>>JONA MICI, A 27-year-old media trader, sits in front of her screen at Varick Media Management, a real-time advertising company in New York, and explains how she uses superfast algorithms to buy 20m-30m advertising “impressions” a day. Today one of her clients, an American bank, has asked her to find new customers. At first she guides the algorithm to buy as many impressions as possible near bank branches. Then she narrows her targets, choosing criteria that produce a better hit rate. For example, she finds that tablet users sign up more often than iPhone users, and afternoon seems to be a better time than morning.

It sounds so hand-wavy. I'm guessing she (as well as the entire food chain from her keyboard onwards - the DSPs to the SSPs and the publisher websites) would be sleeping under a bridge if she was paid for actual conversions, rather than for just filling a purchase order of impressions with whatever she can get.

Sometimes I wish I could get clients with loose wallets like this...

...and what on earth does getting impressions "near bank branches" mean?

I agree that media trading is quite hand-wavy - similar to trading stocks.

That said, it's quite possible that the article mistook 'hit rate' for 'conversion rate'. That would explain why snooping for imps "near bank branches" would make sense. Those customers may be primed for the ad/can convert sooner.

Programmatic ad sales is established and growing rapidly and should be of bigger concern to people than the "native advertising" diversion people are misunderstanding lately.

Programmatic ad sales relies on collecting incredible amounts of behavioral data and letting advertisers target upon numerous criteria. Trying to get someone to sign up for your credit card? Perhaps try advertising to a 20-28 year old male who's hit up both a few debt forums and a few gadget review sites lately? It's all pretty amazing from a scientific POV.

What do you think people are misunderstanding about "native advertising"? Seems like a pretty simple concept.