And it seems like there’s already competition for this… from AdaptiveAds, AdReady, and GoogleDisplayAds.

If the algorithms themselves are genetic, then it becomes extremely trivial.

That’s not to say it’s not a good idea – it’s quite clever, but replicating it would not be hard.

What about incorporating similar technology across the entire spectrum of ‘buzz’- advertising, marketing, PR, blogging, etc? Wouldn’t it make sense to use the same approach for blog headlines, for example?

Or what about genetic multivariate testing that optimizes for the combinations of features customers think are important in a product/service/app? There is a great book on that topic called “Blue Elephants”.

Well done folks. Just wish I had an ad with 1 million impressions so I could tinker with your software. Anyone know of great multivariate and A/B testing software for other domains where combinations of variables matter?

SnapAds isn’t some Photoshop gimmick — we’ve chosen Photoshop as it’s already in the workflow of the existing creative team, and it’s simply used as the input mechanism of choice. Most of our campaigns contain hundreds of thousands, millions or billions of possible creative permutations, a capability existing ad servers do not have, and far beyond simply A/B testing a set of 10 creatives side by side.

Having said that, you’re spot on in the rest of your comment. We optimize to whatever an advertiser considers a success, whether that be higher engagement, more clicks, or greater sales.

We can and do optimize to different data: geotargeting, demographic, etc, depending on what is available. While we are still able to see increases when optimizing to a site’s entire audience (as they often share a certain amount of preferences), greater gains are made when pushing down further and optimizing with additional information.

Hope this answers your questions. Feel free to reach out if you’d like to take this offline.

In designing optimization systems myself, sometimes the opt objective is clicks, sometimes time on site, sometimes lead conversions, sometimes back-end sales. What would be truly nice is if the ad version optimization algorithms can be designed to shape the funnel in a way we could choose.

The other comments about geotargeting or profile-based optimization are spot on–there is a high degree of probability that different audiences (in different cities, at different days of week, times of day, on different websites) will respond differently.

Taking this one step further, the real goal of an ad server should be to know in advance what versions are pre-designed for each target audience. Since publishers (through the collective learning of their ad server software) know which types of creative and campaigns/offers work, there is no reason why SnapAds should be allowed to get away with just the geewhiz of a Photoshop integration. Let’s make sure they’re pushing much further into some of the above areas.

David Shor

Seems also like Adsense or other text ad networks could offer a few different phrases and calls-to-action for a given campaign without breaking the bank. No photoshop integration required.

CG

i still dont understand whats so special about them the video wasnt useful.