DML – Demographic Markup Language

The Current State of Demographic Advertising

Are you sick of all of the irrelevant advertising you get on various websites like Facebook, YouTube, Hulu and blogs? On Hulu I’ve set my gender to Male and birthdate to 1973 in order to get advertising that I actually like. This is because I like ads for technology and ads with funny jokes in them. I’ve begun to do this on many other websites so I can get ads that aren’t as obnoxious. The issue is that my interests don’t match what I am on paper. I’d like to be able to define them myself, and I’m sure many would agree.

DML – Demographic Markup Language

There needs to be a markup language so that you can define your own demographic. No more irrelevant online ads. Did these one-size-fits-all-thinking people get an online marketing degree? Brian Ledger, brought this concept up today. It made me happy to know that other people are experiencing this annoyance.

If you change your demographic, your entire user experience could change for the better. Content would be more relevant and interesting, and the content you received would not be based on very small data points such as age or sex. Those two data points often skew people into groups that don’t mean anything at all.

What there needs to be is standard for defining demographics. There needs to be a demographic markup language and then a query language on top of it. For the user, this would be an extra set of metadata one could attach to their profile that would inform ad engines of the type of ads you’d like to see.

Brian pointed out another concept that might be interesting. “What if you log into your own demographic and say “this isn’t me” or change it? One one could keep their own demographic file and preview the type of ads you might receive because of it, tweaking it until it was relevant”.

What Next?

I’ll be writing more on this later and hopefully working with Brian Ledger on how this might look. What do you think? Would you use something like this if it meant you could curate your ad experience to something more relevant? What we need is a textbook that defines a certain vocabulary to use and map that into a subset of XML.


It might make sense to have DML be a part of as a sort of demoformats. Reasons for this is because microformats are already known and used, and demoformats would simply be another extension or option set within microformats. Another curious thing to look at is

9 thoughts on DML – Demographic Markup Language

  1. Great Concept! I for one would definitely use it. The way I see it there two sides to this problem. One is gathering appropriate demographic data intelligently. The second one is the analytics. Is there enough understanding to around what to do with this data? The traditional methods (both, prescriptive and deductive) in ordinal and cardinal formats are ineffective. One of the reasons for this is mostly non-sequential, isolated data points are use to build a demographic information. Any thoughts on that?

  2. I think a way around that is to request some general information first, and then get enough data to make it granular over time. is already doing this by asking for your preferences for ads in the beginning, and then allowing you to choose whether each ad displayed is relevant to you or not. I think a combination of general and specific voting/choosing is important, as the general higher order data is more a good overview and provides fuzzy data, while the specific vote-per-ad provides granularity and better categorization of content. I think it would work best as an emergent system that improves itself over time vs. a tightly structured system from the beginning.

  3. You are definitely right about emergent systems. I think there is a higher order of complexity that needs to be understood to build an emergent system. There are many examples of this in a “Closed System” for example current webanalytics, RDBMS, phase lock loop (in analog circuit design)etc..But when it comes to “Open Systems” these methods have not worked example current stock market crash caused due to lack of visibility provided into unforseen variations from traditional quantitative heuristics.

    I am comparing the variations of stock markets to the change in what people like over a period of time. Therefore, drastically reducing the viability of data to produce categorization. What I like today, I won’t like in another 6 months. This can be attributred the reducing product life cycles resulting in new things coming to market faster, there by reaching the saturation point quickly.

    Some food for thought. How does one accomodate this in an emergent system. I am asking these questions but I don’t have answers either…just thinking out loud and get your insights and hopefully provide mine too.

  4. I really like the idea. We’ve been researching exactly this area recently and I think the issue is wider than just demographics. If you’re trying to build and kind of meta-matching model, then you need to be using standard data sets for the ad server, content etc that you’re matching to consumers.

    Take music for example. How do you define the audience for motorhead using traditional demographic criteria? I saw them play recently in London and as a marketer, looked at the crowd to ask “how would I reach these people?” The answer has nothing to do with sex, age, education, income level or even Harley ownership 🙂

    Please keep me updated with your progress and if I can help….

  5. There’s something to be said for Facebook’s method, which ranks ad relevance based on your “Like” input. Certainly has the potential to become more accurate over time, as long as you’re not just wantonly deleting ads because they’re ads.

  6. I just sent this post to Klout, who had signed me up to receive some perk that had chicks with glossy hair all over it. I explained that they had me in their database as a female, and that nothing they could send to me would ever match me interests. I kindly asked them to change my gender to male in the database so that I could get tech offers or at least something mildly useful to my existence.

    It will be interesting to see if a response appears!

  7. Extraordinarily beneficial many thanks, I believe your current audience might want significantly more writing like that carry on the great effort.

Leave a Reply