Geek's thoughts…. Purely Online Media

I came across a good post on Adify related to impression and click discrepancy which says some of the common reason for discrepancy, so thought of sharing with you

Understanding Common Causes of Discrepancies
 

• The upstream advertiser’s ad server, which might be an agency-oriented solution like Atlas or a boutique Rich Media content server
• The network’s ad server-Ad server used by the ad networks
• The publisher’s ad server, which may range from a high-end solution like DART for Publishers, to an open-source ad server like OpenX, to something homegrown

… and many impressions may be redirected yet again to someone else — either upstream or downstream or both.

During its travels, an impression may be subject to any or all of these common problems:

1. General Internet latency: From hangups at the ISP to slowdowns in the user’s home network, ad serving can be subject to all kinds of delays. That means the first ad server in the chain may well record an impression request, but this request may never get passed along to the upstream ad server(s). For this reason, the ad server closest to the user typically has the highest impression counts.

2. Low bandwidth: As much as broadband penetration has been increasing around the world, there are still many users surfing the Web with slower modems, adding to general latency issues.

3. Inefficient page design or ad positioning: When a publisher buries an ad under many screens of heavy images, it may not fully load into a user’s browser before the user clicks away from the site. These partial ad loads can cause discrepancies similar to general Internet latency.

4. Heavy ads: Most display ad designers optimize for low-footprint ads, for example, by using a 10KB JPG instead of a 100KB PNG ad file. Heavy ads tend to load more slowly, creating additional latency.

5. Different counting methodologies: Even on a platform like Adify’s, where our ad server has been formally audited according to IAB standards, two ad servers may apply robotic detection algorithms differently, with one rejecting a particular impression as “non-human” and another upstream ad server accepting it.

6. Human error: While the IAB has produced guidelines and best practices for workflow and campaign setup, these guidelines are still relatively new and may not be adopted uniformly across the whole industry. Even if they are, misunderstandings can always occur. For example, what happens if a campaign starts a few days late for some reason? Are the campaign’s impressions prorated over the remaining days, or is the entire campaign delivered in full at a slightly faster rate? Unless the advertiser and the publisher/network agree to the business rules, these operational issues can also lead to discrepancies.

7. Reporting time differences: While not a direct cause of discrepancies, when looking at reports from two ad servers to track the source of a discrepancy, it’s important to line them up hour by hour, day by day, and to ensure both reports are set to the same time zone. Reports for specific ads and line items should also be matched carefully. Otherwise, impression and click counts from two ad servers can appear to diverge wildly.

You can also check out my earlier article about Discrepancy