I’ve only been on the bad end of a conspiracy one time, but that’s one too many for a lifetime. What I learned from this experience is that there’s a brief moment of relief that comes from knowing you’re not crazy, followed by the paralyzing realization that all your darkest fears are coming true. That’s a little bit what it’s like to be a publisher of a website in 2025.
LinkedIn is a good industry barometer, and right now, the automotive news industry is split between people hoping to pivot to some sort of membership model and people losing their jobs. I’ve long tried to offer at least freelance work to individuals who lost gigs in one of the now innumerable prior rounds of layoffs, but we’re getting squeezed by the same factors squeezing everyone else, so it’s a lot harder to keep up with the pace of people who need help.
No mega platform owes us anything. I want to acknowledge this. Google (or Alphabet, if you prefer) doesn’t have to send us any traffic or any advertisers, and it still does a lot of both. My understanding was always that we work hard to provide the content that makes Google’s many products valuable, and the company, in return, provides some reasonably proportional level of traffic. This was also the deal with Facebook before Facebook decided to pull the rug out on publishers. Implicit in this understanding is that, is that under the Digital Millennium Copyright Act and similar laws, companies like Google get a “safe harbor” from certain lawsuits because Google isn’t the one making any content. We are. So we assume that risk.
With the push to make everything about artificial intelligence, Google has tried to integrate its own AI systems into its suite of online products. You’ll notice the Gemini diamond on your Google Sheets, an AI assistant in your Google Meets, and generative AI maps in your Google Street… views.
I mentioned this in my previous article on the topic, but as these AI products started to roll out, it seems like Google understood this could be detrimental for publishers who assumed they’d still be at the top of searches, either via Google News or via regular search. This change didn’t have as big an impact on us as we were new and didn’t have a long tail of search results.
What Google did to make up for this is a product you may not even be aware of, but everyone in the industry knows as Google Discover. Basically, any time you go to any Google product, it’ll serve you up some stories, videos, or social media posts. Here’s what my search page looks like:
This has been a key way that people have found us, because it’s a more competitive place than search and gets more traffic than Google News. If more people clicked on our stories, Google seemed to show more people what The Autopian was writing. We write great stories and headlines, with excellent thumbnails, so we absolutely killed on this platform.
And then, as I mentioned, it started to go away. I attributed a lot of this to Discover ignoring its own guidelines for content and instead promoting a lot of AI-generated slop that somehow managed to sneak past the company’s own crossing guards (so much for Google’s excellent machine learing). The bulk of it may just be Google itself, as the latest report from web analytics firm Marfeel (full disclosure: this is the one we use) shows:
In key markets like the US, Brazil, and Mexico, AI Summaries already make up 51% of the feed. When they appear, AI Summaries often display multiple publisher logos but provide a single actionable click, which in most cases plays an inline YouTube (77% in the US, 100% in Brazil and Mexico). Deeper in the feed, AI dominates even more, quietly replacing publisher links.
It looks something like the graphic below, and you can understand why suddenly publishers aren’t getting the traffic that they expected to get in lieu of the search traffic slurped up by AI. We often get crammed in with other logos at the top in a way that results in little or no traffic to anyone.
Who owns YouTube? Oh, right. Alphabet.
None of this is a huge surprise, as TechCrunch reported on this summer, but seeing it in our reports confirms the fears of diminished traffic were correct. And it’s not just us. It’s either comforting or terrifying that our main competitors are also being smacked around by the same changes.
If you’re wondering why so many of your favorite sites are suddenly even more ad-heavy than usual in the fourth quarter, this is likely the cause. With less traffic, the only way to get more money is to squeeze more ad impressions on the page, but this causes more supply and then forces prices down, creating a need for even more ads… ad infinitum.
The difference between our site and those other sites is that we’re way more careful about how we roll out ads because we don’t look at display as our major source of revenue (my guess is that if you were out of the industry for even a year, you’d be shocked at how crappy RPMs are). We have a membership business, and that means we care about getting people to check out the page and then want to stick around.
