Every time a site comes in for review, the same two-path question surfaces: rescue it or rebuild it? The previous agency will push one answer. The new agency will push the opposite. The buyer, you, is the only one in the room without an axe to grind. Which is why most of you end up making this decision badly.
The default mistake looks like this. The rescue you should have commissioned becomes a rebuild because the new agency makes more money that way. The rebuild you should have commissioned becomes a "salvage job" because the agency underpriced the pitch and is now trying to stay inside scope. Either way, the outcome is wrong, and you only find out six months later.
This piece is the five questions we walk through before recommending either direction. Run them yourself before you sign anything.
01 · How much of the current content is worth keeping?
Not "could you keep it", but "should you". Content written for the previous brand, the previous positioning, the previous product line, that's not content, that's sunk cost dressed up as asset. If more than half of what's on the site today is going to be rewritten anyway, you're not rescuing a site; you're rescuing a domain name and some URLs.
If the content is genuinely good (most isn't) and the platform underneath is the only problem, a rescue wins. If the content needs a full pass alongside the technical work, a rebuild is usually faster end-to-end.
02 · How stable is the platform underneath?
A well-built WordPress on a modern stack can absorb almost any rescue work cheaply. A badly-built WordPress on an out-of-date PHP version, with a custom theme compiled by a contractor who left, cannot. The question isn't "is it WordPress" (platforms aren't the problem). The question is "would a competent developer open this codebase and want to stay or want to leave".
Open the repo. Check the commit history. Check the framework version. Check the dependency tree. If you can't do that yourself, have it done. Platform stability is the single biggest variable.
03 · How good is the current design actually?
Here's where most buyers get sentimental. You paid for a design two years ago. You liked it. You still like it. But is it converting? Is it scaling to the three new product lines you've added since? Is it accessibility-compliant? Is it mobile-first in a way that reflects how your buyers actually arrive?
If the design is working and needs a refresh, rescue it. If it's not working and you keep trying to force it to, rebuild it. Nobody ever regretted a rebuild of a design that was already failing.
04 · How badly is the CMS hurting your team?
Rescue work can fix almost everything except a CMS that genuinely fights its editors. If your team tells you they're "scared to touch the site", that's the CMS, and no amount of rescue work makes that better unless you replace it. Replacing the CMS mid-rescue is usually more expensive than a full rebuild on a CMS that fits.
Watch an editor try to update a page. If they can't do it in ten minutes without asking for help, the CMS needs to go, and that usually means a rebuild.
05 · How much money has already been spent?
This is the question nobody asks because it feels rude. But it matters. If you've paid £40k for what you've got, a £3k rescue can feel like a rounding error and a £12k rebuild feels unbearable. The sunk-cost bias pushes you toward rescue every time.
The rational question is: given what you'd be paying either way from here, which delivers a better site a year from now? Sometimes that's rescue. Sometimes it's rebuild. What you already spent doesn't enter the calculation.
The most expensive decision is the one you made on the basis of what you've already paid, rather than what you're about to pay.
How we decide
On our end, the answer comes out of the audit. Five days, 100 checks, 10 scored categories. We don't decide before we see the findings; we decide because of them. And if a rescue is materially cheaper than a rebuild AND delivers the same or better outcome, we say so, even when a rebuild would pay us more.
That's the point of taking a fixed fee for the audit: it pays us to be honest about the answer before we quote the work.