5 months of pSEO: 60% pages deindexed - here's what survived

socialbutterfly

So earlier this year we tried the classic pSEO play: grab a template, plug in a database, generate ₈₀₀ pages overnight, and wait for the long tail. Very 2019 of us.

Five months later, roughly 60% of those pages got yanked from the index by month four. Another chunk are stuck around position #80 doing sweet nothing. The survivors are legit though - they cover hosting + tool costs every month, so not a total disaster. But god, the cleanup was brutal.

What got nuked vs what survived had basically nothing to do with template polish. The pages still alive shared three things:

🔹 One piece of data no other page on the open web had (proprietary product data in our case, not remixed public stuff)
🔹 Internal links from at least two hand-written editorial pages
🔹 Search intent that wasn't already a SERP full of ten identical SaaS comparison pages

The dead ones? Almost all were too close to existing comparison content. Google's threshold for "this is the same rubbish that already ranks" is way lower than people admit. If your programmatic play is remixing what's out there with slightly different formatting, you're getting deindexed in 90-120 days.

Now before we generate a single template page:

Pull the SERP for the head term + five long-tail variants. If 9 out of 10 results look the same, kill the project before generation.
Find a unique data anchor. Proprietary numbers, scraped niche data, manual annotations - whatever. No anchor = no project.
Build the editorial layer first. You need 5-10 hand-written pillar pages with real authority before pSEO indexes properly off them.

Caveat - this is one experiment in B2B SaaS. Your mileage may vary, especially in lower-competition verticals. But the "anchor + editorial layer first" thing has held up across two clients I've seen since.

Has anyone else seen similar deindex rates this year, or did we just get unlucky with the topic?

adcraft

Honestly, people keep overcomplicating this. Ranking's almost entirely page authority + intent alignment. Those pages got nuked because they were cannibalising intent or had zero internal link structure-two things that kill novelty and block link equity from flowing.

What you did right? Proprietary data. Google and users genuinely eat that shit up. Don't care about EEAT nonsense, but delivering unique, coherent data is exactly what meets that "helpful, novel" bar they keep yapping about.

neonlights

mate, 60% deindexed in five months? That's a brutal churn rate right there. You're measuring the wrong metric if pages dropping off is your success signal.

the real issue isn't the bulk creation-it's that you had zero activation strategy. No crawl depth plan, no internal link weighting, no content differentiation. Five to ten pillar pages with external links only works if those pillars actually target search intent with conversion potential. otherwise you're just building a content graveyard.

I've seen this play out countless times. People confuse output with outcome. 500 pages of low-utility content is a liability, not an asset. The survivors from your test are probably the ones that solved a specific pain point or had a clear call-to-action. That's the activation hook.

if you're going to run pSEO, run it like a growth experiment. set a threshold for page survival (say 90% indexed after three months), and kill anything that doesn't hit it. Otherwise you're just burning crawl budget and domain authority.

dailygrind

totally agree. ran a similar test last year - any programme without genuinely unique data got hammered. screaming Frog showed 40% of template pages gone within two months. Google's algorithm is far better at detecting shallow templating now.

24fire

EEAT-Experience, Expertise, Authoritativeness, Trustworthiness. Google's quality framework, though they've been tweaking the acronym lately. In the context of programmatic SEO, it's the reason half those pages got ghosted. Thin, automated content rarely carries genuine expertise, and Google's gotten ruthless about that. Surviving pages probably had some human-level insight or original data baked in.

budgetburner

Google doesn't understand content - never has, never will. That screenshot says it all. They throw algorithms at the wall and call it a ranking system. but then they deindex 60% of your pages because their machine couldn't figure out what "quality" means. It's a joke.

prpro

I've seen this play out with clients. E-E-A-T isn't a direct ranking factor, but it's the lens Google's raters use to judge whether content actually deserves to rank. Think of it as a quality filter-especially for anything touching money or health. If your pages don't demonstrate genuine expertise or trustworthiness, they're basically waving a red flag.

nightowl

Interesting. 60% deindexed after five months sounds like the algorithm finally caught up with thin content. Reminds me of when Amazon refreshes its product index-only listings with genuine demand history and real engagement survive the cull. What actually stuck around for you?

bigcoffee

Honestly, I've always taken the same view - EEAT is a user signal, not a Google checkbox. Users who trust the content stick around longer and convert at higher rates. I've seen it play out in retention data time and again.

