by Freddy J. Nager, Founder + Director of Marketing Forensics…

With digital media, marketers should report both quantity and quality of results. Raw numbers alone can make a campaign appear effective, while obscuring serious issues.

Case in point: we used a digital-advertising service to run an ad promoting a business article. We paid for only clicks on the ad, since “views” can be deceptive: some sites count views regardless of whether people noticed the ad; even worse, they count views by bots. In addition, we did not show our ad to mobile users to avoid paying for accidental clicks. And yet the data revealed some suspicious results.

The Data

On average, the ad received a few clicks per day. Then, one morning, the number of clicks hit 40 in just a few hours. Did our ad suddenly appear before an ideal pocket of readers? We froze the ad and looked at the data:

  • The analytical report from the ad service could not identify one particular source. Further, it showed over 70 clicks — more than what our two other reports showed. The click-through rate was about average at 0.03% (low CTR’s are another reason to pay for clicks only), but the views had skyrocketed from a few thousand per day to over 200,000. That caught our eye, since we had recently lowered the price per click, and lower prices usually mean fewer views.
  • The report from WordPress showed 41 clicks from the ad but 0 clicks on any links within the article, which was unusual. In addition, no other section of the website that day received more than single digit visits. This meant that visitors to the first article didn’t explore further — also unusual.
  • The next-day report from Google Analytics bolstered our suspicions. GA showed 42 clicks from the ad and a glaring number: each visitor from that ad stayed on the site for only 3 seconds. By comparison, people who clicked on our LinkedIn ad stayed on the site for 6 minutes. One could argue that professionals on LinkedIn would have more interest in us than the general public, but 3 seconds is hardly enough time to read a sentence. Even visitors from Google, who often do not get what they’re searching for, stayed on the site an average of 44 seconds.

Conclusion

People or bots were clicking on the ad, and immediately clicking away. While this didn’t cost us much money (70 clicks amounted to less than $10), it could have added up to hundreds of dollars if we had performed only periodic checks — especially if all we tracked was raw traffic. The ad service reassured us that they have systems to detect advertising click fraud, but they could not explain our unusual results.

(Note: we never suspected the service itself of fraud, since we had used them for months without issues; rather, we suspect the managers of one or more of the sites on which the ad appeared. That said, we have stopped using that particular service altogether.)

Even if the clicks were free, they were worse than worthless. Some SEO experts believe Google tracks user behavior on websites to determine site “quality,” and that factors into how Google ranks search results. If users “bounce” (leave immediately after visiting just one page), Google might downgrade the page in search results.

Key Takeaways

  1. Check your site statistics regularly, at minimum once per day.
  2. Compare all numbers to averages on your site, and note aberrations.
  3. Look for measures of on-site behavior, such as time spent and number of pages visited per session.
  4. Use analytical reports from multiple sources.
  5. Remember that clicks can be just as vulnerable to fraud as views. Whatever makes money attracts scammers.
  6. Don’t celebrate traffic out of context. In fact, in Google’s eyes, your bad traffic may be worse than no traffic at all. You want numerous visitors who also enjoy the site. That means having both a good source of potentially interested visitors and good content to keep them engaged — but that’s a topic for a future post.

For more articles on marketing strategies and tactics, please subscribe to this blog.

Leave a Reply

Your email address will not be published. Required fields are marked *