Publishing automation: 10 jobs to hand the machine, 3 decisions to keep

The phrase “publishing automation” produces two opposite reactions. One is awe: “Let everything run by itself.” The other is resistance: “Publishing is a craft; it can’t be mechanized.” Both share the same false assumption — that automation will take over the decisions.

But good automation targets drudgery, not decisions: carrying information, remembering, chasing, copying. It speeds up the decision; it does not make it. In this post, let’s draw a concrete map: ten jobs a press can hand the machine today, and three decisions that should never leave a human desk.

10 jobs for the machine

1. Submission completeness checks. If required fields and files are defined, an incomplete submission cannot get through the door. The editorial office’s “abstract missing” emails become history.

2. Work codes and record creation. The moment a submission arrives, the work code is stamped and the file and workflow are opened — untouched by human hands, in a standard shape.

3. Reminders. A reviewer is overdue, a proof approval hasn’t come back, a contract expires in a month, a task has sat idle for two weeks. Let the system that knows the calendar send the polite nudge; when a human does the chasing, both the chaser and the chased wear down.

4. Clearing unanswered invitations. When the set period runs out, let the reviewer invitation time out on its own and let the pool suggest a substitute. The editor should manage the suggestion, not the waiting.

5. Document generation. Acceptance letter, rejection letter, contract, receipt — from the institution’s template, with the record’s data, at one click. Every hand-filled document is a typo waiting in ambush.

6. Notification routing. The right message through the right channel: to the author when the decision lands, to the reviewer when the assignment is made, to the customer when the parcel ships. Manual “letting people know” doesn’t count as work until the day it’s forgotten; on that day it counts as a crisis.

7. The publication-moment chain. When an issue or a book is published, the archive page, the work’s status, the store display, and the access rights all update together. A manually run publication day is the day of the forgotten step.

8. The financial records of a sale. When payment arrives: invoice, journal entry, stock decrement, royalty accrual; on a return, the same chain closes in reverse. The end-of-month shift spent “entering sales into accounting” is, in its entirety, work that should never have existed.

9. Stock alerts. A title that drops below its critical threshold should announce itself. An out-of-print book announcing the fact through a customer complaint is the most expensive form of notification there is.

10. Process records. Who did what, and when — the audit trail accumulating on its own. Hand-kept minutes are both incomplete and unfindable at the exact moment they are needed.

3 decisions that stay on a human desk

The publication decision. Accepting or rejecting a work is not the arithmetic of reviewer reports; it is editorial judgment. The system lays out the reports, the votes, the history — the editor decides. (Preventing a vote from being mathematically inconclusive is a different matter; an odd number of reviewers and clear decision rules are about avoiding deadlock, not automating the decision.)

Print approval. Print runs are among an institution’s most irreversible expenditures. The system makes it easy to see that everything is ready; a human gives the “print” order.

Contract and royalty terms. Rates, durations, rights — these are matters of negotiation. The system generates the contract from a template and tracks it; a human decides the rate.

What these three have in common is this: they are costly or impossible to undo, context-sensitive, and responsibility-bearing. That is exactly why the principle “the system prepares, you pull the trigger” exists — even where we are most enthusiastic about automation.

The measure of automation: hours won back

The question “did the automation work?” is answered not by feel but by two simple measurements.

The first is turnaround time: the calendar time from a job’s start to its finish. For instance, how many days pass from a submission’s arrival to the desk-review decision? Comparing that number before and after automation is the most honest report card an investment can get. In institutions that automate the completeness check, this is usually the first number to fall — days become hours.

The second is touch count: how many human interventions did it take to complete a job? For the acceptance letter, the manual routine goes: open the template, copy the details in, convert to PDF, save, attach to an email, send, file it — seven touches. The automated routine: generate and approve — two. Even at a stingy two minutes per touch, in an institution producing three hundred documents a year the difference is measured in weeks.

Even making these measurements has a precondition: the process must leave time-stamped records. A process run by hand does not even produce the data to measure its own inefficiency — the first benefit of automation is, very often, the institution seeing itself in numbers for the first time.

Where to start: the drudgery that hurts most

The classic mistake of automation projects is starting everywhere at once. A better way: ask your team one question — “Which mechanical task eats the most of your week?” In most institutions the answer lands in the same top three: chasing incomplete submissions, reviewer reminders, the sales-to-accounting transfer.

Start with that one job. The hours the first automation wins back become the second one’s best advocate.

If you want to see how the ten items above come ready-made in Nasirus, browse the modules or write to us with your most painful piece of drudgery — we’ll build the demo around exactly that job.