The pricing model for online display ads is shifting from cost-per-thousand (CPM) impressions to the more efficient pay-per-click (PPC). You pay only when someone clicks on your ads.

The problem with PPC is that anyone -- not just valid leads -- can click on your ads and raise your spend. The most notorious bad clicks are known as click fraud – even though fraudsters are not the only clickers who can inflate your ad spend.

Jon Myers, Head of Search, MediaVest, has been working in search and online advertising since 2001. He first became aware of click fraud in 2004, when he and his team realized they had to be more proactive about monitoring the quality of ad-generated traffic. He’s been studying click fraud ever since.

Most major ad networks -- Google, Yahoo! and MSN -- welcome your click fraud research. They even reimburse costs from questionable clicks they miss. Myers’ team regularly checks the quality of their clicks and often receives refunds, he says.

Here is how Myers approaches click fraud, spots obvious signs of fraudulent clicks, combats the problem and gets money back.

Problem: Expensive Yet Futile Clicks

Click fraud has two definitions: clicks intended to artificially boost ad revenue; clicks made for any reason other than the ad’s intended purpose.

The best way to fight click fraud is to try to eliminate all invalid clicks to save as much money as possible. So, we’ll use the latter, more inclusive definition.

Three sources of cost-inflating bad clicks:

Source #1. Competition

Your competitors may click on your ads for two reasons: research and spite. They have no intention of converting – they are merely conducting research at your expense.

In addition to familiarizing themselves with your landing pages, sign-up forms and other marketing materials, they may want to cost you a few extra dollars. If you’re in a competitive industry, for instance, and are paying $5 per click, a CMO at a rival company could click on your ad daily to boost your spend $150 a month.

Source #2. Employees

Not everyone, including your own employees, realizes that clicking on an ad costs someone money. Your employees might regularly find your website by searching for your company’s name and clicking on a search ad. This waste can be eliminated with a quick note to them.

Source #3. Fraud

This category is the headline-grabber you’ve heard so much about. Swindlers hosting ads will click them to generate more money for themselves at the expense of advertisers. Ad networks are more often targeted than search- or site-specific ads.

There are many variations on the theme. The entire process may be automated. A computer program can regularly churn out no-content websites, subscribe to an advertising network, host the ads on the websites and click the ads. Or the process can involve complex networks of computers or low-wage workers clicking on ads in distant countries.

Whatever the means, the result is always the same: Marketers are charged for clicks that waste dollars.

Solution: Monitor Click Fraud

Invalid clicks waste a percentage of your marketing dollars. Myers has seen a consistent range of 5% to 12% of most companies’ online media buys going to questionable clicks – with occasional spikes. That number can rise to 20% in highly competitive industries where CPCs can breach $10.

“If you’re in a financial-based sector, absolutely review your traffic,” he says.

You’ll never be able to eliminate all click fraud, but you can prevent, identify and report a portion of it. Let the size of your marketing budget dictate how often you check your traffic’s quality. There’s no excuse for not checking.

-> Your reports

You must prove the quality of the traffic driven by your online ads. The higher the click quality, the more likely it is that clicks are coming from valid customers and leads.

Myers says a good backend log file is your best option, but you can also use high-quality analytics packages or canned click-fraud solutions offered by vendors. Make sure that whatever system you use will report the following information:
o Time and date of click
o Referral URL and/or keyword
o Clicker’s IP address
o Total traffic (in aggregate and by site/keyword)
o Conversion rate
o Cost per conversion
o Week-to-week, month-to-month, year-to-year reports

Note: For major networks, you want the second-tier URL. For example, Google might report that visitors from its content network are “visitors from Google,” even through the ad was hosted on a partner site. You need to know the partner site to determine if the traffic it provides is good or not. A good analytics package or backend log file will reveal that second site.

-> Networks’ reports

The major ad networks are concerned with click fraud and are actively combating it. Some will provide “invalid-click” reports, telling you which clicks they determined were invalid, and whether they were removed before billing or refunded.

These reports are a good indication of the level of click fraud in your market. If you’re regularly receiving big-ticket refunds, you need to investigate your traffic quality. The networks will not catch everything.

