Are Bots Invading Your Facebook ads?

Limited Run has received a lot of press the last two weeks about their Facebook ad campaigns. The digital startup made the claim that 80% of the clicks they get are coming from bots. And as you know, there’s been a lot of discussion about the effectiveness of Facebook ads in the past year, as Facebook filed its IPO and went public.

Archer has put our Facebook ads and tab to the test with real data about ad activity. We compared Facebook Ad and Insights (their tab-tracking tool) with our own server logfiles and with Google Analytics tracking every day for a week. It turns out that, no matter how we slice the data, Facebook’s traffic numbers are at best optimistic – they’re consistently and significantly higher than any other tracking tool that we’ve tested:

  • Some of the activity reported by Facebook doesn’t happen according to any other tracking tool. Facebook Insights reported 50% more tab activity than our server logfiles did (which include every file requested from the server, whether they were from real people, bots, spiders, or other automated programs).
  • Facebook is definitely reporting tab activity that can’t be attributed to people using web browsers. Facebook Insights reported five times as much activity as Google Analytics did. Google Analytics has been in the market for seven years; nearly half of the Fortune 500 websites use it to track activity.
  • Traditional “click fraud” doesn’t seem to be the cause here. “Click fraud” or “Ghost Clicks” happen when an ad (typically charged on a per-click basis) receives a lot of clicks, but the landing page doesn’t see a corresponding increase in traffic. Facebook Insights for tabs and Facebook ad clicks reports align closely (within 15% of each other every day), indicating that at least Facebook sees all those clicks that land on the tab.

Keep in mind that during the week we tracked ad and tab activity, we weren’t promoting the tab in any other way – no page posts, no other ads, no email, no search indexing. So we were able to isolate a full seven days of activity during which we could directly compare Facebook’s reporting tools (ad activity and Facebook Insights) with server logs and Google Analytics. Now, on to the data…

Facebook reports tab views that don’t occur according to server logs.

The raw Apache server logs for the website hosting our tab, which include every file requested from the server, only saw 80% of the activity that Facebook reported – and that trend was consistent for every day we reviewed. See the following chart:

Facebook vs Server Logs

Facebook sees 5 times as much activity as Google Analytics.

No matter how we cut it, we could never get Facebook and Google Analytics numbers to remotely agree. But the trend is consistent every day: Facebook over-reported activity by at least 400% and by as much as 567% on our tab compared to our Google Analytics tracking tool. Why is that? Let’s dig…

  • What about first-party cookies? Do some tools use them and others don’t? It turns out they are required for Google Analytics – but did you also know they have to be enabled to log into Facebook?
  • What about third-party cookies? Google Analytics doesn’t use them. If Facebook used them, and they were turned off, the Facebook numbers would be lower relative to the Google Analytics numbers – and that’s clearly not the case.
  • What about Javascript? Do some tools use it while others don’t? It turns out that Javascript is required for Google Analytics – but did you also know that it must be enabled to log into the desktop version of Facebook?
  • What about time zones? It’s true, Facebook and Google Analytics might be set to different time zones – which is why we looked at seven contiguous days of data. So if Google Analytics cut off some of the Facebook data on Sunday, that data would show up in Google Analytics as Monday activity. The chart above shows how the relationships between Facebook and Google Analytics remain consistent every day.

None of the “standard” explanations seem to apply in this case. There is something else going on; see the chart for the specific differences we found:

Facebook vs Google Analytics

Facebook dropoff rates are low.

In the week of sample data that we pulled, the activity on Facebook ads pretty closely mirrored the activity on our tab. Of the clicks that Facebook reported, 88% were also reported by Facebook Insights as tab views. The “worst” day for dropoff during our seven-day test had 85% of the ad clicks also reported as tab views, as you can see in the following chart:

Facebook Dropoff Rates

Given these discrepancies, what should you do? We’ve discovered that Facebook’s built-in reporting tools leave something to be desired.

  • Don’t rely on Facebook’s tools alone to report on tabs. Track your Facebook tabs with more than just Facebook Insights. Use Google Analytics, Webtrends, or another tracking tool that’s been tested and proven over time to confirm the activity.
  • Give your ad campaigns a “Purpose beyond the click”. Your campaign should do more than just drive clicks. If your tab has a goal (e.g., filling out a form, sharing a message), track the number of goal completions. After all, who cares how much you spend on clicks when you’re trying to drive goal completions? Figure out that cost, attribute it to the correct campaigns and goals, and move your budget into the tactics that give you the best cost-per-goal-completion.

Leave a Reply

Close Modal

Contact Archer

Close Modal

We know you're still using an older version of Internet Explorer. Do you know how much mind-blowing stuff you're missing because of this?

Save yourself. Upgrade now.
We'll help you...