Tip #40: Understanding “page views” and OpenX impressions

Newcomers to OpenX often ask a seemingly simple question: “Why don’t I have the same number of impression in OpenX as page views in my website stats program?”

As is often the case, the answer is never quite as simple as the question, and you may need to look in a few different places before reaching the “aha!” moment where all becomes clear.

Number of ads per page

The most obvious starting point in the page views vs. OpenX impressions conundrum is to check how many ads you have per page on your site. If you have more than one ad per page, then you should have more OpenX impression than page views. If you have some pages without any ads, then you should have less OpenX impression than page views.

Of course, if you have some pages with one or more ads, and some without, or pages with varying numbers of ads per page, then obviously, you’ll need to do a little bit of maths to figure out your expected ratio of OpenX impressions to page views!

This might all seem pretty obvious, but it’s still worth noting, as it’s often overlooked.

Drop off

In the event that you have consistently less impression in OpenX than you were expecting relative to your number of page views, then drop off could be the culprit. Drop off occurs when a user clicks on a link in your page content, clicks the back button on the browser, or maybe even goes to a completely different page or even site before the ads on the page have loaded from OpenX and had their impressions recorded.

Now, some level of drop off is normal, and to be expected. You will never be able to eliminate drop off entirely, so please don’t waste your time trying!

However, if you are suffering from a high level of drop off — that is, very many less ad impressions than you were expecting — then there are three main possible reasons for this:

You can look at the difference in requests vs. impression in OpenX to help you diagnose drop off issues.

Web log analyzer issues

Web log analyzers look at web server logs to generate the number of page impressions on your site. If you are using a web log analyzer, and you are seeing a ratio of OpenX impression to page views that don’t make sense, then the following are all worth investigating:

  • Check that your web log analyzer is working properly — is it processing all your log files? Is your web server logging everything that it should be?;
  • Is your web log analyzer correctly excluding items from your web server log that are not page impressions? That is, are all calls to images, CSS files, JavaScript files, etc. being correctly excluded from page view counts? Most web log analyzers should do this by default, but it is worth double-checking; and
  • Finally, have you set up your web log analyzer to exclude all of the calls to the OpenX scripts from being analyzed?

  1. Ads placed at the bottom of the page are more likely to result in drop off. []

News: Security vulnerability in OpenX 2.8.1 and earlier

OpenX 2.8.2 was released a while ago now, and the release notes stated that:

We have completed a number of critical security updates to the ad server in OpenX 2.8.2 to reduce any potential vulnerabilities in the software.

However, it seems that the situation is more serious that this – in the past week, an actual remote code execution vulnerability has been announced as being in OpenX 2.8.1 and earlier.

Admittedly, based on the vulnerability report, it would appear that this remote code execution can only be executed if you (or someone with access to your OpenX installation) upload an image banner with a .php file name extension that contains embedded PHP code. If you are careful about what you upload as banners into your OpenX installation, it seems unlikely that you would be vulnerable.

Still, if you have not yet upgraded to the latest version of OpenX, you would be wise to consider doing so!

Tip #37: Read the OpenX “Direct Ad Selling Practices” study

A couple of weeks ago, OpenX released a study of direct ad selling practices, based on survey data from the OpenX community. If you have not read it yet, it is well worth downloading the study.

In particular, the report outlines five different techniques that publishers with direct ad sales income of more than $5 CPM generally use, and which may be worth considering using yourself if you are planning on selling your inventory directly. For smaller publishers, this is very useful and valuable information to have. Kudos to OpenX for releasing the information!

Tip #36: Change the banner cache value when testing

Today’s tip comes courtesy of Seagull Systems’ Demian Turner, who quite rightly points out:

“During testing you can set the cache timeout value to some low value like 1 second so that effectively the cache doesn’t get used.”

If you’re debugging banner delivery, or you want to test the results of changing settings like banner capping or companion positioning, then reducing the banner cache value can be a great way of seeing your changes “go live” much faster.

Of course, if you’re doing this on a production system, you may affect delivery performance, so make sure you keep an eye on your system performance monitoring tool!

Tip #33: Avoiding duplicate ads

If you have more than one banner on a website page, it’s easy to ensure that you never show the same banner twice with the OpenX ad server.

Provided you are using Single Page Call, or the Javascript zone tag, the Local Mode zone tag or the XML-RPC zone tag, then under Tag Settings on the Invocation Code tab, you can set the “Don’t show the banner again on the same page” option, and OpenX will ensure that the same banner is not shown more than once on a page.

The "Don't show the banner again on the same page" option.

The "Don't show the banner again on the same page" option.

There is also the “Don’t show a banner from the same campaign again on the same page” option for Single Page Call and these three zone tags, which can be used to ensure that every banner on a page will be from a different campaign.

Of course, you can use both options if you want to ensure that not only is every banner on a page from a different campaign, but also that there are no duplicate banners.