An Honest Review of the Tools I Actually Use to Automate My Business

A few months ago I watched a founder spend forty-five minutes manually copying leads from a web form into a spreadsheet, then into her CRM, then sending a welcome email by hand. She did this every single day. When I asked why she hadn't automated it, she said she'd looked into the tools and it felt like "too much to figure out."

That's the problem. Not a lack of tools. Too many tools, all promising they'll change your life, none of them being particularly honest about the tradeoffs.

So here's my honest take on the automation stack I've landed on after years of building things, breaking things, and occasionally automating myself into a corner. This isn't a sponsored roundup. It's what I actually use, and why.

Why I Kept Switching Tools

For a long time I chased the "best" automation tool the same way people chase the perfect notebook. Zapier, then Make (formerly Integromat), then n8n, then back to Make, with various detours into no-code platforms that promised everything and delivered about sixty percent of it.

The switching wasn't random. Each tool solved one thing better and failed at another. Zapier is genuinely the easiest to start with. The interface is friendly, the documentation is good, and most of the integrations just work. But the pricing structure punishes you for doing anything interesting. Once your workflows get complex or you need to run them frequently, the cost climbs fast. I wrote about a similar tradeoff when I was rethinking where to host different parts of my stack, sometimes the easy default costs you more than you realize.

Make is more powerful and considerably cheaper for the same workload. The learning curve is steeper, though. Thinking in "scenarios" with modules takes some adjustment if you're used to Zapier's linear trigger-action model. But once it clicks, you can build things that would be genuinely painful in Zapier.

n8n is the open-source option, and if you're comfortable self-hosting, it's worth a look. You own your data, you pay nothing per execution, and the flexibility is real. The tradeoff is that you're responsible for keeping it running. That's fine if you have the technical appetite. It's a headache if you don't.

The Stack I Actually Landed On

For connecting SaaS tools: Make

For most of the integration work I do, pulling data from one place, transforming it, pushing it somewhere else, Make is my default now. The pricing makes sense at scale, the error handling is better than Zapier's, and the ability to run complex logic inside a scenario without paying per step is a genuine advantage.

According to research from Zapier's own State of Business Automation report, small business owners who automate report saving an average of several hours per week. That tracks with my experience, though the number varies wildly depending on what you're automating.

For internal tools and databases: Airtable with reservations

Airtable is great until it isn't. For lightweight databases, shared team views, and things that need to live somewhere between a spreadsheet and a real database, it works well. I've used it as the backbone of a few client projects and it held up fine.

The reservation is that Airtable's automations are limited. You can trigger simple things, send an email when a record changes, update a field based on a condition, but anything more complex needs to reach out to Make or a custom script. That's not a dealbreaker, just worth knowing upfront.

For AI-assisted tasks: a mix of things I'm still figuring out

This is the part that's changed most in the last eighteen months. Adding AI to automation workflows used to mean writing a lot of custom code. Now there are reasonable options for calling OpenAI or Claude inside a Make scenario, which opens up things like classifying incoming support tickets, summarizing long documents before they hit a database, or generating first drafts of routine emails.

I'm still learning where this is genuinely useful versus where it adds complexity without much payoff. My current rule: AI in automation is great for tasks where "good enough" is fine and human review would catch anything terrible. It's not great for anything where you need the output to be reliably precise.

What I Tried and Dropped

Zapier

I still have a Zapier account and I use it for quick prototypes. The speed to get something working is hard to beat. But I stopped relying on it for anything production-level when the bills started not making sense relative to what I was getting.

For context: Zapier's pricing starts free and scales based on tasks per month. Once you're running a few hundred automations a day, you'll want to do the math and compare it to Make.

Notion as a database

I love Notion for writing and documentation. As a database for anything that needs to be queried reliably by other systems, it's frustrating. The API is fine for simple reads and writes, but it's not a real database and it wasn't built to be one. I moved off it for anything that needed genuine reliability.

Integrations built entirely in custom code

Early on I wrote a lot of small scripts to handle integrations. Python scripts running on a server, cron jobs, that kind of thing. Some of those are still running and I regret every one of them. Not because the code was bad, but because maintaining custom integration code is invisible work that doesn't move anything forward. A visual tool that breaks visibly and loudly is better than a script that breaks silently.

What Actually Makes a Tool Worth Using

After enough time with this, I've stopped evaluating tools on features and started evaluating them on three things.

Does it fail visibly? Automation failures are the worst kind because you often don't know they happened until three days later when someone asks why the thing didn't work. Good tools log failures loudly and send you alerts. Bad ones just quietly do nothing.

Can you debug it without a PhD? Some tools give you detailed execution logs. Others give you nothing. Debugging a broken workflow at midnight is much less awful when you can actually see what happened at each step.

What's the real cost at scale? Free tiers are fine for experiments. But run the numbers at the volume you actually need before you build something important on top of a tool.

If you want to see how I apply this to actual projects, the blog has more specific examples, or you can get a sense of the work from the about page.

The Honest Summary

There's no perfect automation tool. There's the one that fits the problem you have right now, at the budget you have right now, with the technical appetite you currently possess.

For most small businesses, Make is worth learning even though it takes longer to get comfortable with than Zapier. For teams that want something lightweight and don't need complex logic, Zapier's free or starter tier is a fine place to begin. For developers who want full control, n8n is genuinely good if you're willing to host and maintain it.

And if you're not sure where to start, or you've got something specific you're trying to automate and want a second opinion on the right approach, get in touch, I'm happy to talk through it. No pitch, no upsell. Just an honest take on whether the thing you're imagining is a twenty-minute setup or a three-month project.