Lessons from Testing Hundreds of Ad Formats — Talk with Roy Peleg
This week on the podcast we talk with Roy Peleg, who is one of the founders of a company called FirstImpression. Roy and his team have built technology that allows a publisher to dynamically create and monetize highly customized new ad placements that can be tailored for any editorial parameter imaginable. This lets publishers determine the best advertising strategy for every circumstance on their website and truly maximize their monetization potential.
I was initially interested in talking with Roy because his company is sitting on a huge data set about how users respond to different types and variations of ad formats. That makes him an expert on identifying what formats work the best in which scenarios.
It turns out, as with many parts of ad tech, general rules of thumb are rare and the performance of almost every ad format is dependent on the unique circumstances of the content surrounding it. There is no one cookie cutter format that works the best everywhere, but with enough diligence and testing you can determine which customized ad formats will work best for you.
FirstImpression, similar to Sortable, is a company that brings large scale infrastructure and machine learning capabilities to a field that has traditionally been run by humans. I think this new style of ad tech is extremely interesting and that companies in this space will make major waves in the next few years as the field of publisher monetization develops.
1:30 — Intros/Roy’s background
2:30 — The problem trying to monetize different strategies as a publisher
3:00 — The origin of First Impression and its idea
4:15 — Lack of innovation in ad formats
4:45 — History of ad formats
6:00 — Are there general rules of thumbs when it comes to ad formats?
6:50 — How granular do you have to get before you see trends?
7:50 — BTF vs ATF and viewability
8:36 — Do certain types of units work best in certain scenarios?
9:45 — Have you guys noticed negative trends in advertising?
11:20 — Should ad ops continue to be done by humans?
12:30 — Do people ever out perform the automation
13:10 — The difference with A/B testing as a person vs a machine
14:10 — Are there permanent ad format changes that last or are users just reacting to a the most recent change?
14:48 — Ad Fatigue & Banner Blindness
15:56 — It takes time for testing new formats to work
16:15 — Current Trends — High impact units are in demand
17:15 — Can you rotate and test formats for advertisers?
18:09 — Are trading desks interested in this data?
18:50 — Are CPM ad used differently than CPC ad by publishers?
19:30 — Are you rotating demand sources as well?
20:19 — How much longer can the old style ad networks hang around
21:10 — How do you balance ad revenue vs user friendly desings
22:20 — What is the advice that you would have to publishers who can’t user your tools?
24:50 — Why the black box model of optimization services doesn’t work
24:50 — What do publishers focus on once you handle their ad real estate?