Great piece as usual Ryan, we are heavily invested in rebuilding our team to integrate AI very aggressively - so the same 3 man SEO content team now creates nearly 2X the content (so we didnt have to let anyone go!) We also use Surfer - how come you dont just rely on Surfer to create the main outline (as you know it allows you to pick your competitors and crawl their pages etc) - to create a solid outline with headings/subheadings. Seems you use the Chat GPT outline first, then bring that into Surfer for the keyword/nlp/density etc stuff?
Do you run the content through an AI checker like originality? I've been seeing a TON Of false positives so re-thinking if that even makes sense or not
To add to Ryan’s answer, it depends on the AI checker you use too. Not all AI checkers are the same. I’ve gotten “written by AI” for content that I wrote and edited heavily (>70%). The best AI checker IMO is the one from Content at Scale.
The problem with waiting to do KW research until you create the outline is that you open yourself up to creating duplicate content. It’s easy to forget you wrote a piece targeting “xyz keyword” when it was a year ago, or when you get a new content writer or SEO. You’re asking a new hire to figure what’s been written already. Why not just have all of that done already and organize the large TAM keyword research document? It completely prevents this issue because all anyone has to do is pick up where the last person left off.
ive been doing this for a very, very, very long time for a lot of clients. trust me when i tell you, the way i laid out is the better way. theyre able to do much better research when they are working in the moment on that piece. if you do it your way, it can take weeks / months to revisit those topics, they are goign to waste time refreshing their memeroy and essentially doing the research over again. trust me, my way is better - ive done them both before, this way works better.
Regarding the TAM kw research, the deliverable should never have been the research, but an organized version of the research. Match kw groups to their existing pages so you know how to update them. Then organize the kw groups by search volume or topic or by value (avg cpc). If you just hand clients a data dump, ya it’s useless. TAM kw research allows you to plan out work for a very long time. It’s still the best value for the client, especially if they have a content team you’re helping.
the reason why I don't send a KW list to clients is that they think they can choose which KWs to work with, and now I have to teach them that I should decide which ones work better! it is all an unnecessary hassle
When clients express concerns about AI-generated content, how do you address their worries? Do you inform them upfront that all the content you plan to publish, edit, or update will be generated using AI tools, and what kind of reaction do they typically have? Do you utilize AI detector tools to identify AI-generated content, or do you believe such tools are ineffective?
great question! 1. we run them through our entire process so they understand the outputs and quality 2. we show them results 3. we offer them human writers instead, but we charge 2x for that they almost always self select ai content