When NOT to rebuild your website: the 12 questions we ask first

“We need a new website” is the most common opening line we hear — and in more than half of those conversations, the correct answer turns out to be “no, you don't”. That answer earns less money up front, which is exactly why you rarely hear it from people who sell rebuilds.

A rebuild is not a purchase, it is a trade. You trade known bugs for unknown ones, accumulated search equity for a redirect map, and six months of roadmap for a site that — on launch day — does roughly what the old one did. Sometimes the trade is right. The trick is deciding with a checklist instead of with fatigue.

The invoice nobody shows you

• SEO equity burns quietly. Even a flawless migration with perfect redirects wobbles for months; an average one loses rankings for good. If organic search feeds your funnel, this single line item can eat the whole business case.
• The roadmap freezes. While the new site is being built, the old one stops improving. Six months of “after the relaunch” is six months of compounding neglect.
• New code, new bugs. The old site's bugs are documented in your support inbox. The new site's bugs are a surprise party.

The three-fix pattern

Most “it's slow, ugly and doesn't convert” complaints trace back to two or three concrete, fixable things. We have lost count of how often the list is exactly this: a multi-megabyte hero image, one synchronous third-party script, and a missing database index behind the slowest page. Days of surgery, not months of reconstruction — and the effect shows up in the same quarter's numbers.

So that is always step one: a findings-first audit. If the top three fixes recover most of the pain, the rebuild conversation is over. If they don't, at least the rebuild starts from evidence instead of exhaustion.

The 12 questions

• 1. Where does revenue actually enter? If organic search is a top-three channel, a rebuild carries genuine risk of self-harm.
• 2. Have you measured where the funnel leaks? “It doesn't convert” without step-by-step numbers is a feeling, not a finding.
• 3. Is it slow on the server or in the browser? A two-second TTFB and a two-second render freeze have completely different fixes — neither of them is a redesign.
• 4. Can your team ship a landing page in a day? If the CMS blocks routine marketing, that is a real rebuild argument.
• 5. Is the design dated, or is the brand changing? Cosmetics renovate in place. A genuine brand pivot doesn't.
• 6. Is the platform alive? Vendor abandonment trumps everything — more below.
• 7. Does the site meet your security floor? Unpatchable dependencies argue for rebuild; missing headers are an afternoon.
• 8. Does the data model still describe the business? Fighting your own schema on every feature is structural, not cosmetic.
• 9. What does a small change cost? If a text edit needs a developer and a deploy window, the platform taxes you daily.
• 10. How large is the redirect surface? Ten thousand indexed URLs are ten thousand chances to leak equity.
• 11. Who maintains it in year two? Rebuilding into a stack nobody in-house knows is a subscription to consultants.
• 12. What breaks if you touch nothing for a year? If the honest answer is “not much”, you have time to renovate instead.

Reading the answers

Mostly renovation-flavoured answers: fix the three things and keep shipping. A mixed picture: consider the strangler pattern — rebuild one high-value slice, checkout or catalog, behind the same URLs while the rest lives on. Structural answers on questions 6 and 8: rebuild, and do it deliberately, with the redirect map treated as a first-class deliverable.

When the answer really is rebuild

We are not anti-rebuild; we are anti-default-rebuild. Last year one of our own projects lost its backend-as-a-service provider — one morning the DNS for the project's database simply stopped resolving. No amount of renovation fixes a platform that has ceased to exist. We moved everything onto plain PostgreSQL on our own server, and the only regret was not doing it before an outage set the calendar for us.

Platform death, a data model that fights the business, licensing that eats the margin — those are rebuild reasons. “The homepage feels old” is not. Audit first. The architecture can wait a week; your funnel shouldn't wait six months.