Weeding through the mess of video metrics

Everyone expects metrics when advertising online, and all publishers and ad networks claim to have the metrics necessary to prove their client's spending was appropriate and did the job. Whether that job was brand lift, customer engagement, driving traffic to a destination site or a consumer purchasing a product, the client is usually given an identical set of metrics.   

Today, we'll consider the metrics portion of branded video advertising online and provide some guidance on sorting through the morass of data so marketers can determine the success (or failure) of their video campaigns. We'll also suggest that the metrics that are generated should be tailored to the goals of the campaign.

The issue is not just about receiving data and metrics. It's about receiving the meaningful data and analysis to help correlate the data to the campaign goals. Otherwise, there will likely be sensory overload. Think about it: If a client is running a $100,000 campaign with an effective CPM of $20, and the campaign consists of a pre-roll, a companion banner and a set of overlays that run over the content, this equals three assets, each with a set of metrics. The total number of "impressions" for this campaign amounts to 5 million. For the pre-roll and overlay assets (if the overlay is click-to-video), this can include not just an impression (i.e., the ad was served by the ad server) but everything on this list as well:

  • clicks (for pre-roll, overlay or companion banner)
  • pre-roll played (from 0 to 100 percent)
  • overlay was clicked and video played (from 0 to 100 percent)
  • viewer engaged with the advert (other than clicking) and took some action, such as a mouse-over (applicable to overlays and pre-rolls)
  • companion banner or pre-roll clicks to website
  • conversion tracking from website based on banner or pre-roll clicks
  • overlay was dismissed
  • pre-roll was dismissed
  • the rate of pre-roll dismissal is increasing during the campaign.

OK, we'll stop now without completing the list with other possibilities, including additional metrics that measure the interaction rates during the campaign. Take the last metric, for example: The client would want to know if the pre-roll was annoying people and making them bail out, wouldn't they?

So, for a simple brand campaign with three assets, that's about 25 different metrics, each provided 5,000 times -- or over 100,000 beacons or tags, to use an ad serving phrase. Even when collated into a spreadsheet, it's just too much data. A post-campaign report showing that the combination of pre-rolls, overlays and companion ads reached 5,000 unique visitors and the clickthrough rate (CTR) was 0.4 percent is not doing justice to the money being spent.

What's needed is an understanding of the basic goals of the ad campaign. Then, during and after the campaign, the publisher or ad network should organize (and reduce) the data and provide summary statistics that are appropriate for the campaign. In addition, the data and metrics should not be provided in a vacuum, but in context. Here are a few hypothetical examples, and what to consider when measuring metrics:

1. The client is releasing a horror movie in three weeks and wishes to measure viewer interest in the trailers (three are provided). The client simultaneously wants to drive traffic to the feature film's destination site because past experience shows that traffic spikes drive a film's opening weekend box office numbers. 

  • In addition to the baseline impression and click numbers, summaries of which trailers are performing (i.e., viewers are actually watching the whole way through) on a daily and weekly basis will reflect whether the creative is doing well.
  • On which pieces of content or websites (used as a proxy for the demographic) are the trailers being watched the most?
  • Are particular trailers receiving a higher clickthrough rate to the website than others?
  • Is there a set of websites or demographic for which the CTR is higher?

2. The client has an athletic shoe release coming up and wants to blanket the airwaves -- or in this case, relevant online video destinations -- with its branding campaign. The client also wants as close to 100 percent share of voice (SOV) on pieces of video content (along with the usual roadblock banner/display elements for the web page to ensure SOV) that are relevant to its demographic of 18- to 24-year-old males.

  • Again, the basic reporting in terms of ad impressions, clicks and on which pieces of content the ad ran should be provided.
  • In addition, to measure SOV (rather than asking the client, agency, etc., to take the SOV on faith), the total number of video content streams should be provided. This way, the client can ascertain how close to 100 percent SOV it actually received. On a side note, this is not completely accurate, since overlays can have multiple impressions per stream. But for now, let's go with it.

3. The client is advertising for the release of a new video game that can be played either online or by using a dedicated system, such as an XBox, Wii or Playstation. The client wants to run an ad campaign that engages viewers and encourages them to use the product. The ad is a custom-developed Flash unit that has loads of interactivity and a free mini-version of the game.

  • The amount of time the cursor, mouse or pointer is moving over the Flash creative is an important factor in determining engagement. How long did the user linger? If the ad creative runs for 15 seconds, did the user pause the creative? Did they start playing the game and stay longer than 15 seconds?  
  • How often did the aforementioned behavior happen and on which pieces of content? Clearly, the content on which this happens more often is more relevant.

As you can see with the examples above, the metrics that can be measured with video are brand lift, interest, driving traffic, share of voice, engagement, interaction and many others. The list goes on. Clearly, one of the major advantages of advertising with video is the correlation that can be drawn between the client spend and ROI, because of the ease with which metrics can be reported. Due to an overload of data, it is critical to tailor the metrics provided to the campaign goals to judge campaign performance.

Jayant Kadambi is co-founder and president of YuMe.

 

Comments