which case, that's actually a pretty fantastic clickthrough rate. At least, that's the understanding I got from this paragraph:

At one point, data from Facebook indicated his ads had delivered 606,000 clicks, but the site itself registered only 160,000 incoming clicks from Facebook, according to data supplied by Brar. (160,000 clicks is a not insignificant return. After all, these are not clicks on a mere Facebook page, these are users who clicked through to an off-Facebook site.)

Otherwise I'm not sure what they'd mean by "160,000 clicks is not an insignificant return." Maybe I'm reading too much into it though.

It sounds like the ads were directed at Facebook pages.

I don't think so, the article says

After all, these are not clicks on a mere Facebook page, these are users who clicked through to an off-Facebook site.

I.e. the link that the user would go to after clicking the ad would be an off-facebook site, not a facebook page.

Right. The 160k clicks are not clicks on a mere Facebook page. Perhaps the original 606k that Facebook is talking about are though. And the 160k are the number of those that click through to the off-site page. Anyway, probably not. Was just a theory. The article could be clearer.

Reply on news.ycombinator.com