When I started learning Python, I tried to think like an engineer. I learned about classes, modules, and design patterns. I read about separation of concerns and clean architecture. None of it stuck, because I kept asking the same question: what does this have to do with closing deals and hitting targets?

Eventually I stopped trying to think like an engineer. I started using Python the way I would use a spreadsheet — as a tool that solves a specific problem I already understand.

That reframe changed everything.

How salespeople think about data

People who grew up in sales think about data in a specific way:

• Revenue is the north star. Every metric connects back to it.
• Pipeline is the unit of work. Not tasks, not tickets — opportunities with probabilities and values.
• Time is the constraint. Forecasting matters because we need to make decisions before we have complete information.
• Relationships have signal. Who talked to whom, how often, and what happened next tells you more than most dashboards.

This is a coherent mental model. It’s just not the mental model that most data tools were built for.

Where Python fits

Python is useful to me because it can work with my data in my mental model.

I don’t need to restructure my thinking to fit the tool. I need a tool that can pull CRM data, calculate pipeline coverage, flag deals that haven’t been touched in 14 days, and format everything into a table I can act on.

Python does that. And once you can do that, you start seeing more:

• Which campaigns actually generate pipeline, not just leads
• Which lead sources convert at higher rates over longer time windows
• Where in the funnel deals stall and why

These are questions a sales or marketing person already asks. Python just makes them faster to answer.

Where AI fits

The AI tools I use most are not the big flashy ones. They’re the quiet ones:

LLMs for text classification. I have a script that reads through CRM notes and tags them by topic — competitive mentions, pricing objections, timeline uncertainty. Manually categorizing 500 notes takes hours. An LLM does it in a minute.

Embeddings for similarity search. I can find deals that look like a deal that closed — same industry, similar objections, similar timeline. That’s useful for forecasting and for training new reps.

Summarization for reporting. Instead of reading every call transcript, I generate summaries and flag the ones that mention specific competitors or concerns.

None of this requires a machine learning degree. It requires knowing what problem you’re solving and using the API.

The practical mindset

The way I approach new Python tools is simple:

1. Start with a problem I already solve manually
2. Figure out the minimum script that automates that one thing
3. Run it for a week before building anything more complex
4. Expand only when the first version is stable

This is not how most software gets built. But it’s how most useful things get built by people who aren’t full-time engineers.

The goal is not to become a developer. The goal is to have fewer manual tasks and better answers to business questions.