Yesterday I briefly presented at eMetrics the results of an independent test that was undertaken to audit the accuracy of the technology behind NextStage Analytics. For those who couldn’t attend, the accuracy of our predictions in this test were between 98% and 99%.

I’ll be publishing a White Paper explaining this test next week. In the meantime, I wanted to explain a little bit more about the value proposition that NextStage Analytics will provide in the near future and specially Age and Gender predictions.

One question that was addressed to me after my presentation was ‘How actionable are Age and Gender predictions’? My canadian friend Jacques Warren also made a similar comment through twitter in which he said:

OK and are gender and age predictive of anything?

So I wanted to respond as I guess that other people might have the same kind of questions.

First of all age and gender aren’t predictive of anything really actionable per se (well, maybe it would allow me to know that I’ll be able to market you some Viagra in a few years ;-) ). More seriously, the purpose of my presentation was to demonstrate the accuracy of our predictions. Let me give you an example of how you will be able to use NextStage and its ability to predict gender.

So let’s take for the sake of this example that I’m the CMO of a brand that targets women.

So I’ve built a great new website about my products targetted to females and I’m doing advertising in 2 different websites in order to attract traffic to my new website. Let’s imagine that the first month I spend 50.000$ in online advertising evenly in the 2 referral websites to which I pay a CPC of 0.1$. This means that I’ll get 500.000 visitors. The question is then, are they my target? Can I optimize my efficiency thanks to gender predictions?

With NextStage Analytics I’ll be able to gauge the gender breakdown for the referrals sent:

Referral A: 30% females 70% males

Referral B: 90% females 10% males

Based on these numbers I realize in fact that even if I’m paying 0,1$ per visitor that comes to my website, but the ‘cost per woman’ is not the same per referral. So let’s look at how many women each referral is sending me and then what’s the ‘cost per woman’ sent to my website:

Referral A: 75.000 women - Cost per woman: 0.33$

Referral B: 225.000 women - Cost per woman: 0.11$

So as you can see, in order to have a woman coming to my website, I’ll need to pay 3 times more to Referral A than B.

After this first campaign I want to optimize my results, the second month I’ll switch then all my budget to referral B, meaning that with the same budget, I can get 450.000 female visitors instead of 300.000.

This means that without increasing my budget, I can increase by 50 % the results of the number of women coming to my website.

Simple and actionable - isn’t it?

René

P.S. You can also read a recent blog post from Joseph that presents another example of how Marketing is a science and how NextStage and its ability to predict gender can help.

Original Comments

Comment 1

Author: Christopher Berry Date: May 9, 2009 Text: I’ll add:

If you have more females than males coming through a specific page, it would make sense to start making changes to the creative in a way to get more females to convert.

It also has implications for personae development. It might very well be that more females, of a particular age, are researching more on the site and then leaving than males, of a particular age, who are converting. These findings could have a very large impact on how the site is tweaked or redesigned.

Very frequently, we make assumptions about who is coming to our websites. The data from Next Stage Analytics will shatter many of those assumptions, resulting in some directly actionable insights, which, if executed upon, will improve the user experience and frequently the performance of the website.

Actionable indeed!

Comment 2

Author: René Date: May 10, 2009 Text: You’re absolutely right Christopher, thanks for adding your thoughts and for being such an early fan of our technology ;-)

And as I said, age and gender are just two of the many variables that our technology can predict, so imagine all the new possibilities for website designers and owners… There’s room for evolution!

Cheers,

René

Comment 3

Author: Jacques Warren Date: May 10, 2009 Text: Hi René,

As much as appreciate the example you give, my first reaction would be that you should have done a better job at ad placement ;-).

From what I’ve heard, your technology would certainly made those dimensions readily available without a marketer resorting to a survey, or panel-based solutions. Then there’s price of course.

I certainly look forward to seeing what other variables you are able to uncover.

Comment 4

Author: René Date: May 10, 2009 Text: Bonjour Jacques,

I confirm that our technology makes these dimensions available without the use of a survey or any declarative method. The technology invented and patented by Joseph is so disruptive that I’ve chosen to really take baby steps for disclosing its power. I understand that savy users like yourself might feel these steps to be too careful so please stay tuned for the next posts that will be unraveling other dimensions.

Cheers,

René

Comment 5

Author: Jacques Warren Date: May 10, 2009 Text: Salut René,

Yes I understand that it does it without a survey. What I am saying is, as for gender and age, how beneficial it will be to use your technology, which I suspect will be expensive, compared to using a low cost survey. That is why I really look forward to your announcing the other variables you can capture, because age and gender are of too low value to lose sleep.

But I’m quite confident you guys are going to surprise us out of our wits!!

Comment 6

Author: Sébastien Brodeur Date: May 26, 2009 Text: Working for a financial institute, I realize quickly this could help us identify frauds.

We already know the gender and age of our visitor (online banking need authentication.) With that information and NextStage Analytics (ET) we could detect people trying to steal money from their phising victim. If a male of 30 year old is accessing the account of a 70 years old female, something smell fishy here. Let popup a question before allowing the visitor to continue.

Ah, what a wet dream :-)

Comment 7

Author: René Date: May 26, 2009 Text: Bonsoir Sébastien,

You’re right, security is one of the applications of Evolution Technology. We had in mind as one of many possibilities that our technology could be applied to ATMs. this would allow to see not only if the person was who he really said he was (the cardholder) but it is also possible to sense if the person is in his ’normal’ state or not. For example we could see if the person was under stress (maybe a thief is forcing him to do the withdrawl?). But we don’t have to forget to allow certain flexibility. The technology could be applied as you suggest, but if I can go a bit further, let’s imagine that I’m one day helping my grandma doing some homebanking transactions. The system would recognize that I have between 30 and 35 years and not 95 as my grandma. Instead of blocking the system we would suggest that the system challenges with extra questions in order to see that my grandma is really behind the computer even if she’s not controling it.

It is not a wet dream, just a nearby possibility… If you want we can discuss and make some scenarios to see how NextStage can help you (rdechamps (at) nextstagevolution (DOT) com). Let me know.

Cheers,

René