bot

screaming world

Facebook Ads Bot Traffic? Get The Real Story

Recently a company we’ve somehow never heard of (until now) released results derived from their data. They called Facebook Ads out as being about 80% bot traffic.  Accompanying details to support their accusations were as limited as their name, in our opinion, and none seem to be forthcoming.

We were open minded until we dug in. Limited is way off the data we saw in our study. In the next couple of days we’ll publish a detailed response post in SearchEngineWatch (SEW).

aimClear’s results do not omit what we believe to be crucial data points and our findings were substantially different.  We’ll let you know when the SEW post is live. It names numbers.  For now, here are our top level findings. We think you’ll really dig the data in SEW so please watch for it.

First, a bot is hard to spot for certain. We should refer to all bots as “possible” bots because a malicious bot sometimes pretends to be something else.  There are good bots and bad bots. A good bot says in it’s user agent, “I’m a bot.” A bad bot can display a user agent that mimics a real user. For the purpose of this conversation, we call anything that has JS disabled a possible bot.

Here are aimClear’s high level findings:

  • The number of visits FB reports from ads appear to be nearly accurate, even if the visitors are bots. That’s, umm… comforting.
  • Possible bot traffic from the USA is much lower than Limited’s study purported, according to our data.
  • Internationally, possible FB Ads bot traffic is higher than in the USA but much lower than Limited reports according to our data.

International bot traffic is difficult to study, given the differences in timestamps between servers involved, international timezones, etc.. When analyzing you need to be extremely careful to understand the timezones that FB is reporting, your server logs, and PHP datestamps on pages.  A parsing error of as little as 4 hours may skew data significantly, especially in small samplings and undermine the entire test. We have no idea the mix of international traffic Vs. US traffic involved in Limited’s study. We do know ours.

Does LimitedRun really have the chops to call out FB? We don’t know. aimClear had a team of three highly skilled tech people, including  CTO Joe Warner, work on it for days.  Reed and Alyssa worked on it as well. Is Limited that good? We can’t say.

What we can say is that Limited is way off in comparison to our data. They reported a total of about 80% bot traffic.  They better show up with a lot more information about their data and methodology if they want anyone other than a bunch of tech IPO paparazzi, reporters and scared investors to believe them.

We’ll call for the data we believe missing and reveal real bot traffic numbers specifically in the SEW post. Hang tight!

-->

12 Comments

  1. Barry Tubwell on August 7, 2012 at 3:27 pm

    Hard hitting reporting from aimclear, LOVE it man, just freakin love it. Can’t wait til SEW Post but if its anything like “ALL” your other articles its sure to be a block buster and I won’t be surprised if it make front page news in the main stream media (I am not saying aimclear is not main stream :)

  2. Vincent Ammirato on August 7, 2012 at 4:51 pm

    This is gonna be good!

  3. Gary Cameron on August 7, 2012 at 5:11 pm

    Makes no difference to me. Facebook is so inconsistent with regard to what ads get approved and what ads don’t that I have switched back to Google.

  4. Bill Scully on August 7, 2012 at 6:32 pm

    It’s great to see a company spend the time and money to confirm a poorly documented report before it becomes the stats people start throwing around. Looking forward to the article!

  5. Marty Weintraub on August 7, 2012 at 6:46 pm

    @Bill Scully: Thanks, we did spend thousands on the study in media spend and time. Your comment made us feel appreciated. :)

  6. Doug Antkowiak on August 7, 2012 at 7:14 pm

    I’m super excited you guys are doing this study. I remember reading the initial claim that 80% of FB ads were bot traffic and thinking it was an outlandishly high number. Also, it’s refreshing that you say how hard it is to find a bot. Looking forward to the findings!

  7. Rob Leathern on August 7, 2012 at 7:22 pm

    In my personal opinion, the journalists who wrote this story without having even the simplest and most important fact (“80% of what number?”) did us all a disservice.

  8. Jeff McNeill on August 7, 2012 at 7:53 pm

    This is a ridiculous article. There is nothing here of any substance. Linkbait nonsense.

  9. Karpos Media on August 7, 2012 at 8:25 pm

    It will be nice to seem some real numbers and detailed methodology. This issue has gained some real press over the past week. If your FB ad does send bots to like your FB fan/business page, it should be quite easy to spot them. Dead giveways for spotting likes from fake profiles:

    1) Profile image of female model
    2) The bot has A LOT of likes
    3) Not many friends (or most of them are bots)
    4) Fake occupations or universities

    After eyeballing a few of these profiles, they become quite easy to identify. Surely there must be pattern recognition scripts that could identify a fake profile with reasonable reliability. Anyway, great reporting. I’m looking forward to a good read on SEW!

  10. charl hoffman on August 8, 2012 at 2:25 am

    And they say old fashion scare marketing is dead! What better way to get noticed. Well… Still would really be interesting to see if they can back up their figures.

  11. ed dawson on September 10, 2012 at 12:15 am

    Hmm… has the bot traffic data post been published over at SEW? Can you guys share the link in the comments or update the post?

    Thanks :)

  12. Alex on September 18, 2012 at 2:28 pm

    So what’s the update with the Facebook bot rumors? I manage social media accounts and have noticed a lot of clicks in only 1 day. Google Ad words would generally take a few weeks to generate that type of action. I’m definitely skeptical.

Post a Comment

*

*