Am I a Luddite for not integrating our payroll and financial systems?

We’re working on a NetSuite go-live for January. An opportunity to eliminate an Excel interim upload to the financial system will save five minutes every two weeks and make it less prone to manual error. Great.

Until something changes in either system. Then we spend more time updating and testing the interface than the upload ever required. Net time savings and error avoidance? Maybe negative.

Perhaps I’m not a Luddite after all.

That does not mean the integration has no value. A consistent interface provides control, repeatability and an audit trail. A well-documented and controlled spreadsheet can provide those things too. Excel can and has done that reliably now for 30+ years, and five minutes every two weeks is barely two hours a year. What continuing obligation are we creating to eliminate that five-minute task?

Every automation solves for something. The danger begins when the thing it solves becomes more important than the reason we were doing it. We’re sticking with the Excel bridge for now.

Consider another productivity opportunity in the basic calendar. We used to keep handwritten notes. You cross out a meeting, move it to Thursday and forget to tell someone. Digital calendars eliminate that miscommunication. We made calendars more efficient. Others could see when we were available and schedule meetings directly. No phone calls, back-and-forth messages and no human interaction required. Excellent.

Have you gotten the calendar invitations from people you have never heard of hawking products you never wanted? It is the new spam. The same open access that eliminated the scheduling friction also eliminated the friction protecting our time.

No problem. We can automate that too. Our AI assistants will screen the requests, manage our calendars, schedule our meetings, remind us to attend and write the follow-up emails. All increasingly unprompted. That rogue email? Sorry. I did not write it. My Muse made a mistake. Except it was my Muse, acting in my name, using my account, communicating with another person who thought they were communicating with me. Can we automate the action and disclaim the responsibility?

Meanwhile, automated outreach becomes so inexpensive that everyone receives more of it. More email, calls and invitations proliferate. More perfectly written personal messages are created without a person. So we automate the defense. Ignore the calls, screen the emails and reject the invitations. Trust nothing unless the known sender knows the secret code word validating their authenticity. Zero trust. Cloistered from the very efficiencies the technology first promised.

Awesome efficiencies and time ROI until they’re not.

Sure, the automated task may be faster and neater. But faster for whom? Smarter about what? What happens to the larger process surrounding the task we just perfected?

The old flexible spreadsheet bridging two systems may look like an obvious candidate for elimination. Perhaps it is. But the spreadsheet is not merely a manual process waiting to be automated. It may also be absorbing the differences between two systems that do not contain the same truth. It bends when the business changes. Someone can see the exception, understand why it happened and adjust.

The automated interface will be more consistent. It will also be less forgiving. Spend too much time aligning disparate systems so they can talk efficiently and the interface itself can become the objective.

Then the proposal expands. Could the interface also give us more precise labor costing?

Now the business adapts to satisfy the integration. People maintain fields because the interface requires them. Operators spend time feeding the costing model instead of doing the work the costing model was supposed to help us understand.

Measure more closely and behavior changes. Sometimes that is the control we needed. Sometimes people simply become better at satisfying the measurement.

This is not an argument for keeping everything manual. Manual processes fail. Spreadsheets can contain errors. Institutional knowledge disappears. A flexible workaround can be another name for a control no one documented and only one person understands. So rigidity is sometimes precisely the benefit.

But automation has an ROI, and the burden it creates belongs in the calculation. Design, testing, monitoring and maintenance. Lost flexibility. New dependencies. Human work transferred from the person performing the old task to everyone now required to keep the automated process alive.

And we all need new defensive automation to protect us from everyone else’s automation.

Maybe the question was never whether we could automate the five-minute upload. Maybe the question was what larger process we would create by doing it, and whether that process would leave us any better able to accomplish the original goal.

No, I’m no Luddite.

Some work even benefits from a slower pace. Direct human interaction. A letter written with pen and paper. Time spent finding the words yourself.

Never mind. Claude can express my feelings better anyway.