I ran a 30-day LLM visibility experiment on my own brand. Here's what the data actually showed.

This is Post 3 of 3 in a series documenting a live LLM visibility audit on my own brand, NextWise Studio. Post 1 covered the baseline. Post 2 covered what I changed and why. This post covers the 30-day results.

Thirty days ago I set out to answer a simple question: if someone searches for what I do in an AI tool like ChatGPT or Perplexity, do they find me?

At baseline, the answer was no. Not when it came to category searches, the kind where someone doesn't already know my name.

So I made a set of deliberate changes, documented everything, and committed to sharing the results honestly. Here's what the data showed.

How I measured it

I used six different measurement sources to get a complete picture.

I ran the same set of prompts manually in ChatGPT, Perplexity, Claude, and Google AI Overviews. This is the most direct test of what a real person would see. I tracked my Google Search Console data to see how my site was performing in traditional search. I used a third-party AI search visibility tool to benchmark my brand mentions against competitors. I pulled my website analytics to see who was visiting and where they came from. And I tracked which AI crawlers were visiting my site and how often.

Six sources. Thirty days. Here's what each one showed.

Where LLM visibility improved

AI tools now know who I am. When someone types my name or my company name into ChatGPT, Perplexity, or Google AI Overviews, they get accurate, detailed results. A month ago that wasn't consistently true. Now it is. Brand clarity is the foundation everything else builds on.

AI crawlers are actively visiting my site. The bots that power ChatGPT, Perplexity, and Google's AI tools have been crawling my website consistently throughout the past 30 days. My content is being read and considered. The plumbing is working.

More people are finding me through ChatGPT. The number of times someone asked ChatGPT a question that led them to my site went up 27% this period. My homepage saw a 50% increase in AI-driven visits. Small numbers, but the trend is moving in the right direction.

My site is showing up more in Google. Overall impressions in Google Search grew significantly compared to my baseline period. My blog posts and key pages are indexed and appearing in results. Google is starting to associate my site with AI marketing topics, not just my brand name.

I appeared in a Google AI Overview for a local category search. Nine days after making my on-site changes, I appeared in Google AI Overviews for the search "bay area ai consultant LLM visibility." That's a category search. Someone who didn't know my name could have found me. It's one data point, not a consistent result yet, but it confirms the signal is there and the door is open.

What didn't move, and why

Broad national category visibility is still zero. If someone asks "who are the best AI marketing consultants for mid-size brands?" without knowing my name, I don't appear. Not yet.

I ran additional geo-specific prompts across all four tools to understand the landscape better. The pattern was consistent across all of them. Agencies with established addresses and years of citations dominate these results. Individual consultants and boutique firms aren't breaking through yet, and not just me. Anyone.

I thought that was interesting, so I kept digging. When I asked Claude for recommendations, it actually told users to search LinkedIn filtered to "LLM visibility" or "GEO consultant" in the Bay Area to find independent practitioners. My LinkedIn profile is optimized for exactly those terms. So I'm findable through the channel Claude recommends, even if I'm not in Claude's direct answers yet.

"Be wary of anyone who just added 'LLM SEO' to their services page recently without demonstrated results."

One response I kept coming back to from Claude: "Be wary of anyone who just added 'LLM SEO' to their services page recently without demonstrated results."

That's exactly what I'm building against. I'm not adding language to a services page. I'm running a live, documented, transparent test on my own brand and publishing the results, whatever they show.

Most of my changes were made in the last two weeks of the window. My Wikidata entry, directory listings, AI Journal feature, and blog posts were all completed in mid to late April. AI systems need time to process these signals. Sixty to ninety days is a more realistic timeline for them to compound.

Claude is my weakest signal. When I ran the brand prompts in Claude, it pulled older information rather than my current positioning. Claude relies heavily on structured data sources that I've addressed but that haven't fully compounded yet. There's also a naming collision I'm still working through. Several other companies use variations of the name "NextWise," and making it clear to AI systems which one is which takes time.

One measurement tool hasn't updated yet. One of the tools I use runs on a 30-day update cycle and its last refresh was before most of my interventions were complete. I'll include that updated data in my Day 60 report.

What I learned about measuring LLM visibility

This turned out to be one of the most useful parts of the whole experiment.

Measuring LLM visibility is genuinely hard right now. The tools are early stage. Some update monthly, not daily. Some require minimum traffic thresholds before they show you anything useful. And the most important signal, whether you show up in category searches, can only be tested manually by typing prompts yourself.

There is no single dashboard that gives you the full picture. You need multiple sources, and you need to understand the limitations of each one.

I also learned that prompt specificity matters more than I expected. Broad category prompts surface agencies. Geo-specific prompts open the door for smaller firms. Ambiguous prompts confuse AI tools into asking clarifying questions rather than giving recommendations. The way you phrase the question changes the answer completely.

What comes next

The foundation is in place. The signals are being read. But showing up when someone searches for what you do without already knowing your name takes longer than 30 days to build.

Day 60 is where I hope to see more movement. By then, the April and May interventions will have had more time to compound. My Crunchbase profile will be active. A schema update connecting my website directly to my Wikidata entity will be live. And I'll have more third-party citations behind me, including this series itself.

The space is genuinely early. Most of the firms showing up in these results didn't exist two years ago. The window for building real authority here, before it gets crowded, is open right now.

I've also put together a full measurement report if you want to see the data in one place:

View the Day 30 Report

I'm taking on a small number of brands to run this audit alongside my own. If you're curious what your baseline looks like, get in touch.

Day 60 report publishes June 20.