P.S. If you want more content-and-SEO workflows like this one delivered to your inbox every week, join 3,000+ readers getting them free: https://wuki.beehiiv.com/subscribe

I Measured the Half-Life of 300 Blog Posts — Most Were Dead Within 6 Months

Eighteen months ago I started treating my blog like a lab experiment. Every Sunday I pulled rankings and click data for every post I'd ever published, logged it into a spreadsheet, and watched the numbers decay.

The headline finding: the median post on my sites loses half its search traffic in 87 days. Two out of three posts are effectively dead — zero organic clicks in a 30-day window — within six months of publishing.

This isn't a "your content is bad" story. Some of the dead posts were my best work. It's a decay story, and it turns out content expires on a fairly predictable schedule. Here's the dataset, what predicts survival, and what I'd do differently if I started over.

How I measured this

I tracked 300 posts across four small SaaS blogs (mine and two products I've built) for 18 months (Jan 2025 – Jun 2026). Every Sunday I recorded:

• Search Console clicks and impressions for ~2,000 queries
• Rank positions for a tracked keyword set
• A "last updated" date for every post
• Internal links pointing at each post

Then I bucketed everything by content type, length, update history, and link structure.

Honest limitations up front: this is one person's sites in a narrow niche (developer tooling and SaaS), so the absolute numbers are mine, not universal. What I trust is the shape of the curves — they matched across all four sites, which a single outlier wouldn't do.

The decay curve is real — and exponential

Most posts follow a classic exponential decay: a burst of traffic in weeks 1–4, then a smooth, relentless fade. Plotting it on a log scale gives you a straight line. It's stable enough that I can now predict a post's traffic six months out with uncomfortable accuracy.

Median half-life by content type:

Content type	Median half-life	Organically dead within 6 months
How-to / tutorial	61 days	74%
Listicle / roundup	54 days	82%
Opinion / essay	90 days	61%
Data / analysis	210 days	22%
Tool / resource page	320 days	12%

The pattern is brutal: the more generic the content, the faster it dies. How-tos and listicles are interchangeable with thousands of competing pages, so search engines keep swapping in fresh ones. Data posts and tool pages accumulate links and citations, which makes them harder to replace.

The one lever that actually worked: updates

The single strongest signal in the whole dataset wasn't quality, length, or backlinks. It was whether the post had been touched recently.

• Posts refreshed within the last 90 days got 3.4x the clicks of stale equivalents with similar rankings.
• Posts updated two or more times had a median half-life of 300+ days versus 61 days for never-updated posts.
• A single refresh typically bought 4–6 months of traffic before decay resumed.

I checked whether this was just Google rewarding freshness signals. It's not — the effect persisted even when I refreshed a post without changing the URL or title. What changed was behavior: a refresh usually means you fix the outdated parts, and ctr-improving snippets (better titles, structured data, new examples) follow.

One caveat: never-updated posts that were still ranking didn't need touching. Refreshing them sometimes reset them downward. The decay curve is the diagnostic — only refresh posts that are measurably losing traffic, not ones that are fine.

What actually predicts survival

Ranking my 300 posts by remaining traffic after 18 months, the survivors shared three things:

1. Internal links from other ranking pages. Correlation with survival was r ≈ 0.42 — stronger than any other factor. A post linked from two or three posts that still get traffic basically never died.
2. Specific, non-generic intent. Posts targeting "how to use X" vanished. Posts targeting a named workflow, a specific tool comparison, or a quantified outcome stayed. Narrow intent pages have no competition for the phrase — and no replacements to swap in.
3. Crawl persistence. Google crawls stale pages less. In my Search Console data, pages untouched for 12+ months saw crawl frequency drop by roughly half. Less crawling → slower re-ranking → faster death. It's a feedback loop: decay causes decrawl, decrawl causes more decay.

Backlinks mattered less than I expected (r ≈ 0.19 in this dataset). They helped posts that were already doing well — they didn't rescue decaying ones.

What I'd do differently

1. Treat content like code. Posting is a deploy, not a finish line. Every post gets a maintenance schedule: review at 90 days, refresh at 6 months, retire at 12 if it's still declining. This is the change that would have saved most of my dead posts.
2. Build a refresh queue, not a publishing pipeline. For six months I published one new post a week while older posts rotted. Re-refreshing one stale post bought more traffic than two new posts. I should have inverted the ratio.
3. Kill the zombies. I had 60+ posts getting zero clicks for months. They were wasting crawl budget and diluting internal-link equity. Merging or redirecting the survivable ones into living posts recovered a meaningful share of traffic.
4. Stop writing one-off listicles. Half-lives of 54 days are a treadmill. Everything I publish now is either a data post (long half-life, earns links) or a how-to attached to a page I intend to maintain forever.

BTW, I built a tool for this

The refresh workflow I described — flagging posts entering the death spiral, tracking half-lives, drafting updates — is now automated. It runs across my sites and watches for the exact decay patterns above. It's called nextblog.ai, and it's basically a CI/CD pipeline for content. If you've ever watched a post you were proud of quietly die in the search results, you know why I built it.

Has anyone else measured this?

I'd love to compare datasets. If you track your own content, what's the half-life you're seeing? Do your decay curves look like mine — or does the shape change in your niche? And for anyone in the HN crowd who's done the "refresh vs. republish" experiment at scale: what did the data say? I'll publish the full spreadsheet if there's enough interest in a comparison.