 |
| Samuel Barber |
Last year I was on a webinar on the tropic optimizing for AI search. We were all learning at that point, I was by no means an expert. The general thesis felt clear enough: structure your press releases for AI parsing, get your brand mentioned on high domain authority sites, make your content easy for LLMs to cite.
Six months on we’re all a little bit smarter. Yet, what’s actually happening and the advice I see being handed out, I've started asking myself: what the hell are we all doing?
The gold rush is on
AI search optimization has exploded as a topic over the past year. Startups like Profound have raised serious funding. Industry voices like Sara Evans have built significant followings sharing expertise on LLM visibility (her work is genuinely great). Established PR and marketing agencies are rushing to bolt GEO services onto their offerings. And new consultancies are launching specifically to serve this need.
The message is that the era of AI search is here and if you're not optimizing for it, you're falling behind. I think that’s probably true, except, I’m increasingly concerned about the paths some are taking.
I give you one example, a financial technology company published over 20 blogs in December alone, all exceptionally long (3-5,000 words), all visible on the company’s blog page. Each one a magnum opus on a very specific or mundane topic such as a comparison between two different companies. Then there are these incredibly well structured and written press releases bearing absolutely no news whatsoever being pushed out onto newswires.
I suspect, those “gaming” the system are playing a risky game and could get found before too long.
Moneyballing AI-search optimization is not a long-term strategy
The way some are creating these AI-optimized content farms feels quite Moneyball-esque. The difference is the basic principles of baseball don't change between seasons.
With AI search, companies are optimizing for a game where the rules will get rewritten constantly; model updates, new retrieval methods, shifting quality signals. The edge some gain because they’ve cracked the code will be lost when OpenAI or Anthropic or Google decide the game now works differently.
History repeats
What I'm seeing in AEO and GEO right now is uncomfortable but hard to ignore. The tactics that seemed reasonable last year have metastasized into something that looks a lot like the content spam of early SEO.
Content farms flourished until Panda killed them. Link schemes worked until Penguin. Keyword stuffing was standard practice until it became a penalty.
LLMs are different. They change continuously. Pattern recognition is core to what these models do – they’re inherently better at detecting optimization than rule-based algorithms ever were. OpenAI, Anthropic and Google are all competing on answer quality. They have strong incentives to filter out content that exists to game their systems.
Companies investing heavily in AI content farms may be building on sand. The optimization window will be shorter, less predictable and harder to sustain than SEO ever was.
Substance over industrialization
There's a difference between gaming a system through industrialized AI content and just being easy to understand. The answer is to marry the principles of effective AI search with the fundamentals of good content marketing.
Start with structure. Clear headings, direct answers near the top, consistent entity naming across your web presence. This isn't manipulation – it's making content accessible to any reader, human or machine.
Then focus on being citation-worthy, not just crawlable. LLMs treat earned media and analyst mentions as authoritative. Be the primary source where you can – original research, proprietary data, named expert perspectives. If you're the actual source of a useful data point, you're more likely to be cited.
And don't abandon the humans. Content that actual customers, journalists and analysts find valuable will perform better as LLMs get smarter at detecting quality. The 3,000-word blog that nobody reads is a bad long-term bet.
Invest in authenticity and win long-term
Some of the expertise being shared right now is genuinely useful. But the AI-optimized auto generated content tools being sold are enjoying a brief window before the models adapt.
The companies, agencies and consultants that win long-term will be the ones who understand the rules of the game and find that authentic middle ground. Araminta Robertson from Mint has a great blog on this.
The penalty for those who take shortcuts may come faster and harder than ever before. I'd rather optimize for a game where the fundamentals don't change than moneyball my way through one where the rules get rewritten every quarter.
***
Samuel Barber is the founder of Pitchr, a financial services and fintech PR firm.
Newsrooms and TV producers are increasingly incorporating AI into their work. Unfortunately, most brands haven’t caught up.
In communications, where trust is earned and lost in moments, human judgment, experience and responsibility will define the future far more than artificial intelligence ever can.
Younger employees are more likely to receive AI training in the workplace as well as guidance for using AI tools, according to a recent report.
For decades, visibility has been a game the largest brands have won by default. Today, as AI search increasingly serves as the Internet’s front door, those rules of discovery are being rewritten, presenting a rare and significant opportunity for communications professionals.
