Does Social Media Analytics Pass The Monday Morning Test?

Is it actionable, prescriptive, pointing you to the right next step? Or is it more in the realm of data summaries – descriptive, maybe even interesting, but leaving you wondering what to do next?

I have been curious about this question for a while so when the AMA invited me for a webcast titled “Social Media Analytics” by Coremetrics, I tuned in.

Well, to make a long story short, the answer to my original question is NO (at least based on this webcast). No Monday Morning Analytics here, just basic reporting.

The time spent on the webcast wasn’t a waste, however. I learned a few interesting tidbits about what online marketers are thinking about these days and how they are “instrumenting” their sites to collect more detailed data. 

• Social media sites like Facebook and Twitter seem to be considered by marketers as just another marketing channel, along with email, paid search ads on search engines, and display ads at third-party websites like nytimes.com or espn.com
• In all these channels, they run ads of different types and measure “conversions” i.e. how effective are the ads in enticing browsers/searchers to do something that the advertiser wants them to do e.g., buy product, sign up for newsletter, register.
• The key “new thing” here seems to be the realization that people interact with and are exposed to a particular brand on multiple occasions before they actually “convert”.
  • Scenario: Prospect views a video (produced by brand X) on a social site, gives it a thumbs-up, but doesn’t click on a nearby “learn more” button. Couple of days later, she types in brand X into Google and clicks on an ad paid for by the brand.
  • If we ignore the impression made by the video on the prospect, we will give all the credit for the conversion to the search ad.  We may even take budget from producing video ads and give it to paid search but that, of course, could well hurt our business.
• So what is the solution?  Measure “view throughs”, not just “click throughs”.
  • Embed Javascript tracking code (helpfully provided by Coremetrics and similar vendors, I suppose) into every ad. This is called an “impression tag”.
  • The first time a web-surfer interacts with any of your ads, a surfer-specific cookie is dropped onto their machine, along with an impression cookie (or tag) for that particular ad.
  • As the surfer interacts with your ads over time, every ad interaction leaves a trace in your weblogs. You connect up all the traces using the unique surfer-specific cookie and thereby build a clear picture of every surfer’s interaction sequence.
  • With the interaction sequence in place, reports can be run to figure out which sequences convert better and where to focus the ad spend.

In summary, the new thinking here is in instrumenting the ads so that more data is collected. The analysis itself seems to be basic reporting. That said, the new data opens up interesting possibilities for decision-oriented modeling and analysis.

Understanding the characteristics of “winning interaction sequences” would be one area where statistical models could be usefully applied.  For example, when several ads are part of the interaction sequence, we need a systematic way to allocate credit for the conversion. This may be complicated if (for example) seeing a Facebook ad followed by a YouTube video leads to higher conversion than the other way around. Without a systematic way to model the sequence data, it will be difficult to glean trustworthy (i.e. an accurate reflection of surfer behavior) and actionable insights.

Nothing that I have described above is particular to social media. It is just better ad impression tracking. I thought that perhaps I will hear something about how a Facebook discussion between two friends about brand X may now be trackable as part of the interaction sequence. I posed the question but the presenter didn’t get to it (likely due to lack of time – there were 1100 viewers on the webcast!).

Bottomline: No Monday Morning Analytics yet. New data, old reporting, nothing particularly social.