This list includes some of the most common questions about Verification and Reporting.
If you are serving with Sizmek, Verification is built in and no additional tag is necessary; you need to select a check box on the account, advertiser, or campaign level. If you would like to track Verification on placements for which Sizmek is not the ad server, you must ad an additional tag.
The platform updates data every 24 hours; however, since all of the data may not be complete, the last available date from which you can view a full day's worth of data in the Verification dashboard is two days back. Note that data for the Verification dashboard is stored for 180 days; data for the Excel report is stored for 18 months.
Verification can be enabled on a campaign, account, or advertiser level.
In the data table, you can see metrics organized by site. For example site AA had 202,386 video impressions, 95.92% of which actually started the video. Yet we also see that 60,825 of these impressions were AutoPlay. This might lead us to believe that 30% of all impressions were AutoPlay (60,825 / 202,386). But wait! We also see that the Video STR is 82.41%. So before we draw conclusions about how video attribute data is representative of the whole campaign, we have to remember that the numbers we have are only based on 82.41% of the total. In this case, that is 166,786 impressions. Therefore, impressions with AutoPlay weren't actually 30% of the total. They were more like 36%! (60,825 / 166,786). The lower the STR, the stronger an impact this will have.
The Sizmek Verification algorithm can assign multiple negative categories to the same page.
For example, a page contains a story about a car crash during an earthquake. The negative categories in this page include Accidents (car crash) and Disasters (earthquake). If an impression appears on this page, the algorithm increases two counters for this impression - one for Accidents and one for Disasters.
Unsafe Impressions is a metric that eliminates overlap when counting multiple negative categories for the same impression. This metric considers all negative categories (except ghost sites). For the page in this example, the Unsafe Impressions count is one.
Yes. Site and Domain level data are available through the user interface or by using an Excel export.
There are several factors that impact whether the Sizmek engine can provide meaningful verification data.
Providing meaningful Verification (STR) requires that the ad server be able to “see through” to the page’s URL. But, the use of cross-domain, nested iFrames can prevent 3rd party ad servers from extracting URL information, resulting in no classification.
STR may be affected by the page's structure. Some page URLs do not have enough information to fully classify the page (for example, limited or no text or error pages), which may result in no classification and will not be calculated towards STR.
URLs with very few impressions will neither be processed nor classified.
Other reasons for low STR might include:
very old browsers
browsers that do not support the technology of the ad (such as Flash ads to mobile devices or HTML5 ads to legacy mobile phones)
users who exited the page before the ad fully loaded
In general these should be aligned, since impressions that are "recordable" have to do with technical limitations and not customized thresholds (as is available for viewable impressions). However, since the engine collects recordable impression counts as separate interactions, it is possible that one registers ("records") an impression and not another. For example, if a user leaves a page or closes a browser in a specific moment, one interaction may have had a chance to fire but not the others.
There are distinct areas in which Sizmek determines player position.
Side/Center logic: If more than 2/3 of the player is in 30% of the screen width from the sides (30% from the left or right of the screen sides), the player will be considered Side. If more than 2/3 of the player is in the middle 40% of the screen width, player will be considered Center.
ATF/BTF logic: If 50% or more of the player is above the user's screen fold, player will be considered Above The Fold. If 50% or more of the player is below the user's screen fold, player will be considered Below The Fold.
The engine reports Video Start any time the video player starts. Impressions with Video Start is a unique metric that is only fired for each user once during each session. Examples:
The engine does not count auto-play in the Impressions with Video Start metric, but does count it in the Video Start metric.
When a user starts the video player multiple times during one browser session, the Impressions with Video Start count increases by one, but the Video Start count increases every time the user played the video.
Note: The Video Start metric is available in the CRB (Analytics), but not in Verification reporting.
With regard to domains, "Unrecognized" includes cases in which a domain is not visible or unknown (impressions with a blinded URL). In the verification dashboard, unrecognized domains appear as "Unrecognized."
Sizmek completely removes server-level bot traffic from all reporting, so that data is neither visible, nor billed to the client. Ad Stacking is removed from viewable counts, and is reported separately. Ghost Sites are included in viewable impression counts.
Sizmek counts Ad Stacking only if another ad completely (100%) covers our ad. The case of expandables may be included if the ad is expanded and fully covers one of our ads.
Ad Collision (Same Placement) identifies placements that run on the same supplier.
Calculation: Ad Collision (Same Placement) = Multiple impressions from the same placement (Sizmek placement) on the same webpage at the same time (which necessarily means, same supplier or Sizmek site)
When determining ad clutter, Sizmek considers various factors such as the number of ads that are in view simultaneously, the ad density, the spacing between ads, and the average screen share. There is no fixed threshold. Sizmek does not share the exact calculation and weight of each factor due to proprietary technology.
We recommend waiting a couple days after the campaign goes live to make sure there is enough information to indicate trends. After that, we recommend checking regularly but at least once per week. This allows for mid-campaign course corrections if necessary.
You can access data that is more than two, complete days old; you can view this data for up to 180 days.
High recordability rates ensure that advertisers can get the most expansive view on Viewability. It is worth relaying plans to Publishers so they can do everything possible on their side to improve the recordability rate.
One thing that prevents an impression from being recordable is the publisher's use of blind iFrames. If publishers have received rich media certification, they will have the Sizmek addineye file on their domain, which allows us to "see through" blind iFrames.
The Verification Suite supports an unlimited amount of publishers, so there is no minimum to the number needed to perform verification.
Since Verification is integrated into the Sizmek tag, there are no differences to enabling the tool by using networks or premium buys.
It is a best practice to inform the publisher when booking the campaign that the campaign will track verification and to highlight the expectations around delivery.
If you have expectations about viewability or contextual targeting, this should be made clear at the start so that the publisher can be sure to secure the proper inventory.
Once expectations have been set and the tags trafficked, the publisher can implement the tags accordingly.
There is a significant advantage to applying Verification to all campaigns because it serves the best interest of the publisher and the advertiser by eradicating wastage.
The agency should agree with the publisher who will be doing the monitoring; the recommendation is that both do it.
When a campaign goes live with Verification enabled, the first step after trafficking is to check that the campaign is live in delivery reports or campaign monitor on day one or day two. Here you are looking to check that impressions, clicks and conversions are being collected – this demonstrates that implementation of the tags has been done correctly.
It is a good idea to wait a few more days before checking the Verification Suite. This ensures the suite has had enough time to collect meaningful information to spot negative verification items. After all, you don’t want to be calling up publishers every day if the problem has only occurred once, and they have spotted it on their side and resolved it already.
The planner (who maintains the publisher relationship) is usually the one who reviews results in the Verification dashboard. Detection of poor brand safety or misaligned content classifications can be reported to the publisher contact. The domains and subdomains can be supplied to the publishers, who can ensure no further impressions are delivered to those pages.
With time, there will be fewer misdemeanors as the publisher gradually eradicates wastage from their inventory pools.