At Cirque du Soleil, I led an insights and analytics team that brought qualitative research, quantitative research, and data science together. That team became the model for OhGee Insights, the firm I founded and own today. Many research firms are organized around one specialty. OhGee brings all three capabilities into the same project. We start with the business problem and then choose the method and the right people for the work. We are also a small, senior team. Everyone has deep technical expertise, and we are business people who connect the research to an outcome the client can use. Clients tell us the process is easy. They also know we care about what happens after the deliverable, which is why many of them come back.

Almost all of our new business comes through existing relationships: former colleagues, past clients, and referrals to their peers. It is a warm, trust-based funnel, and it converts well. The channel I want to grow is inbound. I want people who have never met me to find OhGee through our work and point of view. I have started investing in that by sharing methods and case work on LinkedIn, building tools other researchers can use, and publishing how I think AI fits into research.

Handed over: the first draft of the deliverable. A year ago, turning analysis into a client-ready PowerPoint took a full day of copying, pasting, and formatting. Today we have a set of Claude-based skills that know OhGee’s brand standards, our proposal structure, and our analysis methods. We use a similar setup to scrape and code thousands of public reviews. Took back: qualitative synthesis. I tested AI on first-pass transcript analysis, including themes and key quotes. The output flattened too much of the meaning. It missed hesitation and gave too much weight to the person who talked the most, which blurred the line between what a respondent said and what they meant. In qualitative work, the insight is often in the thing that was almost said. I now do the synthesis myself. Afterward, I use AI as a check: it can surface a theme I may have missed, pull quotes I have already tagged, or challenge my interpretation.

So far, every inquiry I can trace still comes through people, and I am working to change that. The day someone opens a call with “My LLM recommended you” will be a first for us. A large firm may have hundreds of backlinks, while a boutique’s reputation can live mostly in people’s heads. What has changed is how much people know before they reach out. Three years ago, a first call was mostly educational. Today, prospects often arrive after asking AI to explain the methods, sometimes with a draft research design in hand. Some of those drafts are good. The first conversation now starts further along. We can focus on where the AI is useful and where human judgment changes the answer. I get to add judgment much sooner. Visibility is changing too. AI assistants need sources they can cite. A clear, published point of view gives them something to work with, which is one reason I am putting more of our methods and thinking in public.

I would stay curious and experiment with AI. Build a small tool around a process you already understand. If you can describe how you format a deck or clean a dataset, you can encode it. The tool will pay you back on every project that follows. Start anywhere. Put your thinking in public so people and AI assistants can understand your expertise before they meet you. Published methods and case work give them something concrete to find and cite. And ignore anyone who calls themselves an AI expert. The field is moving too quickly for that label to mean much.