How to minimize ad-serving discrepancies
Discrepancies exist throughout the ad-serving ecosystem
It’s a fact that discrepancies exist throughout the ad serving ecosystem from the advertiser’s own server to the ad agencies’; the Gourmet Ads ad server to the Publisher’s website. Find out why discrepancies exist, and how to minimize the discrepancies.
Discrepancies and their definitions: Ad requests versus ads displayed
From the outset it’s important to note that most Publishers compare a web analytics program to the Gourmet Ads Publisher Console statistics, which can be programmatic as they both track different things. For example:
Publishers count the ad requests (Page Views X Number of Ads Per Page) while the Gourmet Ads Publisher Console counts the number of ads requested as well as the actuall number of ads displayed (ie: fill rate)
Here are some ways to minimize discrepancies on your site:
1. Are the ad tags on every page?
One of the most overlooked issues with discrepancies is that the deployed ad tag could be missing on a few page templates. Templates often overlooked include search results pages, category pages, tags pages, and in some cases the entire blogs (where the blog uses a different CMS from the main website).
The Fix: Complete an audit of how ad tags are trafficked to ensure they appear throughout the site.
2. Trafficking ad tags globally
Gourmet Ads is able to monetize your website globally, so to reduce discrepancies, make sure that your ad tags are trafficked globally. Discrepancies can occur when you are only display ad tags in one country, yet your web analytics are tracking globally.
The Fix: Undertake an audit to make sure ad tags are trafficked globally.
3. Ad Blockers
A major source of discrepancies and an enemy to advertisers and Publishers is the ad blocker. There have been estimates across the industry that suggests that 22% of the internet has some sort of ad blocker in place, this has recently risen with ad blocking functionality being added to anti-virus software.
Why does an ad blocker cause a discrepancy? The Publisher’s website issues an ad request to our server, but the ad is prevented from being displayed by an ad blocker. The web page is then loaded but without ads, page impressions are counted as normal but revenue is not earned.
4. Publisher-side ad servers
For the vast majority of Publishers there is simply no need to run an Publisher-side ad server on your site. Not only are they complex to set up (large Publishers usually hire an ad operations expert to set up and maintain them) but they are more likely to cause at least 10% (if not 15%) discrepancy because of latency between ad servers.
Gourmet Ads only recommends that only Publishers with over 1+ million global impressions per day consider a Publisher-side ad server, and then, only if they are selling inventory in different counties to different vendors will it be worth the expense.
The Fix: If you’ve got an ad server and you’re doing under 1 million impressions per day we recommend that you simply hard code our ad tags to your website directly and avoid this discrepancy.
5. Latency – load time for the user
Gourmet Ads’ ad server is hosted at major data centers around the world and is backed on global content distribution networks (CDNs), thus providing quick loading of ads for Publishers. However, any lag in the connection between the actual ad request and the display of the ad on your site and in the user’s browser can create differences in counts. A classic example is web pages heavy with images and content where the user may have navigated away from the site or the next page before seeing the advertisement load.
We recently had a complaint from a Publisher who was saying that ads on her site were not loading. A quick look from our Publisher Support Team found that the homepage consisted of 42mb of content and internal pages where approx. 2MB – 3MB. Our ad tags were the last element of the page to load. Most people were not waiting for the page to load before clicking off. As a guide our maximum size per ad unit is 40k or less.
The Fix: Complete an audit to ensure that page content is optimized and page load size should be no more than 400k in total.
We also recommend that you place our ad tags at the top of the page code to ensure quick loading, thus maximizing revenue.
6. General caching
Modern day browsers are built to cache as much third-party code (like ad tags) as possible, thus reducing the need to call every element of the site. Add into the mix that some Publishers apply website caching (such as WordPress Plugins like W3 Total Cache or WP Super Cache). When the advertising creative is cached in the browser or on a proxy server, no ad request is seen by the advertiser’s server, which results in impression count differences.
The Fix: If you are having discrepancy issues we recommend not caching your site.
7. Cache busters
If you’re using a Publisher-side ad server, you may need to consider a cache buster. A cache busters is a simple dynamic piece of code, usually a unique number, to force a new call to the third party, called a cache buster. Gourmet Ads doesn’t include this by default in our ad tags, as the majority of publishers don’t require it as they don’t use ad servers. You will be required to implement a cache-busting macro in your Publisher-side ad server when trafficking our ad tags into your ad server.
The Fix: Publishers who require a cache buster, should request this via Publisher Console Support
8. Time differences
Not always the case, but discrepancies can be easily seen when ad servers operate on different time intervals or geographic time zones, which results in temporal differences. For example the Gourmet Ads server is set-up to report on the US East Coast (EST) however if your reporting is set-up on GMT or Pacific Time or another time zone – there will always be a natural discrepancies.
The Fix: Ensure that Reporting is on the same Time Zone
Finally if you’ve tackled a few of these points and still seeing higher than 10% discrepancies, please create a Support Ticket from within Console and we’ll look into it.