Four preventive measures for questionable clicks

Measure #1. Avoid ad networks

“[Click fraud] is on the content networks more than the search networks without a doubt,” Myers says. “If you were to advertise purely on Google, Microsoft and Yahoo.co.uk, on the top level of the main search engines, the percentage of click fraud, in my eyes, would absolutely be negligible.”

Measure #2. Get your competition’s IP addresses

An IP address is like a website address. It’s a unique series of numbers that identifies a computer online. If you can learn your competitors’ IP addresses, you can set up filters to be notified when they’re clicking on your ads.

It might be hard to find your competitor’s IP addresses. What you can do is look up their website’s IP address (see tool below) and hope they surf the Web from the same server.

Myers suggests contacting competitors that regularly click your ads. A simple conversation could be enough to end the problem.

Measure #3. Get your company’s IP addresses

Create filters for your company’s IP addresses, too. They’ll alert you to when a coworker is clicking on your ads, and costing the company money. A simple memo explaining how online advertising’s pricing model works should be enough to remedy the problem.

Measure #4. Have a tight campaign

The more focused your campaigns, the less trouble you’ll have. Fraudsters want as much money as possible, so they’re targeting big-money keywords. Keep your campaigns as relevant to your business as possible. Stick to niche keywords over broad ones. However, you may be stuck if broad keywords are your biggest performers.

You’ll notice which keywords generate more suspicious activity after you monitor traffic quality over time.

Five red flags

The typical fraudulent click comes from a region outside of your market, has an extremely short session length, does not convert and comes from a source that frequently sends bad clicks.

Clear signs of trouble can emerge from your reports. When you find a red flag, focus on the traffic’s source and investigate if there is a problem and when it started. Visit the referral website. Look at the history of the traffic the page has sent. Save all your research to report to the network.

Watch out for these five red flags:

Red flag #1. Conversion rate plummets

One of the biggest indicators of fraudulent clicks is a nose-diving conversion rate. A sudden, significant drop means that you’re getting a lot more clicks that are not converting.

Red flag #2. No conversions from a site

Once you can see each click’s referring URL, you’ll be able to determine the quality of traffic each site sends you.

“If we’ve had a thousand clicks from that URL and no conversions, we’re obviously going to question the quality of the traffic and investigate it,” Myers says.

Red flag #3. Visitors from distant regions

Local marketers should be concerned when receiving traffic from Bangladesh or Moscow or other foreign locales. If your marketing material doesn’t reach foreign lands, your customers shouldn’t be coming from foreign lands. You can find your visitors’ locations through their IP addresses, reported in your backend (see tool below to convert an IP address to a geographic region).

Red flag #4. Drop in average session length

Your website’s average session length is a great click-fraud indicator. With CPC ads, a fraudster does not intend to stay on your site. They click, generate the revenue and leave immediately within a few seconds. That will cause your average session length to drop and indicate that something fishy is happening.

Red flag #5. A sudden performance dip

If any metric associated with your online advertising takes a sudden unexpected hit, you want to dig deeper and find out why. Find the keyword, website, region or time of day causing the problem and report it to your network.

How to report click fraud

Reporting suspected click fraud to major networks is easy. They make a boatload of money on their network, so they want to ensure its quality. Smaller networks may give your more pushback.

You do not need to explain a lot when submitting a report, Myers says. Give the network the referring URL, the number of clicks sent and the conversion rate. Put it into a spreadsheet. If the rate is extremely low, you’re likely to get a reimbursement.

Useful links related to this article:

Charts on Identifying Click Fraud:
http://www.marketingsherpa.com/cs/clickfraud/study.html

Data on search marketing click fraud
http://www.marketingsherpa.com/article.php?ident=29768

DomainTools: Whois – Find a website’s IP address, location and more
http://whois.domaintools.com/

Yahoo! Traffic Quality Center
http://searchmarketing.yahoo.com/trafficquality/for_you
php

Google Invalid Clicks Report
http://adwords.blogspot.com/2007/02/invalid-clicks-goog
es-overall-numbers.html

MSN Invalid Clicks
http://advertising.microsoft.com/microsoft-adcenter/faq
/invalid-clicks

MediaVest
http://www.mvmediagroup.co.uk/