Automation: end of year summary

2025 was the year automation became normal. Not an experiment anymore, but an expectation. And like any expectation, it brought disappointments.

Here's an honest summary: what worked, what didn't, and what we learned from it.

What worked

1. Email sorting and summaries

This was the year's most popular solution. Not because it's flashiest, but because it solves a real problem.

Typical result: 3-5 hours saved per week. Some clients saved more, some less. It all depends on email volume and how much was done manually before.

2. Connecting data between systems

"I have to enter this info in three places" - I heard this sentence a lot at the start of the year. By year's end, many had solved this problem. A practical guide to system integration showed how to do it right.

Most popular solution: form submission that automatically sends data to Notion, Google Sheets, and email. Simple, but saves 15-30 minutes daily.

3. Report automation

Monthly report that used to take 2 days? Now generated automatically. This was one of those projects where ROI was immediately visible.

What didn't work

1. Plans too big

"We want to automate the entire customer journey." That's too big. Too abstract. Too hard to measure.

Projects that started this ambitiously often stalled. Not because technology didn't work, but because focus was missing.

2. Automation without a clear problem

"We want AI because everyone uses AI." That's a bad reason. How to choose the right AI tool? has helpful questions to ask yourself. Several such projects ended up in a drawer because nobody knew what was actually being solved.

3. Forgetting users

Technically brilliant solution that nobody uses. We saw this more than we'd like to admit. The best automation is the one the team wants to use.

Numbers

Yes, about a quarter of projects didn't deliver expected results. That's an honest number. Usually because the problem wasn't clear or users weren't involved.

Lessons for 2026

1. Start small, measure quickly

One well-working automation is better than ten half-done ones. Start with what hurts most, measure results, then expand. AI agents and workflow automation can help here too.

2. Involve users from the start

Before you build, ask: will this be used? If the answer isn't a clear "yes", you need to think more.

3. Automate processes, not problems

If the process itself is messy, automation makes it messier. First clarify the process, then automate.

Summary

2025 was a good year for automation. Not because everything succeeded, but because we learned what works and what doesn't.

For 2026, the focus is clear: less ambition, more practice. Smaller steps, bigger impact.