Insights

Designed for People. Written for Robots.
Last year, we embarked on a brand refresh for mdg that included completely rebuilding our website. It felt complex but also pretty straightforward: Design it well, write it clearly, make sure it ranks. We knew how to do that.
Then, over the course of the build, LLM usage exploded. Claude, ChatGPT, Google Gemini — people started depending on these platforms to find information, make decisions and evaluate vendors. And with that shift came an added layer of complexity we hadn’t fully anticipated.
We needed to think about our website in a new way — as a content source. One that LLMs would read, interpret and potentially quote back to someone asking about event marketing strategy or association growth.
The problem with writing for robots
There is a lot of guidance out there on how to make your website LLM optimized. There is less on how to make it human-friendly and robot-friendly at the same time. In theory, those goals align. LLMs like what humans like — clear structure, direct answers, content that gets to the point. In practice, if you aren’t careful, chasing LLM optimization turns your website into a laundry list of keyword-stuffed instructions that sounds like every other site out there. It can quickly turn generic and soulless, making it easy to ignore.
This was something we wrestled with throughout the rebuild of mdg.agency. Here are the principles we kept coming back to when we had to choose between what the algorithm wants and what a human needs.
1. You can sound human and rank high.
The most useful data most associations aren’t looking atKeyword-rich content can quickly start to sound like it was written for a search engine, not a person. And the moment it does, you’ve lost the reader.
To make sure it was engaging, every page had to pass a simple test before it went live. Does this sound like something a person at mdg would actually say? Even if it meant sacrificing some of the best practices for LLMO, we knew we needed it to sound like us. Direct, specific, grounded in real experience and a little bit clever.
Here’s the thing: LLMs are increasingly good at recognizing generic content, and as AI-generated writing floods the web, distinctiveness is becoming a competitive advantage. A site that sounds like every other agency is less useful to an LLM.
Keyword-rich and human-sounding are not in conflict, but it takes discipline to be both. The moment you start writing for the algorithm first, the reader — human or robot — can feel it.
2. Think of the question.
LLMs are essentially built to answer questions. They actively look for content that mirrors that format: a clear question and a direct answer that’s specific enough to be citable. A well-written FAQ gives them exactly that.
A thorough question-and-answer list that’s easy to find helps visitors get quick answers, surfaces questions they didn’t know they had and positions you as a trusted resource.
The trick is using questions people actually ask, not just questions that make you look good. A genuine question — one that addresses real hesitations, real comparisons, real “but what about” moments — is useful to a human and exceptionally useful to an LLM.
3. Some content gets you found. Some content gets you trusted.
Optimizing purely for search produces content that answers questions but never leads anywhere. It’s findable but forgettable and it does nothing to differentiate you from the next organization with a blog.
We think about everything we publish through two lenses — science and art — and we’re deliberate about which one each piece is.
Science content captures demand that already exists. It’s built around what your audience is actively searching for, structured to perform in search engines and surface in AI-generated answers. The goal is to be the most useful answer to a question someone is already asking.
Art content shapes how the industry thinks. These are opinion-driven, human-led pieces that reframe a problem rather than just answering it (like what you’re reading right now). They build reputation and showcase expertise at a level that goes beyond tactics. LLMs value this content too, because it shapes the broader understanding an LLM builds about who you are and what you stand for.
Science content helps people find you. Art content gives people — and the platforms surfacing you to them — a reason to trust what they find.
4. Don’t forget the power of emotion.
If you spend all your energy optimizing content for robots, you can end up with a website that is clear and informative but doesn’t resonate. And a website doesn’t resonate doesn’t convert.
While the content work — structure, clarity, FAQs, specificity — is in service of both LLMs and people, the experience, including micro animations, transitions, moments of delight (like party mode 🎉) that make a site feel considered rather than assembled, is purely for the human on the other end.
No LLM will ever notice a well-timed animation. But a person will. And that person is making a judgment about you in the first few seconds of being on your site. About what you do and how you do it. About what the experience of working with you might feel like. And whether you care about the details.
I mean, our industry is all about human-to-human experiences. So, our websites should be no different.
This is new ground for all of us
Like any new frontier, navigating the balance between human and LLM optimization will require learning as we go. Right now, it can feel cumbersome. One more layer of complexity on top of an already demanding content workload. Another set of rules to keep in mind. Another reason to second-guess a sentence that felt perfectly good before.
But this is how every shift in the marketing landscape felt at the start. SEO felt that way. Social media felt that way. The organizations that leaned in early, figured out the principles and built habits around them pulled ahead.
A year from now, writing with LLMs in mind will be second nature. The structure, the FAQs, the discipline of being specific and direct with clear H3 headers … it will just be how you write for a website. And the good news is that most of what LLMs want is what good writing has always required: clarity, specificity and a genuine point of view.
In the end, the readers are who we want to reach. The robots are just the middlemen (or is it middlebots?).

Sara Baliti is the group director, agency marketing and an account strategist at mdg.
Questions & Answers
Whether it’s AI for associations or AI for nonprofits, there has been a shift in the way organizations and events are found and evaluated. People increasingly ask LLMs like ChatGPT or Claude for recommendations before ever visiting a website, so content needs to be structured to compel human readers and attract AI tools — with clear headers, direct answers and genuine specificity, not just keywords.
An LLM-optimized event website template pairs a clean, scannable structure using H3 headers, FAQs and direct answers to real questions with a distinct point of view. This format performs in search and surfaces in AI-generated answers, while still feeling human — not generic or keyword-stuffed.
Good membership website design needs an FAQ section because LLMs are built to answer questions, and a clear Q&A format gives them the content they need. FAQs also help members find quick answers and introduce questions they didn’t know to ask, building trust with both people and AI platforms.
Yes — this is the balance mdg calls “designed for people, written for robots.” Keyword-rich and human-sounding content aren’t in conflict; using AI for events or membership takes a thoughtful approach: Every page should sound like something a real person would say, even while following LLM-friendly structure.
Strategy. Ideas. Proof.
Delivered Monthly.
Get insights from real event and association marketing work — not just theory.
"*" indicates required fields
Related Insights
