Why I Reject 15,000 Guest Pitches And Accept 1,200

The inbox nobody sees

Somewhere between thirty and sixty guest post pitches land in my inbox every week, and about two of them get a yes. That is not a typo. Over the life of my site I have accepted roughly 1,200 posts and turned down more than 15,000. I screen every one of them myself, no assistant, no filter bot, because the moment you hand that judgement to someone else you start publishing things you would not put your name near.

Most weeks the maths feels harsh to the people sending pitches, and I understand why the silence is frustrating. Nobody wants to hear that their email is one of fifty landing in the same window. But the volume is the whole explanation. When fifty pitches arrive and one or two deserve a place on the site, the rejection rate is not a mood; it is arithmetic. I read the subject line, I read the first paragraph, and within about thirty seconds I already know which pile it is going in.

Three paragraphs and I know

I can spot AI-written copy within the first three paragraphs, and once I see it I stop reading with any real intention of publishing it. The tell is rarely one word choice. It is the rhythm, every sentence roughly the same length, every claim rounded off with the same tidy confidence, nothing awkward or specific anywhere in the piece. Real writing trips over itself occasionally. AI copy never does, and that smoothness is exactly what gives it away.

I am not precious about this out of principle. Four months into rebuilding my own site with AI, I tried producing blog content entirely with it and tripled my output. It read fine on a first pass, honestly, and I let myself believe the volume was a win. Six weeks later Google's helpful content update hit the site and organic traffic dropped roughly 35 percent. I spent three months clawing it back before new posts started ranking again. So when a pitch reads like that first batch of mine did, I know exactly where it ends up, and I am not doing that to my own site twice. These days AI has a place early in my process, for research and structure, never for the paragraphs that go out under a name.

One link, not two

Editorial guest posts get exactly one contextual backlink to the writer's own site or a relevant resource. That is the deal, stated plainly on the guidelines page. A second link, however naturally it is worded into the piece, changes what the post is for. One link reads as a writer pointing a reader somewhere useful. A second one usually drags the whole paragraph around it slightly off course, and that alone is often enough to tell me what I am holding.

I do not chase down every anchor text trying to prove intent. I do not need to. By the time I reach the second link the post has usually already stopped talking to the reader and started talking to a metric, and you can feel that shift happening mid-sentence if you are reading closely enough. It is a small rule on paper and it does most of the filtering work on its own.

What clears the bar

Relevance matters more than polish. The topic needs to sit inside a lane my readers are already in, not somewhere adjacent to it that a keyword tool suggested. Past that, I check the writer's LinkedIn profile against the expertise claimed in the pitch, which takes about ninety seconds and saves everyone a lot of back and forth later, and I look for an angle written for a reader with a problem rather than for a search engine.

What this leaves

Two out of fifty is not a wall so much as what is left once the obvious problems are gone. My own filter never changes from one week to the next: subject line, first paragraph, thirty seconds, then the one-link check if the piece survives that far. I still open every pitch myself, with nothing screening ahead of me, because the day I hand that over is the day something gets through that I would not have put my name near.