Stop hacking LLMs (Why and How)

SEOs are doing what SEOs do. And Mordy has some feelings about it.

Two tactics have become the go-to playbook for gaming LLMs right now. The first is chunking: stripping your entire site down to bite-sized snippets — nothing longer than two sentences — so AI models can grab and cite them easily. The second is self-serving listicles: publishing “best [category] tools” posts where you rank yourself number one and fill your competitors in below you. Big companies are doing both. Google has explicitly said they know about both and are working to stop them.

And here’s the part people are underplaying: these hacks are working in LLMs — while quietly tanking organic search traffic at the same time. Glenn Gabe, one of the sharpest SEO analysts tracking this, has a name for it. He calls it Mount AI: sites that optimize hard for LLM visibility and watch their Google traffic fall off a cliff. So before you double down on the hack, the question worth asking is: which channel are you actually willing to lose?

Mordy and Miruna don’t say never hack. They say understand what hacking costs you — in positioning, in credibility, and in the organizational mindset it creates. Growth hacking strategies don’t build momentum. They borrow it from the future.

Nick Leroy, founder of Nick Leroy Consulting, SEOjobs.com, and the SEO for Lunch newsletter, joins to answer how Google will most likely get a handle on LLM quality — and why the timeline is shorter than most people assume.

Transcript

Mordy Oberstein (00:00)
You hear that? Of course not. That’s the whole point.

Welcome to Say It Anyway, where we say things about the digital marketing space that you’re supposed to kind of keep quietly to yourself. And we — wait for it — say it anyway.

Mordy Oberstein (00:21)
Each week, we’re taking an honest and maybe uncomfortable look at something marketers need to know but might be too afraid to say anyway.

I’m your host, Mordy Oberstein, the Head of SEO Brand at SE Ranking, and I’m joined by the outstandingly outstanding — I totally... I’m joined by the outstandingly outstanding CMO of Planable, Miruna Dragomir. How are you?

Miruna Dragomir (00:43)
I think I should start doing intros when you’re about to rant.

Mordy Oberstein (00:47)
That’s it. You should just say my name because I can’t get my name out.

For the audience, I flubbed through that last — because they’ll edit it out — I flubbed through the “I’m your host” part 20 times in the last two episodes. Yes.

Miruna Dragomir (00:59)
And he likes the part of being bombastic about my presence here. And I think when you rant, first of all, I might be better at saying your name and title than you are. But second...

Mordy Oberstein (01:12)
That’s so clear. I think a goldfish is better at saying my name and my title than I am.

Miruna Dragomir (01:18)
And maybe you could get that boost of confidence and compliments before you rant away at whatever issue.

Mordy Oberstein (01:26)
I am not good at compliments. I don’t know how to take them. I just find them awkward, like, okay.

So on social media, if you write something nice about me, I’ll say something sarcastic back. And someone said to me, “Mordy, you just have to say thank you.”

So now, every time I write “thank you,” I see that person who told me to do that. And they get super...

Or no, I’m just kind of trolling her. Yeah. Yeah, see?

Miruna Dragomir (01:45)
That’s the cutest thing ever.

Miruna Dragomir (01:51)
Well, that’s weird. Then I’ll take it as cute. But yes, I understand the problem.

Mordy Oberstein (01:57)
There’s a fellow in the SEO world. His name is Barry Schwartz. He’s like the godfather of all SEO, and he has the same kind of “not good with a compliment” thing.

So on his birthday, everyone would be like, “Happy birthday, Barry,” all over every social media platform. And he replies back, “Thank you.” Period. To every single one of them.

Miruna Dragomir (02:12)
That’s training?

Mordy Oberstein (02:15)
Not even saying thank you — period. End of conversation. Do not say anything else to me about my birthday.

Miruna Dragomir (02:21)
Thank you for acknowledging I have grown up another year.

Mordy Oberstein (02:25)
Period. All right.

So today, we’re talking about why you shouldn’t hack LLMs. And this is controversial.

But before we get into the controversy, let me just tell you that Say It Anyway is brought to you by SE Ranking and Planable, two sister companies helping brands grow their presence across channels.

SE Ranking helps you understand your search and LLM visibility, and Planable helps you create and collaborate on the content that improves that visibility across social, website, PR, and beyond. So you can track, distribute, and validate the impact of your data in an AI world.

Okay, now back to the hacking.

