Thank you for your kind words on the unpronounceable Aztec god! Viewability always comes with a trade-off I feel. To maximize viewability, you have several choices:
1) You place more ad units above the fold
There's a limit to how much you can do that, as it not only creates bad user experience but may also earn you an SEO penalty by Google. For us as a non-profit, user experience is super important, so we have opted to only place a single 300x600 ad unit (which also receives bids for 300x250 and 160x600) above the fold.
2) You load ad units dynamically as the user scrolls
That's quite a bit of development work where, especially when combined with header bidding, so we opted not to go down this route. Also, you're losing out on bids that you might receive for less-viewable inventory.
When we switched to header bidding, we were surprised to see how even our 728x90 ad unit at the very bottom of the page was receiving good CPMs. In the waterfall, we got maybe $0.20 CPM out of this unit, and now we're getting something like $0.80 CPM. Probably only about 10% of our users view that ad unit: In order for dynamic loading to be worth it, we'd need to average out at $8.00 CPM for viewable-only impressions... unlikely!
3) You use Onscroll
Luckily, Sovrn has a technology that allows us to combine the two: OnScroll. We have got our standard header bidding ad units on the page, always loaded. We get the CPMs for those. Whenever a user scrolls down to such an ad unit, and views it for 30 seconds or so, OnScroll comes in and reloads the ad unit with an OnScroll ad (if it has demand for that impression). The CPMs there are good and the fill rates are decent.
We're also going to try adding a fifth ad unit which is purely OnScroll (and sticky to the browser window) on the side of our super long articles (only inserted for articles of 2,500 words or more). Let's see how that will perform.
Lastly, I also feel that viewability is often overrated. Back in our waterfall days some of our networks kept pushing us for higher viewability. So we ran several experiments, moving ad units around and changing which ad network was above the fold or below the fold. We saw only very minimal average CPM differences. That was over a year ago, so maybe this has changed now, I'm not sure.
@RobBeeler, as you're from AdMonsters, I'd love to hear your opinion on this, too.
Disclaimer: Sovrn didn't ask me to say any of this, it's my honest opinion, even though it sounds like an ad for Sovrn.