Posted on Jul 19, 2008


The first ever SWAT Summit kicked off this last week in San Francisco. The aim of the conference was to help advertisers effectively enter and work with social networks.

Due to unfortunate scheduling, I was only able to make it for the last half of the one-day event (earlier in the day I was attending PSFK’s conference). My friend and colleague, Steve Hall of Adrants, had the honor of speaking at two sessions, one of which was The Science of Measuring Campaign Success (User Metrics and Engagement). The session included Ian Swanson (Sometrics), Kim Kochaver (Federated Media), Troy Young (VideoEgg), and Anna Banks (McCann Worldgroup). Steve grilled everyone on if the case studies they were presenting actually worked at the end of the day, and it was obvious that some of the panelists were agitated by this question.

The panelists discussed how demographic targeting/analyzing tools in Facebook were making it easier for clients and ad agencies to measure success in social media. I couldn’t help but raise my hand and ask if this was actually considered progress. By these standards, it seems like advertising is making little to no progress by taking the same solution (demographics, impressions, etc.) and trying to force-fit it into a new problem (social media). I’ve been ranting for a while that demographics are dead. It seems like advertisers are taking the easy way out by using traditional metrics and refusing to spend effort towards educating clients about what is relevant in social media. Additionally, I never once heard the panel mention the idea of building custom metrics based on social media analysis and relevance to the individual project.

I think it’s important for conferences like SWAT Summit to talk about these issues, but I wish there was more representation from the social media side (for the half of the day I was there, it seemed very advertiser-heavy).


  1. Jim Driscoll
    July 20, 2008

    It sounds like the conference was designed as a “look at all the money you could be making!!!!” effort: it’s no real surprise that it was packed with advertisers.

    As for the matter of demographics, you really need to define what you mean by “traditional demographics”. Grouping people at all by age/gender? The actual bounds of traditional age/gender groups? Trying to target contiguous ranges in the age/gender set? Other possible demographic data? It sounds like you have an interesting point of view, it’s just not entirely clear what that is.

    I’ve thought for a long time that what on-line (particularly social network, but ultimately anything where you’re logged in) could do with is “I like this” and “I don’t like this” links by each advert, providing trend data effectively allowing people to pick what adverts they see themselves; add to that a system akin to Amazon’s “recommendations for you”, and you could have very quick and effective advert auto-targetting. You could also, in social networks (assuming SOME level of feedback, even if that’s only whether an advert has been clicked on), use the friend data to bias a person’s friends’ adverts according to the person’s own preferences, which presumably are generally similar.

    But I don’t think any kind of intelligent targetting system of that kind is likely to be agreeable to the advertising industry, and that’s all because of numbers. Advertisers like to think “I’ve just bought 10,000 eyes” or whatever it might be. They don’t care if there are duplicates in the set, they don’t much care how many people actually see it (it’s a probability-driven industry), they may not even care if much of the set is outright disqualified from whatever the product is (eg. people outside the US getting US adverts), but they do really like firm numbers to base everything on – and that means the set of people who MIGHT see the advert. In effect, that requires some form of demographic data: you have to be able to say, in advance, that a certain number of people should see it, and provide reasonable assurances that most of those people are well targetted. I don’t think that the idea of quality over quantity, at that level, is conceivable to advertisers right now, even in the face of more advanced targetting such as Google’s AdWords (which still provides firm predicted numbers).

  2. giles palmer
    July 21, 2008

    I like your thinking, and i’m sure in the coming years, you will be proved right (and hopefully be given some visionary status!). Building custom metrics based on social media analysis however is
    1. very tough – as a conversation is not something that is easy to analyse and quantify in the first place and
    2. doing it relevant to the individual project sounds expensive and expertese-intensive which is going to put off almost everyone right from the off.

    However, it looks like you are right into this space – i’m going to read the rest of your blog and discover more….

    and Jim, that has to be the most thoughtful and thorough comment i have ever seen on a blog – kudos

  3. Scott Oppliger
    October 7, 2008


    Your comments are on target with respect to the “old-school” mentality of measurement as applied to successful marketing in social networks. It’s a Catch 22. Companies want to market in social networks. If they use traditional advertising methods, they can easily measure the success (or lack thereof). However, if they go about it correctly – using the medium as it’s intended to generate viral spread and word of mouth, it becomes much more difficult to measure success, therefore leading to the potential that a company could ditch it’s social marketing efforts due not to a lack of success, but rather, to a lack of available metrics (or metrics that they understand).

    New medium requires new tools, maybe even an entirely new way of defining success. Companies who properly use social networking as a vehicle should want to know what people are saying, who they’re saying it too and where they’re saying it. Those tools do exist, but old-school thinking hasn’t learned to shift from “click-through rates” to “velocity of talk”, etc.

    Scott Oppliger