Four months in, I have 30 workflows sitting in my n8n dashboard. Four of them are actually doing anything. That's an 87% abandonment rate on my own automations, which I'd find embarrassing if I hadn't already made peace with it.
The part where I thought I was being productive
It started the way it always does — a specific, real problem. I needed to pipe deal alerts from the BOGO deals app into a Slack channel so I wouldn't miss time-sensitive stuff. Built the workflow in maybe 40 minutes. It worked. It's still running.
That first win is dangerous. Because what followed was three months of building automations the way some people collect kitchen gadgets. Each one felt justified at the time. Each one solved a problem I was pretty sure I had.
By the end of month two, I had 18 workflows. By month four, 30. I self-hosted n8n on a $6/month VPS — so the infrastructure cost is basically nothing, which is exactly the wrong incentive structure for someone with my particular flavor of builder brain. When something costs nothing to store, you store everything.
What the 4 survivors actually do
Let me be specific, because I think the pattern matters.
Workflow 1: Deal alert → Slack. The original one. Fires when a new BOGO deal gets scraped, posts a formatted message with expiry time. Runs maybe 40 times a day. Never touched it after the first week.
Workflow 2: Error log → email digest. Something breaks in the app, n8n catches the webhook, batches errors into a daily summary email instead of pinging me in real time. Saved me from at least a dozen 2am anxiety spirals. Runs daily, zero drama.
Workflow 3: New user signup → Airtable row. Dumb simple. A user signs up, a row appears. I look at that Airtable table maybe twice a week. Didn't need n8n for this — could've been a direct API call — but it's running and I'm not touching it.
Workflow 4: Weekly stats email. Pulls numbers from a few places on Sunday night, formats them, emails me a summary by Monday morning. I've read every single one. This one actually changed how I think about the week.
Four workflows. All of them are either alerts or data moving from A to B. Not one of them is clever. Not one required more than an hour to build and test.
The workflows that survived are boring. The ones I killed were interesting.
The graveyard
Here's a sample of what's sitting dead in that dashboard.
There was a workflow that was supposed to auto-categorize incoming support messages using an LLM call — Claude Haiku via API, about $0.002 per call — and route them to different Notion databases based on sentiment. I ran it 11 times over six weeks. Then I realized I only get maybe 3-4 support messages a week and I was reading them anyway. Total spend: maybe $0.04. Total time saved: zero.
There was a competitor price-tracking workflow that hit three different grocery store websites on a cron schedule. It worked great for two weeks. Then one site updated its HTML structure, the scraper broke, I got a cascade of error emails, and I never fixed it. That one cost me about 4 hours total including debugging time and I got maybe 12 useful data points out of it before it died.
There was an "ideas inbox" workflow. Voice memo → transcription → Notion page. I used it exactly once, realized I don't actually trust voice memos as a thinking tool, and that was that.
In total, I estimate I spent around 40 hours building workflows that are now doing nothing. At even a modest rate, that's real money sitting in a folder called "Inactive."
I kept building because building felt like progress. It wasn't.
The thing I kept missing
There's a specific seduction to n8n (and tools like it) that I didn't fully see until I was staring at 26 dead workflows. The interface rewards construction. Every new node you drag in, every connection you draw — there's a little dopamine loop there. It looks like productivity.
What doesn't show up in the UI is the maintenance cost. Not in dollars, but in attention. Every live workflow is a thing that can break. Every workflow that could break is a low-grade background worry. By the time I had 18 running simultaneously — briefly, during month two — I was spending more time checking on automations than I would've spent just doing the things manually.
I ran some rough numbers. The 4 surviving workflows probably save me around 3 hours a month combined. The deal alerts alone are worth it — I'd have missed things that mattered. The weekly stats email is worth it. The other two are neutral at best.
Against that: the 40 hours building things that didn't stick. Month one and two had basically negative ROI on the automation effort.
Where it actually tipped positive
Month three, I stopped building new workflows and just used the four I had. That's when the real value showed up — not because the workflows got smarter, but because I stopped paying the construction tax.
The weekly stats email, specifically, has compounded. I've now got 16 weeks of Monday morning summaries. That's a dataset. I can see that signups spike whenever I post anything (still figuring out the conversion side), and that the deal scraper has a ~94% uptime on weekdays but drops to about 80% on weekends for reasons I haven't fully diagnosed. That visibility took 16 weeks of a boring, consistent workflow to produce. I wouldn't have it if I'd kept chasing the next clever automation.
Consistency from a dumb workflow beats sophistication from one that breaks.
The honest math
- 30 workflows built over 4 months
- 4 currently active (13% survival rate)
- ~40 hours spent on things now in the graveyard
- Infrastructure cost: $6/month (VPS, same one I mentioned back in episode 14)
- Estimated monthly time saved by survivors: ~3 hours
- Break-even on the graveyard hours: never, realistically
If I'd spent those 40 hours on literally anything user-facing for the BOGO app — more deal categories, a better onboarding screen, fixing the weekend scraper reliability issue — I'd probably be further along. That's the part that stings a little.
The $6/month VPS is worth it for the four that are running. The habit of building automations as a form of procrastination dressed up as productivity? That one I'm still working on.
n8n didn't cause this. My own brain did. n8n just made it cheap enough to indulge.
TL;DR: Built 30 n8n workflows, only 4 survived — the automation graveyard cost ~40 hours and delivered near-zero ROI; the 4 that stuck are all boring, reliable, and do exactly one thing.
Next episode: I finally look at why the BOGO app's weekend scraper reliability drops by 14 points every Saturday, and whether it's worth fixing or just worth admitting.
Get the $31 Stack Prompt Pack — $19
The exact prompts, cost calculator, and n8n templates from this stack.
Top comments (0)