Selling line · stop 03 of 6 · 22 min · members
Two hours a week reduced to one approval, with the parts that must stay human.
Free with an account
Membership is free: an account opens all 86 script pages. The Lab, Studio Canvas and the paid guides need the $99 pass, paid once. Already signed in on this browser? The page opens by itself.
The problem
Not from lack of ideas. From the cadence being expensive enough to skip.
Every shop that goes quiet went quiet the same way: a busy week, one missed listing, then two, then the habit is gone. Nobody decided to stop.
The cause is that publishing a listing involves a dozen small tasks — pick a topic, make the product, make five images, write a title, write a description, upload, tag, price — and each is trivial while the pile is not. When the week is full, a two-hour pile gets postponed, and postponed indefinitely.
The fix is to make the pile small enough that a busy week cannot displace it. Not by working faster, but by removing decisions.
The principle
Almost everything in the pile is mechanical. Almost nothing needs you.
Go through the list and mark each task as a decision or as production. Decisions require judgement: which topic, is this good enough to publish, what price. Production is everything else.
In a typical listing, one or two items are decisions and the rest are production. Yet they are usually done together, in one sitting, which means the whole block requires your full attention and gets scheduled accordingly.
Split them and the shape changes: production happens automatically ahead of time, and your involvement collapses to approving or rejecting something that is already finished.
Building it
Three pieces, none complicated, that between them remove most of the work.
The topic pool. A file of candidate topics, added to whenever one occurs to you. The weekly job draws from it rather than asking you to think of something on the day. Include a used-marker so the same topic is not drawn twice.
topics.json — [{ topic, status: pending|used, added }]The template. The listing image layout, the description structure, the tag set. Once these are fixed, producing a new listing is filling in fields rather than designing.
The schedule. A job that runs on a fixed day, draws a topic, produces the assets and the draft listing, and leaves it somewhere for you to look at.
The output of the automation is a draft, never a publication. That boundary is what makes the whole thing safe to run unattended.
What stays human
The automation should stop short of these, deliberately.
The final yes. Nothing publishes without you looking at it. This is the one rule that makes everything else acceptable — an automated pipeline that can publish is a pipeline that can embarrass you at three in the morning.
Pricing. It is a judgement about positioning, not a field to fill. Set it deliberately, per product.
Anything responding to a customer. Reviews, questions, complaints. Automated replies read as automated and cost more goodwill than they save time.
Everything else — asset production, image layout, description drafting, tagging, scheduling — can run without you.
Failure modes
Design for the week when it breaks and nobody notices.
The characteristic failure of automation is silence. The job stops running, or starts producing something wrong, and because the output was going into a folder you check on Saturday, you find out days later.
Two defences. First, the job should report whether it ran, successfully or not — a message either way, so absence of news is itself information. Second, it should refuse to proceed on obviously bad input: an empty topic pool, a missing template, a failed asset generation should produce a stop and a notice, not a half-built draft.
Also build the duplicate check early. A pipeline that produces the same topic twice because the used-marker did not write is embarrassing in a way that is entirely preventable.
The result
And the ten minutes are the part that actually required you.
The end state is a message on a fixed day saying a draft is ready, a link to look at it, and a decision: publish, adjust, or skip. Ten minutes, at most.
The cadence now survives a busy week, because ten minutes always fits. That is the real return — not the time saved, but the fact that the habit stops being fragile.
Build it once, with the boundary in the right place. Everything up to the approval is machinery; the approval is you.
1:1 · two hours