Question for the ad-ops side: those native content-recommendation widgets, the you-might-also-like grids of thumbnails at the bottom of a video or between rows, do they actually out-earn regular display for you, or are they just a lower-quality way to fill space? the pitch is they blend into a tube layout better than banners, get more clicks because they look like more content, and do not trip ad blockers as hard. the catch is the click quality can be junk and they can cannibalize your own internal navigation and pageviews. so has anyone run native recommendation widgets against display on the same inventory and actually measured which nets more per session, and did the native clicks hold up or turn out to be worthless?
the hidden cost of native widgets that almost nobody measures is that they cannibalise your own navigation. the widget earns you a visible click while quietly stealing the internal pageview the user would’ve made on your own related videos, so you book the ad revenue but lose the session depth and the downstream impressions. i only caught it when pages-per-session dropped right after i added a recommendation widget that looked like a clear win on its own click number. measure the whole session, not the widget, because a unit that out-earns display on its own line while shortening the visit can be a net loss once you count what it ate.
the click quality is what bit me with native. the widgets pulled a great click-through rate because they look like more content and people tap them on reflex, but that traffic converted terribly for the advertiser, which came back around as my rates getting cut the next cycle because the source performs badly. a high click rate on junk clicks isn’t a win, it’s a short-term number that poisons your standing with the buyers. judge native on the conversion and the rate it holds over time, not the flattering click-through. manufacturing clicks that look valuable is basically its whole trick, and the market corrects your rates once it sees through it.
the clean way to settle it is a holdout. run the widget on half your traffic and not the other half, then compare revenue and pages per session across the two. judged on its own stats the widget always looks great. the real question was never whether the widget earns, it’s whether the page earns more with it than without, and only the holdout actually answers that.
the lighter blocker-tripping is a real advantage of native units, if it’s actually true for your implementation. served the right way a recommendation widget reads as site content rather than an ad call, so it survives on traffic where your display just gets stripped, which is real incremental money from users you were earning zero on. the caveat is that the second you start tunnelling obvious ads through it to dodge blockers, you’re back in the arms race and putting your domain at risk. it’s worth having when the unit is genuinely content-shaped and worth nothing plus some risk when it’s just a disguised banner.