The Puki Blog

Data, Analytics and Open Source

The 21st Century is a fun time for data analysts...

OLD Paradigm: Guy/Gal gets a brilliant idea. An ad is created. More money, focus groups are asked stupid questions about their 'purchase intent' (YUK). The ad makes it to air. 'Its 'doing well' for a few months (again, focus groups). Another 6 months later, Goosebumps Analytics comes along and analyzes the commercial with a mix model. The ad ‘targeted’ people. Sort of. We ran the add on 'Ellen' - so we are only getting "Ellen" viewers right? Eh..Kinda. Our regression had to be fuzzy: data were collected at a market level on both sides of the equation.

NEW Paradigm: Quite quickly, an ad is fielded on FB. Initial outlay, $10,000. It gets a decent click rate. The ad is tweaked, the dog changed to a cat. Another Test, $10,000, click rate doubles! Yes, we are onto something. Week three (yes, that's right, we are in week 3). A full $million dollars is spent, we look at the click to conversion, it is good – ROI on the program exactly $1.10. Okay. Our targeting is much better (not perfect, not perfect, but better) The new paradigm gives advertisers basically real time data on what is working and what isn't. Set the Social Media ‘buzz’ aside for a moment. One of the most important reasons advertisers will go social is the data. So while I might sound like a corporate hack for Facebook - Facebook is not so much what intrigues me - it is more what they are doing RIGHT that is important.

By the by, have you ever seen an TV ad for Facebook itself?


We work like dogs