I spent a weekend wiring up an n8n flow for the BOGO deals app before realizing I was automating something that ran maybe twice a month. The automation took longer to build than doing it by hand would have for the next two years.
The tool discussion I keep having with myself
Every time a new task feels tedious, my brain goes straight to "okay, n8n or Make?" I start sketching nodes in my head. I think about webhooks and cron schedules. It feels productive. It's not, always.
The BOGO app had this thing where convenience store chains would update their weekly deal lists on slightly different schedules, and I'd manually check them. Annoying? Yes. Frequent enough to automate? I never actually checked. I just assumed.
So I built the flow. Took about six hours including debugging auth issues with one retailer's flaky endpoint. It ran. It was kind of beautiful in that "look at all these connected boxes" way.
Then I checked my logs two months later. It had triggered eleven times. Total.
The question that should've come first
Not "which tool," but "does this actually need to happen without me?"
Turns out there's a rough filter I wish I'd applied earlier. I started asking three things before touching any automation canvas:
How often does this actually happen? Not how often I imagine it happens — how often it actually happens. I was wildly wrong about the BOGO update frequency. I thought it was daily. It was closer to weekly, and half the time I didn't need the data anyway.
What's the real cost of doing it manually? Six hours of n8n work versus maybe thirty seconds of manual checking, multiplied by eleven occurrences over two months. The math was embarrassing.
What breaks if the automation fails silently? This one matters more than people think. A flow that dies quietly is sometimes worse than no automation at all. I had one Make scenario that stopped running after an API schema change and I didn't notice for three weeks. The manual version would've caught that immediately because I would've looked at the thing.
Automation debt is real — a broken flow you're not watching is worse than a manual step you do every time.
Where I actually landed
I didn't delete the n8n flow. But I rebuilt it much smaller. Instead of trying to scrape and normalize and notify and log everything, it now does one thing: pings me when a specific change threshold is crossed. The rest I just do by hand when I feel like it.
The other thing that shifted for me was separating "I hate doing this" from "this should be automated." They feel the same in the moment. They're not. Some tasks I hate because they're genuinely repetitive and rule-based — good automation candidates. Others I hate because they're annoying and unpredictable, which means automation just adds a maintenance layer on top of already messy work.
Hating a task isn't the same as having a task worth automating.
The BOGO scraping was messy. Store websites change their markup. Sale categories shift. Every month I was tweaking the flow anyway. I basically had a part-time job maintaining an automation I built to save time.
What I actually use automation for now
Stuff that is genuinely repetitive, genuinely rule-based, and happens often enough that the setup cost pays back in under a month. That's a much shorter list than I expected. Sending a weekly summary digest — yes. Reformatting uploaded CSVs before they hit a database — yes. Monitoring whether a deal is still live and updating a status field — yes, because it runs hundreds of times a day.
Everything else? I just do it. Or I put it in a list and revisit whether it's still a problem in a few weeks. Half the time the problem goes away on its own.
The best automation is sometimes the one you decided not to build.
n8n and Make are genuinely good tools. I use both. But I burned enough hours on flows that outlived their usefulness before they were even finished to know: the tool choice is the easy part.
TL;DR: Before picking n8n vs. Make, ask whether the task is frequent enough, predictable enough, and high-cost enough to justify the build and maintenance — most of the time, at least one of those fails.
Next episode: I finally tried giving the BOGO app a real monetization layer — and discovered that "people said they'd pay" is doing a lot of heavy lifting as a business plan.
Get the $31 Stack Prompt Pack — $19
The exact prompts, cost calculator, and n8n templates from this stack.
Top comments (0)