Hacker News new | ask | show | jobs
by adambratt 3651 days ago
I've spent a good amount of money on Facebook ads in Asia (>$300k) and while there may be some fraud, there's definitely a completely different pattern of Facebook use in some countries. I think what you're observing is cultural differences in Facebook use, not fraud.

It's incredibly hard to get people to like or engage with your posts in Korea/Japan unless you have a ton of social proof already. This is definitely a cultural aspect as no one wants to be the first or one of just a few to engage with the post. In general, I've found these two countries to be more expensive than the US for getting likes even though it's about 1/3 the cost for getting post clicks.

Meanwhile, in Indonesia/Vietnam/Philippines it's incredibly easy as many people just "Like" everything in their feed as they scroll down. I've talked to several people from these countries about Facebook usage and they say it's their way of marking that they've seen a post. It's funny but I frequently get more likes than clicks on my posts that I run in these countries. Some may be fraud, but after seeing how people actually use Facebook in these countries I'm inclined to believe it's legit.

One of the hacks that works well for me is to take something I want to run as an ad in Korea or Japan and run it in Indonesia first. $2 for a post engagement campaign will get me around 500 likes. That seems to be enough to tip the scale in Korea/Japan and get them to start liking it en masse as well. It's crazy but my like rate goes up around 30x in Korea when I use this strategy of pre-seeding likes.

Anyways, going back to this author's post, the Facebook algorithm seems to be trained to follow what works and gets you the cheapest engagement rate. This is why I never ever run a single ad set with multiple countries and interests as it will just all end up saturated on the one that starts out working best. You should be running separate ad-sets for each country/interest, doing it any other way is a COMPLETE WASTE. Do not run Facebook ads like this.

Furthermore, these really don't look like spam accounts. These look like real Southeast Asian FB accounts. Most of my friends in that part of the world have very similar looking Facebook accounts with a shit ton of random friends/likes.

4 comments

One of the hacks that works well for me is to take something I want to run as an ad in Korea or Japan and run it in Indonesia first. $2 for a post engagement campaign will get me around 500 likes. That seems to be enough to tip the scale in Korea/Japan and get them to start liking it en masse as well. It's crazy but my like rate goes up around 30x in Korea when I use this strategy of pre-seeding likes.

Interesting that, if we accept the click-farm premise of the article, your hack puts the farms in your service to get the initial likes, and then drive the kind of engagement you are actually seeking (in the crowd-influenced countries like Japan/Korea)...

In an environment like this, where the "likes" themselves are suspect, what counts as a success story for FB promotion for the work that you do?

I focus mainly on driving clicks to both content and eCommerce. The actual likes and comments on a post don't really matter much to me but it does help for increasing the CTR
- Re: cultural differences in Facebook use, not fraud. Plausible. Based on this and the previous ads I've run in E/SE Asia, though, I'm still under the believe that something is askew.

Re: Japan/Korea - Yes, collectivist culture, conservative buying habits (e.g. not wanting to be first mover, etc.) can very well translate to FB activities.

Re: VN, etc - Agreed, they play faster and looser there with their likes and friending.

Re: Your hack - It's buried at the bottom of the Updates section, but I did do something very similar. As it was becoming clear to me that the VNese engagement was garbage, I figured that I still might be able to leverage that to get more conservative places like Japan and Korea to jump on board. No dice. This might have to do with how much budget I was contributing (another topic I talk about in the post).

Re: Per-country ad sets. - Again, I did this.

I ran a boost targeting JP/VN/KR, got 100% engagement from VN. Revised the target set to focus just on Japan, crickets, and that's even AFTER there were already 300 likes for that boost, which falls in line with your hack. Ok, not crickets...I got three likes, one Japanese...and then one Cambodian and one Filipino, both living in their respective countries.

Re: Doesn't look like spam accounts. - They look spammy or garbage to me. 18yr old girl from one of the small province towns in VN has 3,000 or 4,000 friends? Could be explained away as just cultural differences, or it could be complete garbage.

Another traveler with a lot of VN female friends on Facebook checking in, those accounts don't look very odd for the region. It's not uncommon to have a girl have to remove friends to add you. I've never really asked them why they have so many friends ... sorry.
Ooh. So Facebook's “Be the first to Like this” text, assuming it's translated directly to Japanese, might be a deterrent there?
Excellent information.

May I inquire as to what product/service you promote with Facebook ads?

Is it me or does it smell like aged accounts in here?
Aged accounts?
The practice of creating accounts on a site and making "real" seeming posts for a long time so that when you want to flip the switch and use them for something corporate, they look like they have a background and aren't flagged for being 1 day old accounts with 1 corporate post.
this is very common on sites like reddit and it's even used by intelligence agencies [1]. Politicians or firms can hire companies to sway public opinion [2], check their clients [3]

[1] https://theintercept.com/2014/02/24/jtrig-manipulation/

[2] http://www.thedailybeast.com/articles/2016/04/21/hillary-pac...

[3] https://modsquad.com/work/

pretty easy to programme a bot that acts like a teenage girl. * parse magazine * foreach ad/article * capture image (sources: google images, tubmblr, #tag) * Post article to FB feed When idle; * every 30mins wakeup * search through network, like (rand * (TotalPosts * 1/3 * 1/8)) * like all sponsored posts