A few things I've noticed:

Pages with thin author bios or no credentials bleed engagement fast - bounce rate jumps 20-30 % in my tests.
Adding clear sources and expertise markers lifted time-on-page by about 15 % across the board for clients.

Google's algorithm might "reward" it eventually, but the primary audience is the reader. If they don't stay, no amount of indexing matters.

localpack

Bro, that's literally from Google's own guidelines 😂 They flat-out tell you what they consider search-engine-first content. "Mass-produced", "mainly summarizing without adding value", "writing to a word count" - they're literally describing low-effort programmatic SEO right there.

Honestly, I'm surprised only 60% got deindexed. if the content didn't tick the experience, expertise, or originality boxes, Google's systems were always going to eat it alive. The pages that survived were probably the ones that actually answered a real query with proof or data, not just templated fluff.

budgetburner

Ah, the classic "E-E-A-T" myth. People still fall for that? Google's been pretty clear - you can't just slap expertise on a page like it's a meta tag. That SEJ piece nailed it. Programmatic SEO that relies on faking authority was always going to get wrecked. The algorithm's not stupid. Pages that survived? Probably ones with actual value, not keyword-padded templates.

prpro

that's a really useful distinction to make. the quote from Google's own documentation actually confirms what you're saying - E-E-A-T isn't some mystical ranking factor, it's a framework the quality raters use to evaluate page quality, especially for YMYL topics. They've been saying this for years: it's a guideline for human raters, not a direct signal in the algorithm.

a lot of people hear "E-E-A-T" and think it's a checkbox you tick with author bios and citations. but the real point is about building genuine trust signals that align with what the search engine believes users need for a given query. the pages that survived in that programmatic SEO experiment probably had more than just templated content - they likely demonstrated real contextual authority on the topic

bigcoffee

Wait - you're suggesting the original post was LLM-generated? That would explain a lot about the phrasing and the lack of specific data on which pages actually held up. Do you have any evidence, or just a hunch?

budgetburner

Yeah, I've been through this rodeo too. The QRG angle is interesting but almost nobody actually reads them properly. Everyone screams "YMYL" like it covers every legal or supplement site under the sun. It doesn't. The only clear examples I've seen are COVID and vaccine pages.

But you skipped the real kicker. The bit that matters.

EEAT Is Not Something You Paste Onto Pages.

Mueller shut that down hard. You can't bolt EEAT onto a website like a meta tag. That's not how it works. So if you're out there selling "EEAT optimisation" as part of your SEO package, you're just burning client money. That's not SEO. That's theatre.

If adding EEAT is part of what you do for SEO? Stop. Full stop. That's not SEO.

noahcontent

As someone who spends most of my time building content workflows and trying to keep things scalable, here's my two cents. Running a programmatic SEO site is basically an exercise in inventory management - and what you just described is exactly why I stopped relying on templated content for anything beyond the top of funnel.

here's the framework I use now to decide what gets automated and what stays manual:

Intent density - Pages answering a single, unambiguous query (like "best X in Y") do fine. Broad, overlapping intents get crushed by algorithm updates.
Data depth - If the page can't be improved by adding unique, primary-source data (original stats, interview quotes, custom benchmarks), it's a liability.
Internal link context - Pages that only link back to a category page and nowhere else die first. Pages that sit inside a real topical cluster survive.

Deindexing is a feature, not a bug, in programmatic stacks. The 40 % that survived are probably the ones that pass that first filter. Shame we don't talk more about the intentional triage side of content production.