I think about it this way: If your friend told you about this great restaurant with a chill vibe and a perfect juicy hamburger, you’d probably go. If you got there and the restaurant was full of TV screens blaring ads, the menus were dirty, and the waiter kept trying to shoo you away from your table, you wouldn’t care how good the burger was. You’d never come back. That’s what most websites are like now.
When your friend (in this metaphor, Google) sends you our way, we want you to love it here. We want it to be a nice experience. We want the burger to be delicious. And we want you to come back. Slamming the page with ads is counter to this, and other publishers in our space that have experimented with a lot of ads and membership have found that people didn’t really want to become members.
Our goal is to eventually turn as many of you as possible into members by making membership super valuable, but we also want more people to find this place, which means we’re not entirely paywalled. It’s a tough balance, but our plan has always been to lower our dependency on revenue from all platforms, not just Google, while also keeping our discoverability high.
I write this because it’s going to be difficult for publishers who have insisted on the other model to function going forward, and you’ll probably see other automotive websites you love either get smaller or do all sorts of seemingly nonsensical things (AI slop, trending TikTok articles) just to make up the difference. I also expect a lot more layoffs in Q1 of 2026.
Here’s another chart from Marfeel:
Do you really think the AI summaries are going to stay at the bottom of the page? I don’t.
Marfeel’s report pretty much nailed it when it said “Google Discover is shifting from a traffic distributor to an attention controller.”
This is also bad for Google! Going back to the original conceit of the Google-Publisher relationship, AI needs human thinking to feed its models. We provide that thinking. The more watered down and terrible and sloppy the inputs, seemingly, the harder it is to get a good output and the less valuable that input becomes.
Again, we’re building our business to be resilient against these sorts of changes over the long term. That means merch, toys, partnerships, some display advertising, and, most importantly, membership. If you’ve been thinking about becoming a member but haven’t yet, and are able to, now is a good time to do so.
Our plan is to review what sort of membership we can project out into 2026 based on 2026, and we’ll use that data to decide how much we can spend on freelance, the website, more events, et cetera.
And if you’re already a member, as always, thank you for your support. It makes moments like this a lot more tolerable and keeps us focused on making this the best place to hang out on the web for car people.
Keep fighting the good fight! I appreciate your high quality bar – not just the authored content, but you have the only non-toxic community I’ve encountered on the Internet.
AI will kill the internet. It is fast forwarding dead internet theory at a remarkable speed. I only worry about what will happen in the background after. Some countries are already letting it have government jobs. That terrifies me.
Just for anyone doubting or didn’t hear: https://time.com/7324934/albania-ai-minister-diella/
The number of people I say “Oh I wrote an article about the Rodius on the Autopian” and then have to explain what The Autopian is (people who are in the car community) is sad 🙁
They seem to like it once they look it up though.
The enshitification of the world, enhanced through, and by, AI, progresses…
Unfortunately it also seems like AI is eating its own tail.
AI generates content that gets placed on a page and provides misleading or incorrect information.
AI searches the web for content to integrate, stumbles upon its own slop and regurgitates an even more misleading version of that very content.
It can be tough to spot AI content and this process will make it so that is all the internet will be in a couple of years.
It’s bad enough that it was already crowded with idiots and miscreants, it’s only gonna get worse.
It’s interesting to read about how these things work behind the scenes. It sounds like there is almost no chance of success anymore because popularity and a good product don’t equate to money.
Other than being a member and buying a little merch, do you have any thoughts on how we can help this place become more successful? For that matter, do any Autopians have any ideas?
“I’ve only been on the bad end of a conspiracy one time, but that’s one too many”
I heard. It’s a shame what they did to your pizzeria. It didn’t even have a basement.
Lol!
Perhaps we will go back to word of mouth, person to person trust. I took Amtrak into Chicago and back a day this week for a meeting. The train was packed with college students. The kids that sat next to me were talkative – one was headed to student teach HS english next semester, one a junior in econ and polysci to become an economist. They both struck up conversations with grouchy old looking me – both commenting on my paper copy of The Economist.
