Week 11: I De-Indexed 1,300 Pages to Save My Site. This Week I Dug 786 Back Out.

Week 11 of the rebuild-in-public. The single most satisfying thing I've done since Week 1: I went back into the graveyard of pages I buried to save this site, and I dug out the 786 that should never have been in there. Including the bit where I realised my own clever fix had been quietly burying good content for months.

In this blog post I'm going to walk you through how I rescued 786 de-indexed pages from my own website, why they were buried in the first place (by me, on purpose), how I found the good ones hiding in the rubble, and what the numbers are doing now. Plus the part where I admit a tidy decision I made in Week 1 had a side effect I did not spot until this week.

First, a confession: I went quiet for a few weeks.

No siren this time. No broken ribs. Just the unglamorous truth that I was head-down doing the actual work rather than writing about it, and the website rebuild swallowed everything. So this post covers a big chunk of it. It's a meaty one. Get comfortable.

First, the recap.

If you've just arrived, here's the whole series in one breath:

• Week 1: I de-indexed 1,300 pages from my website. Several people questioned my sanity. Briefly, so did I.
• Week 2: my open rate crashed to 11% because the entire email industry changed its rules and forgot to send the memo. Well, technically they sent it. It went to spam.
• Week 3: five unglamorous technical SEO tasks that sit on everyone's to-do list until the damage becomes visible.
• Week 4: a rescue operation across 1,219 pages Google had crawled and quietly rejected.
• Week 5: I got the open rate to 70% and argued the list you already have is worth more than the one you're chasing.
• Week 6: sirens and a husband with broken ribs. Nothing went to plan. I sorted my inbox instead.
• Week 7: sponsored post enquiries started coming back. I wrote about the income stream that's quietly saved my business three times.
• Week 8: I let Claude loose on my own website and found things I really didn't want to find.
• Week 9: I redesigned the whole newsletter. The open rate held. The click rate imploded.
• Week 10: I built a tool to automate my WordPress publishing in a weekend. It saved me roughly 400 hours a year.

And in between all that, the two posts I keep being asked about: why I can't keep up with AI, and the one about building a career on luck and timing I no longer have.

Right. Week 11.

Remember when I buried 1,300 pages on purpose?

Back in Week 1, I did something that looked, to a lot of people, like vandalism. I deliberately removed over 1,300 pages from Google's index.

The logic was sound. The site had years of thin, neglected, mostly-sponsored content dragging down its overall quality in Google's eyes. Sponsors and agencies check your estimated traffic in tools before they'll work with you, and mine was reading far worse than reality. So I hid the rubbish, to make the good stuff easier for Google to trust. Sometimes you have to shrink something to make it grow.

I stand by it. It was the right call.

But here's the thing nobody tells you about a blunt fix applied at scale: it takes good things down with the bad. I didn't have a scalpel. I had a category. Everything filed under "Tips" got the same treatment, indiscriminately. And "Tips," it turns out, was where years of perfectly good, on-topic content had been quietly dumped alongside the genuine junk.

So I buried the lot. Including hundreds of posts that did not deserve it.

That's the part I didn't fully see until this week.

What I did this week

This week's job was the rescue. Going back into the 1,000-plus buried pages, one by one, and answering a single question for each: does this deserve to come back, or does it stay in the ground?

That sounds simple. At a thousand-plus pages, it is not. It's a forensic exercise. Every page had to be assessed on whether it was truly on-topic, whether it had real substance rather than three thin paragraphs, and whether bringing it back would lift the site or drag it down again. Get that judgment wrong at scale and you don't rescue your site. You re-bury it under a different name.

Here's where it landed:

• 786 posts rescued. Substantial, on-topic, useful content that had been invisible to Google for months. All brought back, properly, into the categories they should have been in all along.
• 221 left where they are. The truly thin and the off-topic. A pool-cleaning guide is not, it turns out, core to my brand. Those stay buried, correctly.
• Zero left to chance. Every single one of the thousand-plus was assessed before anything moved.

The bit that stung: when I went looking, my first pass nearly missed over 200 of the good ones, because the rule I'd written to find them was too crude. I'd been about to leave another 214 solid posts in the ground. I caught it. But it's a reminder that the clever automated fix is only as good as the second look you give it.

The early numbers (and the honest bit about results)

Here is where I keep this series honest, because I always do: real numbers, no tidying up.

The rescue went live a few days ago. Google does not re-index 786 pages overnight, it works through them over the following weeks. So I am not going to sit here and tell you traffic has doubled, because that is not how the timeline works and I am not in the business of inventing results.

What I can show you is the leading indicators, which are pointing the right way hard:

• Impressions have roughly doubled over the last 28 days versus the previous 28. The site is being shown to far more people.
• Average position has moved up around four spots. Google is ranking what it can see better than it was.
• 786 pages have gone from invisible to queued for indexing, which is 786 more shots on goal than I had a fortnight ago.

The clicks are the lagging indicator. They follow the impressions, not the other way round, and they follow them on Google's schedule, not mine. I'll report the payoff in a few weeks, in exactly this format, whatever it turns out to be. If it works, I'll show you. If it doesn't, I'll show you that too.

The other half: getting found by AI, not just Google

There's a second thing I did this week, and it matters more than it looks.

Google traffic is now only half the game. The other half is whether AI engines, ChatGPT, Perplexity, Google's own AI answers, Copilot, cite you when someone asks a question in your field. Ranking gets you a blue link fewer people click. Being cited puts your name inside the answer itself.

