Why Marketers Can’t Ignore Bot Traffic on their Sites

As a thorn in the side of marketing teams of all sizes, awareness of ad fraud has grown in recent years due to the sheer amount of money it can cost advertisers. In one famous case, Uber discovered fraudulent app installs attributed to its ads had cost the company $100 million.

But it’s not just overtly malicious activity like ad fraud that marketing budget-setters need to be concerned about. Marketers must be aware of the potential damage bots of all kinds can do, intentionally or otherwise.

Here’s why marketers need to get educated about bots:

• Bots are rampant across the web (on average, 50% of traffic is bots)
• Sophisticated bots are indistinguishable from human visitors without specialized tools
• Ad fraud costs an average of 4% of online revenue
• Bot traffic skews marketing data, leading to misguided decision making, which costs businesses, on average, 4.07% of online revenue

Read on to find out more about bots and how much they affect marketing, often without businesses realizing the full implications.

Intentionally or not, bot traffic affects marketing analytics

Bots are programs automated to carry out actions on websites, apps or APIs, just like a human would, but faster and without the manual effort. Sometimes bots are helpful: for example, search engine crawlers are required for SEO, and publishers may want a search engine bot to scrape their articles for syndication to new audiences.

But bots can also act maliciously, exploiting websites and customers whilst hiding amongst real users.

What risks do bot traffic pose to marketing?

Good or bad, traffic created by these bots has a secondary effect on the websites that marketers might not know about. We invited two experts in ad fraud and marketing, along with one of our own bot experts, to discuss why marketers need to think about the impact bot traffic is having on all aspects of their jobs – not just ad spend, but also strategy, planning, and reporting to the wider business.

Catch up with the webinar on demand: How are bots skewing marketing analytics

What kinds of ad fraud are affecting businesses today?

According to Neil Andrew, founder and CEO of ad fraud solution Lunio, “it’s extremely easy to commit ad fraud. Anyone with a working knowledge of Google and an internet connection can get started in five or ten minutes. You can go to a dodgy underground hacking forum and pay someone to set all this up for you and it’s really not difficult or expensive.”

There is also little to no risk of prosecution if caught (ad fraud itself is not a crime).

Click fraud by competitors

Andrews also pointed out the common misconception that ad fraud is mostly driven by competitors automating clicks on adverts to use up ad spend and have their own adverts seen more prominently. In fact, this activity makes up only a tiny fraction of ad fraud.

Publisher fraud

More commonly, publisher fraud sees fraudsters set up fake websites for the target business’s adverts to be displayed on, and then use bots to manufacture a large number of clicks on these ads. The victimized advertiser then unknowingly pays the fraudulent publisher for these ‘fake’ clicks.

Attribution fraud

Attribution fraud misattributes organic clicks as many more paid clicks to artificially inflate the success of PPC advertising. Not only is this ad spent wasted, but the advertiser is tricked into thinking their ads are more successful than they are, and so is likely to increase their spend on this channel.

How much does bot traffic skew marketing data?

Netacea’s Head of Threat Research, Matthew Gracey-McMinn, highlighted that malicious bots are very good at hiding within normal peaks and troughs of everyday expected traffic. Bots often match their traffic patterns to human behavior (e.g. active during peak times) to avoid detection.

A side effect of this bot tactic is that anomalous traffic blends in with real users on marketing reports. The domino effect spreads widely through businesses, causing often unseen damage.

What happens if we can’t trust our analytics?

To provide effective reporting, forecasting and allocate spending effectively, marketers need good data. This point was expanded on in more detail with several examples by Patrick Curran, Head of Paid Media at marketing agency Spike.

Auto-optimizations react to bad signals

Automation is a powerful tool for reacting instantly to large amounts of user behavior data, allowing spend to be allocated to the most successful ads and audiences without manual intervention. However, if this action comes in response to a bot scalping items instead of genuine customers browsing products, for example, these automations are pouring money into the wrong places at the wrong times.

Over-reporting of traffic and attributed revenue

Certain pages or areas of a website can see their popularity artificially boosted by bots. Even social media posts can seem more popular than they are if they receive impressions and engagement from bot accounts. Performance-based marketing strategies are skewed as a result, leading businesses down the wrong path and focusing time and effort on campaigns that may not have been as successful as they appeared.

Chargebacks due to fraudulent sales

If products are purchased fraudulently by bots as part of a carding attack or using stolen accounts, the business will suffer chargebacks to the legitimate owner of the credit card or account. However, marketing teams rarely factor this into revenue figures when reporting on channel or campaign effectiveness.

Bad data leads to bad decision making

“If a bot is actually filling out forms, you’re wasting time resource by trying to contact those people,” Patrick added. “If they then fill out an email inquiry as well, that’s going to affect email metrics because obviously, no bot is going to actually read your email and buy. All of those things are going to massively impact your budget decisions.”

As more marketers strive towards data and performance driven models, it is vital to gain an understanding about whether traffic represents genuine users or bots, and how these bots might affect data and thus decisions based on it.