An industry visit, a conversation with a BI analyst who spent his whole career in finance, and the realisation that shallow insights aren’t a data problem — they’re a domain problem.
This week we had an industry visit.
I’ll be honest, I wasn’t expecting much. Industry visits have a way of feeling like a field trip more than anything else. You show up, someone talks at you in a boardroom for an hour, you nod, you leave. But this one was different. Not because of the seminar. Because of what happened after it.
The company is based in Midrand, a logistics and retail operation that supplies luxury perfumes to retailers across the country. Before we even got to the entrance, you could already read what kind of place it was. Chalk marks lining the parking lot where the trucks pull in, big and deliberate, the kind of infrastructure you only build when you’re moving serious volume. Then the reception, perfumes displayed everywhere, the whole product range just sitting there like a statement. This is what we do. This is how we do it.
The seminar room itself was plain. Minimalistic. Corporate in the way that says we didn’t decorate this room because nobody here cares about decorating the room. And honestly, that was fine. The room wasn’t the point.
The web analytics conversation was brief. For good reason.
Someone in the group asked about SEO and web analytics. The answer was short and it made complete sense, they’re a B2B middleman. They supply to retailers. Their clients aren’t finding them through Google. So the web presence isn’t really the thing. That conversation closed almost as soon as it opened, and I understood exactly why.
But I wasn’t really there for that conversation anyway.
I chased down their BI and DA analysts on the way out.
I’d asked a question during the session. Nothing groundbreaking, just enough to get on their radar. And as people started filing out of the room, I called them back. Just hey, can I ask you something? They stopped. They talked to me. And I’m genuinely grateful they did because what came out of that five minutes changed how I’m thinking about everything.
I asked the thing I’ve been trying to figure out for months now. How do you get to the point where your insights actually mean something? Where you’re not just producing numbers on a screen but producing something a business can actually use?
The analyst looked at me and said something I wasn’t expecting, he told me he spent his entire academic career in finance. Master’s degree in finance. Switched to BI later. And because of that, when he walks into a meeting with the finance team, when he’s building a report or a dashboard for them, he already knows. He knows what a number means before he’s finished typing it. He knows what to track and what to leave out. He knows what question the business is actually asking before they’ve finished asking it. That fluency, that second nature, didn’t come from learning a tool. It came from years of living inside the domain.
Domain is everything. The tools will come.
That’s what he said. And I’ve said versions of that in this newsletter before, but hearing it from someone who’s built an entire career on that principle, it landed differently.
And then something clicked while he was still talking.
Not a lesson. More like a cold realisation.
I’ve been spending a lot of time on datasets. Grinding through the process, repeating it, trying to get reps in. But here’s the thing, if the process is wrong, doing it twenty times doesn’t make it right. It just makes you faster at being wrong. I could spend months analysing e-commerce data and never notice that the metric I keep reaching for doesn’t actually reflect what’s driving the business. Not because I’m lazy or careless. Because I don’t have enough domain experience to know the difference.
Without that reference point, I can’t ask the right questions. I can’t tell whether what I’m building matters. I can’t feel when something looks right on paper but doesn’t track with how the business actually behaves. The shallowness I’ve been frustrated with, the feeling that my insights don’t go deep enough, isn’t a data volume problem. It never was. It’s a domain problem. And more data was never going to fix it.
What I need to build — and what I’m only just starting to understand.
The gaps that matter most for where I want to go are marketing trends and consumer psychology. I’ve touched marketing but not deeply enough. And without that, there’s a whole layer of context I’m missing — why a product that looks like it’s bleeding might be sitting on the shelf intentionally, what’s happening in the market that makes a certain category behave the way it does, why a consumer drops off at exactly that point in the checkout.
Retail is the sector I want to be in long term. I’m drawn to it because what you’re tracking is tangible in a way that feels real to me — stock moving in and out, products sitting still when they shouldn’t be, spikes that tell you something changed. These aren’t abstract numbers floating in a spreadsheet. They’re connected to physical things happening in physical spaces. That grounding matters to me.
But to do that work properly, especially as a freelancer who has to wear every hat by themselves. I need to actually understand what each hat means before I put it on. Right now, I’m wearing some of them blind.
What the visit confirmed about the classroom.
Last week I wrote about how the classroom prepares you for corporate environments. Walking into that building and seeing how it actually operates confirmed something. Nobody there works alone. The BI analyst isn’t sitting in a corner producing dashboards into a void, he’s embedded in teams, surrounded by people who carry the domain knowledge he leans on. The finance team teaches him things no course could. The operations team gives him context. That scaffolding exists in a corporate environment almost automatically.
As a freelancer, none of that scaffolding is there unless you build it yourself. That’s something the classroom is actually getting right, even if it doesn’t realise that’s what it’s doing for someone in my position. It’s showing you what the pieces of the machine look like before you have to run the whole machine alone.
So here’s what I’m changing.
I’m going to stop grinding through datasets in isolation and start working alongside other people. Volunteering. Getting into rooms where I can ask the questions that don’t have answers in a spreadsheet. Learning the domain the only way you actually can, by being around people who already live in it.
Because if you’re reading this and your insights feel shallow, I don’t think the fix is more data. I think the fix is more context. The numbers aren’t the problem. Understanding what they’re supposed to mean is.
Domain is everything. The tools are just how you get there.
Until next time — ponder this:
Are you getting better at the analysis, or just getting better at doing the wrong analysis faster?