We have 400 pages and a redesign coming — how do we decide what to keep, merge, rewrite or delete?

Look at a single page and ask: if this URL vanished tonight, who would notice and complain? If you can name them, or name the search, the invoice, the email or the sales conversation that sends people there, the page survives. If you cannot, it goes.

That is the whole test. Everything below is either how to gather the three facts that let you answer it, or the handful of pages where the answer is yes even though the traffic says no.

Four verdicts, one per page, no page left blank. Keep as is means it moves into the new design with its copy untouched. Merge means it folds into another page and its URL redirects there. Rewrite means the page needs to exist but the copy is wrong. Delete means the URL goes away and redirects somewhere relevant. A page marked “maybe” is a page you will pay to redesign.

Most teams start an audit with everything on the keep list and look for reasons to remove things. Removal feels risky and keeping feels free, so almost nothing gets cut. That default is how a site reaches 400 pages in the first place, and it is also what makes the redesign expensive.

Migration cost tracks page count and template count more closely than anything else. Each page needs its content moved, checked against the new layout, and reviewed by someone on your side. A page nobody has opened in two years costs the same to move as your pricing page.

So flip the default. A page has to argue for its survival. In our projects, a mature site that has been publishing for five or more years usually has something like a third of its pages that nobody would miss, and often more once near-duplicates get merged. That is not an industry statistic, it is what the exports tend to look like when someone finally sorts them.

Deleting is not an SEO trick here. It is a cost and clarity decision. A smaller site is cheaper to redesign, faster to review, and easier for a visitor to navigate.

Export your URLs with sessions for the last 12 months, entrances from organic search, and conversions or assisted actions. That is enough. Twelve months catches anything seasonal.

Resist the urge to enrich the sheet. Word counts, readability scores, last-updated dates and a composite score out of 100 all feel like progress and none of them change a verdict. The one column worth adding is free text: why this page exists, in your own words. If you cannot fill that in, you have your answer.

You do not need a paid content audit deliverable for this. Three columns and a week of your own judgement will beat a consultant’s spreadsheet, because you are the only person who knows that the “Partners” page exists because a partner asked for it in 2021.

Sorting by pageviews and cutting the bottom half will delete pages you need. Traffic measures how many people arrive, not what the page is for. Some pages exist to be there when one specific person looks for them.

These override low traffic and stay, whatever the numbers say:

• Legal and policy pages. Terms, privacy, cookie notices, security pages. Nobody reads them until someone in procurement does.
• Pages linked from outside your site's navigation. Invoices, receipts, transactional emails, in-product help links. Deleting these breaks something you cannot see in analytics.
• Pages your sales team sends by hand. Low traffic, high value. Ask them before you cut anything that looks like a use-case or comparison page.
• Pages ranking for something narrow and commercially valuable. Ten visits a month from people searching your exact niche term is worth more than 2,000 visits to a listicle.
• Pages with real inbound links from real sites. A page with a link from an industry publication carries value that a redirect only partly preserves.

If three pages answer the same question worse than one page could, merge them. Pick the strongest URL, write one good page, redirect the other two into it.

A worked case. A software client had a page on integrations, a page on their API, and a blog post titled “How our API works”, all written eighteen months apart by different people. Each answered part of the question, none answered it fully, and prospects landing on the blog post never found the integrations list. Merged into one page, the answer became complete for the first time. Nothing about the visual design achieved that. The merge did.

This matters commercially too. Refuse to pay a per-page redesign fee for pages that are near-duplicates of each other. Merge first, then price the design. Three pages folded into one is two pages of migration work you never pay for.

Rewriting old blog posts wholesale as a launch requirement. Redirecting them costs nothing, and archiving them costs nothing. If a post from 2019 is not worth rewriting on its own merits, it is not worth rewriting because the site is changing.

A scored audit report. You need a verdict per row, not a methodology.

And the honest version of this: cutting your page count before you get quoted makes the project smaller. Glasspage prices partly by page and template count, so a client who arrives with 180 pages instead of 400 pays us less. We would rather say that than watch you pay to redesign pages nobody opens.

There is also a chance you do not need a full redesign. If the audit says most pages are fine, and the real complaints are one ugly template plus navigation nobody can use, that is a template refresh and a merge exercise. Cheaper, faster, and it fixes the thing you were actually annoyed about. Templates are worth counting separately for this reason: six templates can drive hundreds of URLs, so a site that feels broken everywhere may only have two designs that are wrong.

Every deleted URL gets a 301 redirect to the nearest genuinely relevant page. Not the homepage.

Bulk homepage redirects are the single most common reason a redesign gets blamed for a traffic drop. Someone searched for a specific answer, landed on your homepage, found nothing about it, and left. Search engines read that as a page that does not match the query, and the ranking goes with it. Sending fifty deleted blog posts to the homepage in one commit is the cheap option that costs the most.

If no relevant page exists, that page probably should have been merged rather than deleted, or the destination page needs a section added to earn the redirect.

Before deleting, archive the copy somewhere your team can reach: a folder of exported text or PDFs is fine. In month three, someone will say “we had a page about that.” Being able to open it in twenty seconds ends the conversation. Not being able to open it restarts the audit.

The finished audit is a list of URLs, each with one verdict, and it is the list your quote gets priced against. Sign it off the same way you would sign off a design.

Pages added after design sign-off are the most common source of scope arguments we see. A page arrives in week two, then three more in week three, and by launch the client feels the studio has expanded the bill while the studio feels the project changed underneath it. Both are right. The fix is a frozen list agreed before anyone opens a design file.

If a page genuinely needs adding later, add it deliberately: name it, price it, approve it in writing. The problem was never the extra page. It was the extra page nobody decided on.

You will not get every verdict right, and you do not need to. A page you deleted and should have kept is one redirect to unwind. Four hundred pages you never decided about is a redesign that costs more than it should and still feels cluttered on launch day.