Whether out of simple ignorance or calculated deception, publishers sometimes misrepresent the traffic to their Web sites. These inflated figures can give a false impression of how popular the site actually is;resulting in highly skewed metrics for marketers or prospective buyers. Eric Shanfelt, Penton’s vice president of eMedia strategy, says that there are at least a half-dozen tricks that if properly examined should raise red flags.

1. Hits Per Month
“I can’t believe that some sites still try to tell advertisers how many hits their Web site gets a month, but it happens every day. A hit is simply an object being pulled from your Web server. Every graphic, every HTML component, every javascript file, even the style sheet;every one of them is a hit. A single view of a single Web page can have 20 to 30 hits. You had a million hits to your site last month? Hope all 400 visitors enjoyed them. (400 visitors x two visits per month x five pages per visit, x 25 hits per page.)”

2. Impressions Per Month
“This isn’t necessarily a bad metric since views of an ad (impression) are often the base currency of Web advertising. But it’s terrible for evaluating how big a site is. Let’s say I have a Web site with 200,000 page views per month with a 728×90 leaderboard at the top and a 160×600 skyscraper on the side. That would be 400,000 impressions per month. But this month I decide to add two badge ads to the bottom of the page. Voila! All of a sudden I have 800,000 impressions per month and my traffic didn’t increase a bit although my share-of-voice for each advertiser just decreased.”

3. User Visits Per Month
“This is another favorite tactic of Web publishers. Rather than report unique monthly visitors, they like to report user visits (also called user sessions) per month. Why? Because typically people come back two to three times per month (and a lot more often on some sites like forums and blogs) so reporting user visits makes me look like I’m reaching two to three times more people than I really am. Remember that visits is a totally valid metric, but be aware of the very significant difference between visits and unique visitors.”

4. Daily Uniques Versus Monthly Uniques
“This is probably my favorite and the most subtle way that a Web publisher can boost their stats. Many metrics applications report monthly unique visitors and daily unique visitors. A daily unique visitor is exactly that; a unique visitor for that specific day only. Come back tomorrow, however, and you are counted as another unique visitor but for a different day. Some publishers add up all their daily uniques and report it as unique visitors for the month. The problem is that this doesn’t de-duplicate visitors who came back on different days. What you need to specify is monthly unique visitors and make sure that they don’t add up daily uniques. The difference can be as much as 15-20 percent.”

5. Improper Filtering
“I’ve been through many Web site acquisitions and I never take a Web site owner’s claims of traffic at face value. They’re not trying to deceive anyone, but often Web publishers use metric systems that don’t filter out spiders (also called automated agents or bots) from their metrics. Search engines and other Web tools go out and check Web sites now and then. This is non-human traffic and can account for 50 percent or more of a Web site’s traffic, and should not be reported as real traffic. Publishers should also filter out traffic from themselves. I know that I’m paranoid and always looking around our Web sites, so my own traffic would probably boost Penton’s Web traffic by 10 percent if it wasn’t filtered out.”

6. Reporting Registered Users
“It’s impressive to see all those millions of users to a Web site. But how many of them have been active on the site over the past month? How about the past year? Marketers beware. Unlike controlled-circulation magazines that must requalify their circulation periodically, most publishers never requalify registered Web users.”

For more thoughts on Web site development, visit Eric’s blog at emediastrategist.com/blog