Why Ethan Marsh says website maintenance needs a tested revert path

Ethan Marsh, Growth Director at Freeman NYC
The Unpopular Opinion

Why Ethan Marsh says website maintenance needs a tested revert path

Ethan Marsh is Growth Director at Freeman NYC, a consulting agency in Rutherford, New Jersey that runs e-commerce operations for brands selling into the US market. Marsh explains why maintenance should prioritize tested reversibility and checks of what visitors see over update reports.

Ethan MarshGrowth Director at Freeman NYC
Published November 22, 2026
The takeaway

A maintenance plan should include a tested way to undo every change and a check that confirms the expected content or actions are visible to customers. A rollback alone is not enough if a failure goes unnoticed.

01

What’s one opinion about your industry that most of your peers would disagree with?

A website maintenance plan should be judged by how fast you can undo a change, not by how current the plugins are. Most of what clients buy as maintenance is a patch report. What protects revenue is a revert path you have already tested.

02

What convinced you? Tell us about the moment, client or result that changed your mind.

A one line helper script. I inserted it with a pattern that matched the first script tag on the page, and that tag was not really a script, it was the section’s own JSON data block. The JSON broke and a block that had been rendering 498 items rendered zero. No console error, no page error, nothing in the logs, so every quick check said the site was healthy. The only thing that caught it was re-counting the rendered items. Because the theme had been duplicated before the edit, publishing that duplicate put the 498 items back in about a minute. Under a plugin update report, that section would have stayed empty until a customer complained.

03

What’s the usual way people in your field handle this, and what do you do instead?

The usual approach is a monthly report: plugins and theme updated, uptime figure, security scan clean. Ours starts before the change. We duplicate the live theme first, so publishing the duplicate is the revert, download the files, and write a short restore note listing every change and how to undo each one. Then we verify by counting what a visitor sees, not by confirming the file saved. We also keep an inventory of which core files carry custom code before accepting a platform update. On one store two core files held undocumented changes, including a 3,480 byte inline style block that no manifest listed, with a theme update pending.

04

Where has this view cost you, or where have you had to admit you were partly wrong?

Two places. It costs time and goodwill, because nobody enjoys paying for backups they never use, and on platforms with a theme library limit the duplicates compete for slots. The bigger correction is that I was partly wrong about rollback being enough. A fast revert does nothing for a failure you never notice, and the 498 item bug produced no error anywhere, so verification had to sit next to the backup. I have also been caught by my own testing. Reading an unpublished backup theme set a preview cookie on my session, so for two hours a page I had fixed that morning looked broken again. The regression was in my browser, not on the site.

05

What should someone in your field stop doing, and what should they start doing?

Stop selling update logs and uptime as maintenance, and stop editing a live file by pattern match. Start writing the revert path for each change before you make it, and start closing every task with a count of what the user sees: items rendered, links resolving, orders completing. If you cannot say in one sentence how you would undo today’s change, the change is not finished.

Thank you to Ethan Marsh and the team at Freeman NYC for sharing an honest view with Leaders Perception readers.

Share this interview

Have a view your industry needs to hear?

Leaders Perception publishes interviews with business owners and leaders who see their industry differently. Five questions, about ten minutes, no calls. It is free and every completed interview gets published.

Get featured on Leaders Perception

Explore additional categories

Explore Other Interviews