There are so many hot takes. SEOs are doing what SEOs do best. And I’m going to separate out some SEOs right now. There are some SEOs — God bless them — I’ll mention Lily Ray, Glenn Gabe, Aleyda Solis. Great SEOs who do not think this way.

Then there’s the other group of SEOs who will just hack everything possible. That’s what SEO is, right? If we can get into the algorithm somehow, manipulate it, and get our rankings up there, great.

I mean, who cares if Google will kill it off tomorrow? We live for the moment. We’re SEOs.

And they’ve done basically that for LLMs. They have figured out all sorts of ways to hack LLMs. Even though Google’s kind of come out and said, “This can actually ruin your organic traffic.”

Glenn Gabe is a famous SEO. Follow him on Twitter and LinkedIn. He calls it Mount AI.

So you see these companies or these websites start optimizing for LLMs, doing all these sorts of hacks I’ll get into in a second, and then you see their organic traffic fall off a cliff.

So they might be getting into the LLM — let’s just say they are — but they’re getting out of Google Search. So which one is worth it? At a minimum, which one is worth it?

Miruna Dragomir (04:03)
Most likely, they’ll get out of LLMs at some point because...

Mordy Oberstein (04:08)
At some point. That’s it. And we’ll get into that also.

Google’s got to get a handle on this. Their algorithm has made strides in quality. I know it’s not great sometimes, but it has become something otherworldly compared to what it used to be.

And at some point, Google’s going to take what they’ve done in their search algorithm and figure out how to do that in Gemini also. And then these things will stop working.

Google has said — there was a Verge article that talked about it — because one of my favorite tactics... There are two favorite tactics. One is called chunking, which I have a whole video series you can look for on the SE Ranking social channels called Let’s Chunk.

But chunking is basically turning everything into tiny little pieces of information so LLMs will like it. So nothing longer on your entire website than, like, two sentences.

And the other one is self-serving listicles. Big companies are doing it. Zendesk is doing this.

Basically, you write, “Who offers the best software for whatever, whatever, whatever?” And you list yourself as number one. Then you list other companies, maybe numbers three, four, and five. You’re one, two, three, and some other companies are four, five, and six.

And these self-serving listicles get picked up by the LLMs.

The Verge wrote a whole article about how SEOs are ruining the world doing this. And The Verge has a whole bone to pick with SEOs and them ruining the world.

I think people are ruining the world, and SEOs are just people. So what’s the difference?

But they got a quote from Google saying, “We are very aware of the LLM hack of self-serving listicles, and we’re going to figure it out at some point.”

So it’s mind-boggling to me to see this go on.

Miruna Dragomir (05:40)
I am hearing the counterargument to that, saying, right, it’s hacking, and we are in the classical SEO algorithm mouse-being-chased-by-cat scenario, where we constantly try to outspeed the cat, get what its path is, whatever.

I will not continue with this metaphor.

But long story short, we know we’re hacking. We know we’re in this ongoing chase, and we know that sooner or later, various things we’ve made will be caught and maybe penalized for it, or they won’t work anymore.

But it’s fine because Google may take another six months to get a crack on this. And during that time, we rank number one in LLMs and in Google, and we get the rewards for it. We get the customers. We get the money.

I understand, in a capitalist world, that that is the winning argument.

What I’d say is that it’s not “don’t hack, only do long-term stuff.” I think it’s about balance. There’s a spectrum.

And when you decide whether to do a hack or not, think if it has any reasonable logic to it, right?

If you do a hack that is undeniably a hack, it does not add value in any way, it does not do anything — then most likely, it will be cut down on hard.

If it’s midway, maybe that’s something that should still be on the table. I don’t know. What do you think?

Mordy Oberstein (07:08)
I think the biggest thing is mentality.

First off, I think the broader problem is that the hacks sometimes work. And if you see a hack...

I used to run a podcast called The SEO Rant a long time ago. It was a side thing I did for fun. And I realized that Google had this carousel of podcasts. I don’t know if it still shows up anymore or not. I guess it would still show up, or I don’t know. Maybe the AI is there.

When you search for “best SEO podcast,” a lot of the podcasts just have the word “best” in it.

So I’m like, okay, I’ll throw “best” in it. And I literally, on the homepage, wrote, “Is this the best SEO podcast? No. But I am going to say it is the best SEO podcast because this will help me get into the carousel.”

And it did.

