Don't automate what you've never done. Run it by hand first. A lot of operators
By Jaspinall · November 21, 2026 · Curated by George's Blog
Don't automate what you've never done.
Run it by hand first.
A lot of operators go straight to the schedule. New idea on Monday, daily routine by Tuesday. A bot pulling the search term report every morning, or rewriting bullets every night.
Then the output comes back wrong and they blame the tool.
The scheduler didn't fail. The job was never run once.
Here's the order I use now.
First, do the job with the bot in chat, one time. Pull the report. Write the bullet. Build the reorder list. Watch what it gets wrong.
Then fix it right there. Wrong date range, wrong ASINs, copy that sounds like a robot wrote it. Keep correcting until the output is something you'd actually use.
Then save what worked as a skill. That's the SOP the bot follows next time, written down so it doesn't have to guess.
Only then put the skill on a schedule.
Real example. This post went out at 10 AM from a routine. That routine follows a saved skill with rules like 300 to 450 words, no hashtags, no em dashes, and links in the first comment instead of the post. None of that came from the scheduler. The schedule just runs whatever method you hand it.
Skip the hand run and you're automating a guess. A consistent guess, every day, at scale.
This matters more right now than any other week of the year. Black Friday is a week out. A routine watching ad budgets, stock levels or price changes over that weekend should not be doing the job for the first time on Black Friday morning.
So pick one thing you were planning to schedule for BFCM.
Run it once by hand this week. Fix what comes back wrong. Save it as a skill.
Then let it run.