Q: Ours is just a showcase site, a dozen or so pages. Can we skip the admin and have you edit the HTML directly and upload it?
A: Yes. And if you won't change it more than twice a year, I'd say don't build one. An admin isn't a standard feature — it's a cost.
Q: Where's the cost?
A: The visible part first. A minimal content admin — section management, rich text editing, image upload, account login, four modules — was roughly 8 to 10 working days by our 2016 numbers. That money could redesign your homepage twice. Then the invisible part: once an admin exists, someone maintains it — PHP versions, vulnerabilities in editor components, who holds the admin password, how access is revoked when someone leaves. Skip that upkeep, and the admin is more dangerous than static pages.
Q: But our marketing team says they'll publish news themselves going forward.
A: I've heard that line many times, so I ask one more question: how many did you publish last year? We actually checked their site — the previous year's 'News' section held 3 posts, 2 of them reposted from elsewhere. Then the answer is clear. Better to pin down the publishing process first — who writes, who approves, how often — run it for three months, then decide if you need a system.
Q: Is there a middle ground?
A: Yes, and we use two. One: extract just the frequently changed pieces — the homepage's three 'latest updates,' the contact phone number, the carousel — into a minimal admin, or even an online spreadsheet, with a cron job generating a static JSON every 10 minutes for the frontend to read. Small blast radius, easy rollback. Two: use an off-the-shelf CMS but grant a single editor account with templates locked — no layout permissions at all.
Q: When would you talk someone into a full admin?
A: When two of three signals appear. One: content items exceed 50 and keep growing. Two: more than one department needs to add content, so permissions matter. Three: the content must connect to other systems — a product database, a store list, registration data. At that point hand-written HTML is the expensive option: every change means finding a developer, deploying, verifying.
Q: So what's the conclusion?
A: No conclusion. I usually ask for their update history from the past six months — even a chat log will do — and quote after reading it. 'We'll update it often,' said with nothing to show, usually means they won't.
That day the client chose static pages plus one online spreadsheet, leaving news out entirely. Six months later they did come back for an admin — they'd hired someone dedicated to content. This time we didn't argue; we scheduled it.
