Are you being cheated on PPC advertising?
December 13, 2005
Another common stats question that comes up involves pay-per-click advertising (such as Google AdWords and Overture). Often, a PPC advertising system will show that it sent more traffic than your own statistics say you received.
Does this mean you’re being cheated? Maybe.
But maybe not.
For starters, you need to understand why comparing two stats packages of any kind will seldom if ever yield the same result. To learn why, click here.
That said, there are a lot of reasons (besides fraud on the part of a vendor) that pay-per-click vendors will report more sessions than any given stats package records.
Some of these are:
- Users clicking on an ad and hitting back before the page completes loading. This could be that they immediately saw something they weren’t interested in, or it could be because they’re on a slow connection (or the page is bloated and loads slowly, particularly on users with slow (dial-up) connections) and they got tired of waiting, etc.
- Users clicking on an ad, reading the page (or not), hitting “back” to the originating page, and then clicking the ad again and returning within a timeframe that the stats package considers to be a single session. For example, I’m on a page with five ads for red widgets — I may click one, go back and click a second one, a third, etc. and eventually come back and click on the first ad again to check something for comparison.
- If an ad campaign is aggressive, and the user is persistent on searching on a given term, he may (unwittingly or otherwise) click on an ad from the same site multiple times on multiple pages. Assuming the advertiser is (wisely) varying his ad text, the user may well not realize this until the page loads. If these multiple clickthroughs occur within 30 – 60 minutes of each other, the web server may well consider the user as a single session.
- Malicious users can be a problem; an increasing number of users are well aware that an ad costs the advertiser per click — some users will make multiple clicks to be a pain in the ass. It’s also not unheard of for unscrupulous web marketers to click multiple times on competitor’s ads in order to drive their costs up.
The first three items can be largely addressed by good page design and marketing planning — make tight, fast-loading landing pages and marketing funnels for each term, and combine that with server-side scripting to try to insure as complete a picture as possible of where each user came from and what they did.
For example, instead of sending all of the users to “www.site.com/landingpage” (or worse, just “www.site.com”), send users from AdWords ad one to “www.site.com/landingpage/a1” and users from AdWords ad two to “www.site.com/landingpage/a2”, users from Overture ad 1 to “www.site.com/landingpage/o1”, etc. Use unique urls not only on each landing page, but on every link in the marketing funnel to see exactly what the user is doing — if he clicks one, you can record that via server side scripting, if he didn’t you can assume he hit back or closed the page.
A lot of successful marketers not only do this, but they also combine it with scripting to send a percentage of each clickthrough on each ad phrase to separate landing pages to determine which pages work best with which ads.
This can be more expensive and time consuming than just putting ads out there and seeing if the user buys something, but it can also tell you a lot about what you are and aren’t getting for your money, and what approaches work best with each source of user.