There's been a few people who commented this in the past, but as an advertiser on Reddit, I want to share the numbers I see.
First, there's a few things to understand in the world of advertising:
- Cost Per Impression - Usually shown as a cost per 1000 impressions, this is how much it costs to run a regular ad
- Cost Per Click - This is a different type of ad where you only pay for who clicks. It's also the reason sometimes you see really bad ads - They're only paying per click, so they want the most gullible customers
- Analytics - I can watch who comes to my website and what they do. I can actually watch a lot more info than that, but it's all I need to run my businesses
- Organic User - Someone who came to my website without an ad
PSA: If you're not using uBlock Origin to block ads, please install it. Firefox - Chrome. Every other mainstream adblocker sells your data in some capacity, but uBlock Origin is open source.
Now, with those things in mind, I pay for Cost Per Click, and I target a more expensive user group. In the ad I'm about to show you (picked at random, but it's within +-20% of most my ads), it costs me an average of $0.82 every time someone clicks my ad:
(Yes, it's brutally expensive. If you really hate ads, install AdNauseam. You will cost advertising companies thousands of dollars.)
But okay that's fine, because roughly 2,000 people went to my site, right? Lets see what they did when they went there
See - There's something interesting about this, and it's less apparent in other advertising networks. You see while Reddit charged me 1,600$ for 2,000 users, my own analytics show only 1,142 people came to my site in the same time window - and that number also includes my organic users, by the way.
So what happened to almost 50% of the users I paid for? Some people accuse Reddit of inflating the numbers, but that's illegal, and there's a much simpler explanation. Reddit's PMs and are deliberately designing ad placement to maximize clicks (and get more money). What they don't realize, is they've made everyone miss-click on ads, so both users and advertisers miss out.
In fact, that miss-clicking part is trivial to prove. Guess when I ran advertising campaigns on Reddit?
Anyways, that's all for now. Reddit doesn't only screw over their users, but their advertisers as well.
This reinforces my belief that online advertising produces a lot of objective data ("how many times was my ad viewed? clicked?") but benefits from not being able to tie that to outcomes companies are actually interested in ("are the ads expanding business?").
A number of years ago I read an analysis on how some large social media site had changed the order of a few important buttons out of the blue. This was likely from A/B testing showing increased engagement, but it was probably just confused users clicking on it. I bet similar things happen all the time in ads, possibly inadvertently. If an A/B change shows increased ad clicks, it's unlikely not to be adopted, even if it's not intentional clicks.
A lot of fast paced companies, big on "ownership" and data give promotions based on how well your feature performed. You need a measurable metric, so they usually go for something like clicks.
They absolutely know it's making the product worse. But for a 100k/yr bonus, they don't care.