Both conversations centered around regaining trust in society, finding ground truth in the AI slop era, finding real connection when most stories are so charged with search hit words. We shared the hope that those who write will survive this upheaval. And joked that live performance may surge in the quest for authenticity.
So – I pay for subscriptions to news that I use and want to be around. And then I tell my friends. And for my fun reading? Used books from the second hand store, predating the AI era. At least until this sorts out.
I know this isn’t technically the place to bitch about Google’s AI(or maybe it is), but I’m old and have been a consumer of the internet since Mosaic, and NOTHING has pissed me off as much as Google’s AI summaries, they’re far worse than “slop”.Why? Because they’re FUCKING WRONG MOST OF THE TIME.
Look, I’ve been a mechanic most of my adult life, now I’m a fleet manager/shop foreman, I know my stuff when it comes to automotive repair, as much as one human being can in today’s complex world, and damn near every time, when someone around me asks a question, the first thing at the top of the phone screen is Google’s AI summary and it’s incorrect. Not in a pedantic way, or have a small part incorrect of a complex question, it’s just completely wrong.
And it’s wrong in an infuriatingly smug manner, no equivocating, no weasel words, it just spits out a wrong answer and unless you’re an expert or at least a learned novice, you’re not gonna know. Then the poor bastard that asked the question either goes about their repair job and screws it up, or asks someone else, or a specialty forum, and that person now has to A) answer the question correctly and B) convince the person that the first answer they received is bullshit.
It’s tiring and infuriating and I’m afraid the AI models are going to scrape already-incorrect content and get increasingly incorrect as the wrong info propagates across the Web. God help us all.
I can’t even trust Google anymore for a cursory check to see oil type and quantity when I’m helping a friend change the oil in their car.
Wait, this place is a burger joint? I thought it was an ice cream shop!
Anyway the internet is becoming exactly what you’d expect when the gatekeepers are psychopath billionaires who don’t recognize the dystopian parts of the dystopian scifi they worship.
Sorry, sorry… it is an ice cream shop. That also serves burgers.
“I attributed a lot of this to Discover ignoring its own guidelines for content and instead promoting a lot of AI-generated slop that somehow managed to sneak past the company’s own crossing guards (so much for Google’s excellent machine learing).”
I’m wondering if slop is really sneaking past crossing guards. Is Google actually promoting it?
I wouldn’t be surprised if Google execs are either indifferent to AI slop or even actively worried that AI is struggling. And they may well be leaning on algorithm developers to let more in.
I doubt it’s an explicit policy, if for no other reason than they want to maintain deniability. But a steady stream of memos and meetings questioning all kinds of decisions about what information gets promoted can easily make it clear to developers to stop favoring mainstream sources or setting up rigorous tests for information reliability.
Thanks for writing these posts! For many here, the backside of internet publishing and how google and similar work to filter traffic, is largely unknown – particularly as it rapidly changes.
Burn it all down. Internet 1.0 for life!
I make a point to send as many interesting articles to friends and family as I can. IMO, good websites should be household names — not something you stumble across via search…while that’s a valid source, of course, my layman’s guess is that ~50% of traffic coming directly to the site is an ideal goal. Then you know people are being pulled to you, and nobody can take that away.
The way searches go these days, I have to put my BS filter on “high” just to even read results from any major search engine. And I rarely click on anything where I don’t recognize the URL because odds are good it’s either all AI or enhanced to the point of blathering circular references and puffy, uncanny valley language.
For something so important to the billions of dollars they make, no one seems to want to destroy the internet as much as Google.
Who cares about tomorrow when there’s profits to be made today. Google search and the internet at large being mostly useless sounds like a tomorrow problem
I actually think it has less to do with any current profits and more to do with betting on a sketchy vision of future profits. And I think there’s a strong chance they’re pretty far down a sunk cost fallacy path they took several years ago.
Don’t be evil they said… LOL