#TraderTalkTV: Mark Zagorski, CEO of Exelate, Explains Real-Time Data Modelling

Interesting video and great to see how custom segments are developing.
The idea of real-time look-a-like optimisation suggests that an inaccurate look-a-like might be created in the first place though:
If the correct common attributes distinguishing converters from browsers (or the rest of the internet population) are identified from the 10k converting users (the seed) to create the look-a-like, then users that subsequently convert (used for the real-time optimisation) should be the same as those at the top of the funnel, in the look-a-like… your look-a-like should therefore remain unchanged… If they are different, your look-a-like isn’t working/delivering performance.
The key to look-a-likes is not the number of users you see, but the number of data points per user you look at – this helps you distinguish the subtle differences between converters and browsers.
That’s not to say that real-time optimisation isn’t valuable – an audience changes over time but it’s generally other media activity with different target audiences that causes this.

Very true – multiple look-a-likes should be produced from multiple seeds including converters, downloaders, crm, offline customer profiling etc