I’ve worked in media and digital advertising and, on multiple occasions, actually implemented ads on sites. The proper way to do it is to A/B test different ad SDK configurations (refresh rate, delay to initialization, etc) and measure the impact on various user engagement metrics (time on page, session duration, etc).

These metrics can sort of tell you how people are reacting to the ads, though it isn’t perfect obviously. At the very least, it will let you know how violently people are reacting to the ad experience, which is also a good indicator for how it will impact SEO.

Or course there’s an inverse relationship between ad revenue and user experience: with a very light ad experience you basically make nothing, with a very heavy experience you will make a lot in the short term but in the long term you will lose users, traffic, and ultimately money. If you do things right, you can strike a balance at least.

I’ve also managed websites that hired some outside firm who worked on a revshare basis to come in and load ads on the site, and they didn’t do this kind of testing, and though initial revenue was high their traffic ultimately tanked in a matter of months.

A big problem I’ve seen is you find a nice balance between revenue and UX and then that becomes the new baseline/control that future people start testing against. So slowly over time it’s a death by a thousand cuts.

Turns out a lot of business people are bad at business. It’s hard to even explain how these metrics work to less technical people and it takes a special company to just trust the engineers and turn down the short-term revenue.