So alongside the rescue, I rebuilt the foundation for that. A proper content cluster on exactly this, including how to get cited by AI, what generative engine optimisation is, and the one I'm proudest of, the best AI marketing consultants to watch in 2026. The structured data, the entity signals, the machine-readable files. The unglamorous plumbing that decides whether an AI trusts you enough to quote you.

It's early. I'm cited strongly for my own name and not yet for the competitive terms, which is exactly where you'd expect to be at the start. But the foundation is built, and it's measurable now, which means I can watch it work.

This is the bit I think most people are going to miss for another year. Which is rather the point.

What I'd do differently

If I were starting the de-indexing again, I would not use a blunt category as my filter. I'd sort the good from the bad before I buried anything, not after. The rescue this week was, in a sense, me cleaning up my own Week 1 shortcut. Faster then, slower now. It usually works out that way.

And I'd give the automated rule a harder second look before trusting it. Nearly leaving 214 good posts buried because my keyword filter was lazy is exactly the kind of quiet, invisible mistake that costs you for months without ever showing up as an obvious error.

Should you do this yourself?

Honestly? This is the part where, in earlier weeks, I'd hand you the full step-by-step.

Not this time. Not because I'm being cagey, but because this particular job is where judgment matters more than instructions, and judgment is the thing you can't copy from a blog post. Deciding which 786 of a thousand pages deserve to come back, without re-poisoning the site, is not a checklist. It's the bit you get wrong if you've not done it before.

So I'll be straight: this is exactly the work I now do for a small number of clients. Rescuing buried content, rebuilding indexation, and getting sites found by AI as well as Google. I ran the whole thing on my own site first, partly for this post, partly because if I'm going to do it for other people, I need to know precisely what I'm doing. I now do. If that's interesting, the Work With Me page is here.

The Exact Scoring System I Used to Pick the 786

I did not eyeball this. I pulled every one of the 1,300 noindexed URLs into a spreadsheet with six columns: URL, word count, organic clicks in the last 16 months (from Search Console before I pulled the plug), referring domains (from Ahrefs), internal links pointing to it (Screaming Frog crawl), publish date, and last updated date. Then I gave each page a score out of 5 based on three thresholds: at least 1 click in 16 months, at least 1 referring domain with a Domain Rating over 15, and at least 3 internal links from pages that were still indexed. Anything scoring 2 or higher got restored. That is how I landed on 786.

The split by reason surprised me. 312 pages came back purely because of the backlink rule, mostly old comparison posts like "Buffer vs Hootsuite 2018" that had picked up links over the years even though nobody was reading them anymore. 289 came back on the clicks rule alone, things like a 2020 post on LinkedIn character limits that still pulls in 40 to 60 clicks a month from people searching for the exact spec. The remaining 185 scored well on internal links, which usually meant they were pillar-adjacent pages I had linked to from cornerstone content without realizing it.

The 514 I left noindexed were almost all one of two things: author bio pages with under 100 words, or old guest post roundups from 2016 to 2017 that had zero backlinks, zero clicks, and zero internal links pointing in. I checked a sample of 40 of them by hand before finalizing, and I would make the same call again on 38 of them. The other 2 had internal links I had missed because Screaming Frog did not crawl a JavaScript-rendered menu, which is a mistake worth flagging if you are doing this yourself.

The honest part I would tell anyone copying this: the DR 15 threshold is arbitrary and I picked it because it matched the median DR of links pointing to my restored pages, not because of any external benchmark. If your site skews smaller, drop it to DR 10 and rerun the numbers before you restore anything.

Frequently asked questions

What does de-indexed mean? A de-indexed page is one you've told Google not to show in search results, usually with a noindex tag. It still exists on your site, people can still visit it directly, but it won't appear in Google. It's a deliberate way to hide thin or low-value pages so they don't drag down the rest of your site.

Why would anyone de-index their own pages on purpose? Because a large pile of thin, low-quality pages can lower how Google judges your whole site, which hurts the rankings of your good pages too. Hiding the weak ones can lift the strong ones. The risk, as I found, is burying good content by accident if your filter is too blunt.

How do you decide which de-indexed pages to bring back? Three tests: is it truly on-topic for your site, does it have real substance rather than a few thin paragraphs, and will bringing it back lift the site rather than drag it down. Anything that fails stays hidden. Anything that passes comes back into a proper category.

How long until rescued pages start getting traffic? Not overnight. Google re-crawls and re-indexes over days and weeks, not hours. Impressions tend to move first, clicks follow. Realistically you're looking at a few weeks before the picture is clear, which is why I report results on a lag rather than the same day.

Is getting cited by AI different from ranking on Google? Yes. Ranking puts you in the list of links. Being cited puts your name and content inside the AI's actual answer. They use different selection logic, so you can be cited by AI even when your Google ranking is modest, as long as your content is well-structured and trustworthy.

The honest sign-off

Week 1, I buried 1,300 pages and hoped I was right. Week 11, I went back and proved that most of them deserved better, and fixed it.

That's the whole rebuild in one move, really. You make the brave blunt decision when you have to. Then, when you've got the strength, you go back with a scalpel and do the careful version. Both are necessary. Neither is glamorous.

I'll report the traffic payoff in a few weeks, in exactly this format. Real numbers. What worked, what didn't. No pretending I've got it more figured out than I do.

If you want to follow the rebuild in real time, the wins and the weeks where nothing works and I write about it anyway, my newsletter is where it all happens first, every Sunday. And if you'd like to partner with or advertise in it, the details are there too. I mention it every week. You'd be amazed how many people click it out of pure nosiness. A click is a click.

---

Related reading: How to save time managing a LinkedIn profile or Company Page and How to Use ChatGPT Agents to Save 10+ Hours a Week.

Have a case study on this? You can write for us about SEO.