The short version: most small business blogs are graveyards, not assets, and the fastest content win available right now isn't writing more, it's using AI to work out what to delete. I ran this on my own site, killed 40 old posts, redirected 21 more, and organic traffic went up 18% in ten weeks. This is the step by step of how I did it.
214 blog posts and I could only defend about 60 of them
I've been blogging since 2011. Fifteen years of posts about social media, content marketing, personal branding, all the stuff that made me an "influencer" back when that word still meant something useful. When I sat down this year to rebuild my business, I exported my full post list from WordPress and just stared at it. 214 posts. I didn't know what half of them said anymore.
So I did something most people never bother doing. I pulled a full Google Search Console export, three years of clicks, impressions and rankings per URL, dropped it into a spreadsheet, and asked ChatGPT to help me sort the whole thing into buckets. Not to write anything new. To tell me what was quietly dragging the rest of the site down.
Some of what turned up was almost funny. A 2015 post called "Best Twitter Tools For Small Business" that hadn't had a single click since 2019 and was still ranking, badly, for a platform that doesn't even have that name anymore. A guide to "using Vine for business marketing." An entire cluster of posts about Google Plus strategy. Dead platforms, dead advice, still sitting on my site collecting nothing but the occasional spam comment.
Why we all keep the dead posts anyway
Nobody deletes content on purpose. Every post represents an afternoon of your life, or worse, money you paid a freelancer, and deleting it feels like admitting the work was wasted. That's the sunk cost fallacy doing exactly what Richard Thaler spent his career explaining: we keep throwing good time after bad because we've already invested, not because it still makes sense. A blog post with zero traffic for three years isn't an asset sitting quietly in reserve. It's a liability that's diluting your site's authority, confusing your internal linking, and giving Google more reasons to think your site is a bit of a mess.
Deleting it doesn't undo the afternoon you spent writing it in 2016. It just stops that afternoon costing you more today.
The actual process, step by step
This took me two afternoons, roughly six hours total, and cost nothing beyond the ChatGPT subscription I already had.
- Step 1: Export everything. Full list of URLs from your CMS, plus a three-year performance export from Google Search Console (clicks, impressions, average position). If you use Ahrefs or a similar tool, pull organic traffic estimates too, that's the same instinct Ahrefs built their whole content strategy around, treating data as the starting point rather than gut feel.
- Step 2: Merge into one sheet. URL, publish year, clicks last 12 months, clicks 12 to 36 months ago, current ranking position for its main keyword.
- Step 3: Feed it to AI in chunks. I pasted 20 to 30 rows at a time into ChatGPT with the prompt: "Sort these into delete, redirect, merge, or refresh. Delete means zero traffic for 24+ months and no ranking. Redirect means dead but the URL still gets occasional links pointing at it. Merge means it overlaps heavily with another post on my list. Refresh means it's getting traffic but the information is outdated." I gave it the topic of each post from the title, not the full text.
- Step 4: Sanity check the AI's calls. It got things wrong. It flagged a post about email deliverability for deletion because traffic looked low, but that post was bringing in exactly the right five people a month, agency owners searching a very specific long-tail phrase. I overrode maybe 15% of its suggestions.
- Step 5: Execute in batches. Delete with a proper 301 redirect to the closest living page, never just a blank 404. Google treats a graveyard of 404s as its own kind of red flag.
What came out of the 214
Final tally: 40 posts deleted outright and redirected to relevant hub pages. 21 more redirected because they still had inbound links worth keeping, even with no direct traffic. Twelve thin, overlapping posts on similar topics (I had four separate posts about "social media tips for small business" from different years) got merged into four proper long-form guides. Nine posts kept their URLs but got rewritten with 2026 data, current tool names, and updated screenshots.
That left roughly 132 posts standing, down from 214, a 38% cut to the total site.
Ten weeks later, organic traffic across the whole domain was up 18%. Average position for my remaining posts improved because Google was no longer spreading crawl budget and authority across dead weight. Bounce rate on the money pages, the ones that lead to consulting enquiries, dropped by about a fifth, because visitors weren't landing on a 2016 post about Google Plus and bouncing straight back out.
Where AI earns its keep here, and where it doesn't
AI is brilliant at the sorting logic once you've given it clean data. It's terrible at judging content quality from a title alone, and it has no idea which five-click-a-month post is secretly your best lead source. Treat its output as a first draft of a decision, not the decision itself. The same rule applies if you ever bring in outside help for this kind of thing: a decent AI consultant for small business will build you this exact audit process in a first working session, but they'll still make you sit in the room while the calls get made, because you know which posts bring in the right people and they don't.
One more thing worth saying plainly: this isn't a one-off spring clean. Content decays. A post that's earning its place today won't be in three years. Brands that stay useful, the way Headspace has kept its content tightly on-message for a decade rather than letting the blog sprawl, do this kind of pruning as a habit, not an event. Same goes for the format, not just the topic. If half your traffic is landing on static "top 10" list posts, an interactive format like a calculator that engages visitors directly will often outperform and outlast a text post covering the same ground, because it doesn't date the way advice does.
The mistakes I'd tell you to avoid
- Don't delete without redirecting. You'll lose whatever residual link equity that URL had, permanently.
- Don't judge a post purely on last-12-months traffic if it's a seasonal or evergreen reference piece, check the 24 to 36 month window too.
- Don't let AI see the whole 214 in one prompt. Context windows aside, the quality of its sorting drops noticeably past about 30 rows at a time in my experience.
- Don't skip the manual override step. The posts AI wants to kill because they "look" underperforming are sometimes your quietest, best-converting content.
- Don't do this once and forget it. Put a date in your calendar, I do mine every January, and run the audit again.
Small businesses spend a lot of time and a fair bit of money asking AI to write things. Almost nobody asks it to help them get rid of things. That gap is exactly why this worked as well as it did for me, and why it'll probably work for you too, whether you run it yourself in a spreadsheet or bring in help through something like an AI implementation coach to build the process once and hand it back to you.
Frequently asked questions
How many blog posts should a small business site have before doing a content audit like this?
There's no magic number, but if your site has more than 50 posts and you couldn't list the topics of the last 20 from memory, it's due an audit. Age matters more than volume. Anything over three years old that's never been revisited is worth checking regardless of how many total posts you have.
Want AI doing the heavy lifting in your marketing?
I build the systems that handle the boring 80 percent, so you get your week back. Done properly, with the human kept in.
Will deleting old blog posts hurt my SEO?
Done, with 301 redirects to the closest relevant living page, deletion tends to help rather than hurt, because it stops Google spending crawl budget and trust signals on thin or outdated pages. The mistake that damages SEO is deleting without redirecting, which just creates a pile of 404s.
Can I use free AI tools for this or do I need something paid?
A standard ChatGPT subscription and a free Google Search Console export is enough. You don't need enterprise SEO software to run this audit, you need clean data and a clear prompt telling the AI exactly what "delete, redirect, merge, refresh" means for your site.
How often should I repeat a content audit like this?
Once a year is the minimum for most small business blogs. If you publish frequently, every six months keeps the graveyard from building back up before it starts dragging down the posts that are working.
Related reading: Your Business Now Has an AI Reputation, and You Probably Haven't Checked It and Why ChatGPT Ignores Your Small Business (And What Gets You Mentioned Instead).
If you want the full breakdown, here is everything I know about AI marketing.
Prefer to hand this over? This is write for us about social media.
Free resource: grab The Time Audit Cheat Sheet from the resource library.
