Rebuilt in 18 days · Part 3 of 6

Blog / Search & AI visibility

Size a rebuilt site from what people search for, not from its old traffic

When you rebuild a website, its old traffic tells you how the site was run, not what is possible. How we sized a new site from search demand: 13,252 searches sorted into the owner's plan, 1,005 searches decided one by one, one page per purpose.

When a website is rebuilt, the usual way to decide what goes on the new one is to open the old site’s analytics: keep the pages that got visits, drop the ones that didn’t. It feels safe. It is also how a new site inherits the old one’s mistakes.

Old traffic measures how the site was run, not what is possible. A page with no visits may be a page nobody wants. Or it may be a page Google never found, never trusted, or showed for the wrong search. The analytics can’t tell you which.

If you lead the team: what to ask

  • Is the page list for the new site based on what people search for, or on what our old pages happened to get?
  • For every search we chose not to build a page for, is there a written reason?
  • Do we have two pages answering the same question in different words? Which one do we keep?

Why the old numbers mislead

We recently rebuilt a directory and review website that had been live for more than twenty years. Before any rebuild, we audited it from its own data. Most of the pages that earned clicks were missing from the sitemap it gave Google, and Google had never seen a large share of the addresses we checked. On that site, a page’s traffic said more about the sitemap than about demand for the page.

That is common on older sites. Rankings lost in an update, pages left out of the sitemap, links that no longer point anywhere: each one lowers traffic for reasons that have nothing to do with what people want. If you size the new site from those numbers, you rebuild the gaps.

The old traffic still has a job: it is the list of what you must not lose. Every address that earns gets a place on the new site and a forward from the old one, as we described in “Redesign without losing what ranks”. It just doesn’t decide how big the new site should be.

Start from demand, inside the owner’s plan

The owner had written his own plan for the new site: its sections, its hubs, its kinds of pages. We kept that plan as the frame. The market data went inside it, not over it.

We gathered 13,252 searches people make and sorted each into the page of the plan that should answer it. That turned a list of sections into a list of pages with a known audience: which pages need depth, which can stay short, and which searches had no page to go to yet.

Then decide the rest one search at a time

Searches that didn’t fit cleanly went into a second list: 1,005 searches, each sorted by hand with one action and a written reason:

  • Merge into a page that already serves the same purpose.
  • New page, when nothing on the site served it.
  • Tool, when the person wants a number worked out, not an article.
  • Attach to a company’s review: a search for a company by name belongs on its review, or on a new one.
  • Hold, when the page would need facts we can’t source or can’t keep current.
  • Drop, when the search doesn’t belong on the site.

Most of the rows were merged: the search was a new wording for a question the site already answered. 61 new pages came out of the sort. The held ones stayed held. A page that needs facts nobody can check is a page that will be wrong one day, and on a review site a wrong fact costs more than a missing page.

The reason column matters as much as the action. Months later, when someone asks “why don’t we have a page for this?”, the answer is written down, and it can be revisited when the facts change.

One page per purpose, not one page per wording

People ask the same question in many ways. If the site builds a page for each wording, those pages compete with each other: Google usually shows one page from a site for a search, links from other sites split between the copies, and each copy is thinner than one good page would be.

So the rule was one page per purpose. Before launch we checked the whole site for it again, and 60 near-duplicate addresses were merged into the page serving the same purpose, with their searches moving to that page. We saw the same effect on another site, where cutting near-identical list pages gave each remaining page a clear reason to rank (“Less site, more traffic”). Hub and category pages need the same care, because they are where a purpose gets its home (“Auditing the pages nobody reads”).

What you can do with free tools

You don’t need paid software to size a site this way. A spreadsheet and two free Google tools are enough for a first pass:

  1. Keyword Planner (inside a Google Ads account, free to use) gives search ideas and rough monthly searches for the words of your business. Its “competition” column is about advertisers, not about how hard it is to rank.
  2. Search Console shows the searches your current site already appears for. Use it to see where Google already connects you with a subject, not as the limit of what you could win. If it is new to you, start with “What is GSC?”, and read “Impressions but no clicks” for searches you show for but don’t win.
  3. Make four columns: search, the page that answers it, the action (merge, new, tool, attach, hold, drop) and the reason.
  4. Group by purpose, not by words. Two searches that a person would be happy to see answered on the same page belong on the same page.
  5. Count the new pages last. The size of the site is the result of the sort, not a target set before it.

The full story of this rebuild, from the audit to launch in 18 days, is in the case study.

This article is drawn from a real engagement. Client details withheld; the figures describe our own work, checked against the project’s records.

Read the full case study →

Want us to look
at your site?

Tell us where traffic, revenue or your numbers stopped making sense. We will tell you what we would check first.

Privacy

Talk to an engineer

No sales theater. Tell us where your operation feels slow, repetitive or difficult. An engineer reads every message and replies by email.

Prefer to talk? Pick a 15-minute slot →

Your message goes straight to our engineers.