The hacks are fine. I’m not against hacks. A less bad word for that thing is being opportunistic.

But I think the problem ends up being mentality, especially if you’re going higher-end or upmarket with whatever you offer. If your strategy is built, or your focus is built, on growth hacking, that spills over into everything.

I say this to my wife all the time: you can’t have your cake and eat it too. There’s a price for everything.

To quote Jim Morrison, “You get yours, I’ll get mine. No one here gets out alive.”

I’ll keep going on with euphemisms. There’s no free ride. I just did it again. I didn’t even mean to.

If you have this mentality of, “Let’s just go to the next hack,” the long-term strategy and the long-term building that you need to do to grow your digital footprint the right way doesn’t happen, because they’re two different mindsets.

And it’s very rare to find a company that’s a unicorn that can do both to the right amount.

So balance I agree with, but also realizing that there are opportunity costs. What happens on one side really impacts the other side.

If you’re in it for six months and you’re going to sell this damn thing, all right, what can I say? No.

But if you’re in it for the long haul, and you’re not selling, I don’t know, glasses cleaners, which is what I’m holding in my hand, and you’re selling microphones, which is what I’m looking at right now, it’s a little bit higher-end of a product. You need to think about more quality strategies than just hacking.

Miruna Dragomir (09:07)
A good, let’s say, similar situation — I’m taking this SEO hacking as very aligned with the growth hacking mentality, which had its prime age and is somehow still there, but it’s not as big as it used to be.

But I lived in it. When I joined Planable, and Planable was a tiny little baby that was barely able to feed itself, it was the era of growth hacking.

You had the “how to scale your startup fast” thing. The get-rich schemes for startups were, again, in their prime age.

You could scrape groups. You could do all sorts of shady things that would get you customers.

And again, it’s not looking down on those hacks. It’s that if you went — Reforge had this post at some point, Brian Balfour, I think. They had this theory about the canyon, the Grand Canyon.

It said you can grow a startup on hacks, but you will reach the point of a canyon where you will have no momentum to live by.

You’re supposed to build momentum. And if you only do hacks, those don’t build momentum. So you’re just going from one step to another.

Mordy Oberstein (10:23)
It’s endless. It’s endless.

First, when I was doing some consulting, I didn’t like working with startups for this reason. But when I did, I would tell them, “You need to grow. Whatever you do, you need to grow.”

Agencies also run into this problem. Agencies are just, whatever, “We do link building.” Fine. You do your thing. “You need us to do whatever? We’ll do whatever.” Fine.

And the problem is that they end up in a little bit of a corner because they never wanted to be that thing. And now they don’t have the space to pivot. They never left room in the cup for milk.

It’s not like you can’t do those things. You can.

First of all, the era of growth hacking — the internet was a completely different place. The internet was... it’s a whole different conversation.

The whole incentive cycle was different because what Google, in particular, was able to recognize and not recognize was more limited. And that era doesn’t exist anymore.

This whole hacking thing is really a remnant of an era that, in my mind, doesn’t exist.

When you’re in a startup and you want to do those hacks, I think it’s important to understand that they are what they are. They are temporary tactics so that we can get started.

But they do hit a ceiling. It’s very true. They hit a ceiling, and then you’re stuck, and you don’t know what to do.

And I’ll say another spin on the same point: no one ever wants to be the company shouting around the internet, “Here we are. Here we are. Here we are. Here we are,” versus the company that everyone just goes to.

That’s not the goal. The goal is you want to go where everybody knows your name. You don’t want to go shouting around the internet like a lunatic telling everyone your name.

Miruna Dragomir (11:53)
And that’s a very good point, because there is a point in life where it’s fine to yell, “Here we are. Here we are,” when you’re just starting out and there’s no one else vouching for you or pointing to you.

But you’re supposed to grow out of that era. And you cannot grow out of it if that’s the only thing that you do.

Mordy Oberstein (12:09)
Yep. I know it sounds like such a soft-skill thing to say it’s a mentality thing, but it is a mentality thing.

But going back to the question of Google eventually dealing with the quality issue in LLMs, I actually asked Barry Schwartz, who I mentioned before — Barry, the king of SEO — in my Let’s Chunk series: why is the quality in LLMs not on par with the quality in the Google algorithm?

And he had a whole point. It’s that LLMs want to give you a positive answer. They want to give you something. I saw a whole article about this that was really interesting, but that’s their whole outlook.

But at a certain point, Google is going to — out of all of them, the Anthropics, the ChatGPTs of the world — Google has many years of experience figuring out quality, and they’re going to figure it out.

So we asked Nick Leroy, the founder of Nick Leroy Consulting, the founder of SEOJobs.com, and the founder of the SEO for Lunch newsletter — Nick, if I left out any of your accolades and projects, please forgive me.

We asked him: How will Google most likely get a handle on the quality issues in LLMs, and how long do you think it is before they do?

And here’s what Nick had to say.

Nick Leroy (13:18)
Hey, Nick Leroy here, Nick Leroy Consulting.

And the question: how will Google most likely get a handle on quality issues in LLMs, and how long do you think it is before they do?

First and foremost, we need to understand that Google can control Gemini. It can’t always impact other LLMs like ChatGPT, Perplexity, or Claude.

So we want to make sure that we understand that Google is going to use Google sources to help optimize Gemini.

And so I don’t think it’s necessarily that they need to solve the LLM quality. They just need to control it.

What I mean by that is they’ll likely lean on what they already trust, using their index, user behavior, and authority signals to filter and ground AI outputs.

The messy phase is a bug — or I shouldn’t say a bug. It’s just an early adoption of AI technology.

I suspect that what we see today will be completely different in 12 to 24 months.

We just have to remember that Gemini and these LLMs are the light-speed version of where Googlebot had 20 years to be able to get the results to the quality that they are today.

Mordy Oberstein (14:31)
Thanks, Nick.

I really appreciate you clarifying my question, saying that it’s only talking about Gemini. Yes, we’re only talking about Gemini.

I feel like Nick, knowing our personal relationship, did that as a little dig to me, and I appreciate that. I like that level of trolling.

It’s a good point. I think what he’s saying is, first off, Google in the algorithm had 20 years to do this. But also on the LLM side, or the Gemini side in particular, they’re just trying to keep up with the Joneses.

They have so many competitors. They have ChatGPT, they have Anthropic, and they have their own stock price to worry about.

And they’re just trying to build, build, build, build, build and keep up. The quality will come later, but it will come once that market starts to settle down.

And it does kind of feel a little bit like it is settling down. Thank you, Sam Altman, for being so ridiculous to sort of settle that market down by default.

Miruna Dragomir (15:15)
I loved what Nick said, and it got me thinking though that there again, we’re comparing AI algorithms to Google’s old algorithms in the 2000s era.

And we’re comparing a race where humans could temporarily win against the algorithms. But the span of time for fixing whatever loops are broken — loopholes — is shrinking, because now you’re against robots.

You’re not against human-built algorithms, human-coded algorithms. You’re fighting against robots that can fix themselves in milliseconds.

So that’s something to really keep in mind.

Mordy Oberstein (15:57)
Really interesting, because I think you have to realize they have to figure out the quality problem. It’s not an if. It’s a when.

Because if they don’t, people will not use it. Which is why Google was so smart about making sure that its search algorithm was of a certain quality, because people are just not going to use it if the results they get are not good.

So they will have to fix this. Smart companies like Google — which, I know I say Google is smart. It doesn’t mean I always agree with them, but they are smart — will figure out the quality issue at a certain point.

I guess what we’re clearly saying, as we move into the “clearly saying” part of the show, is don’t hack LLMs.

Yeah, just don’t hack the LLMs.

Go out on social and say it anyway. Say, “Don’t hack LLMs.” And then tag all of the SEOs who are really into hacking LLMs and see what happens to your social media account when you do it.

Don’t actually do that. That’s a bad idea.

Miruna Dragomir (16:31)
Yeah, skip that part.

Miruna Dragomir (16:45)
You get popular.

Mordy Oberstein (16:49)
That is absurd. That is a hack.

Picking fights. Social hack. You heard it here first. Do it. Maybe not.

Maybe not. SEO is a weird space for that kind of stuff.

Anyway, I think we’ve said quite enough this week.

Join us again next week as we ask — or as we take up — marketing’s next big struggle. That sounds mysterious.

Look for a new episode of Say It Anyway on the Planable and SE Ranking sites and wherever you choose to listen to podcasts.

Please don’t be shy about leaving us a review and a rating on iTunes and Spotify.

Look for more from the podcast on the Planable and SE Ranking social media channels.

And until next time, don’t keep quiet.