We’re a small startup and I’m essentially the marketing team. I’ve been leaning on AI drafts for almost everything for the past year and a half. The speed is real, the volume is real, and for a bootstrapped team it’s been necessary.
Here’s the problem I’m running into now that we’ve grown a bit and our content volume has increased. The humanization step is becoming its own time sink. I can generate faster than I can review and edit. The bottleneck shifted.
I’ve tried batching the review. I’ve tried more specific prompting to reduce how much editing each piece needs. I’ve tried having a contractor do a pass before it gets to me. The contractor approach is the closest to working but introduces a consistency problem because their edits don’t always match our voice either.
I’m curious how other people running content at scale have structured this. Is there a point where you just accept that some pieces will be less polished? Is there a different way to think about where the human time actually needs to go?
The ‘accept that some pieces will be less polished’ question is one teams avoid asking explicitly but answer implicitly anyway. Worth making it explicit. Decide which content categories have a lower bar, document it, and stop spending equal time on everything. That’s not lowering standards. That’s resource allocation.
The contractor consistency problem you describe is a voice documentation problem more than a contractor problem. If your brand voice isn’t documented in enough detail to brief someone to match it, the editing bottleneck moves rather than resolves. The investment that tends to pay off at scale is building a detailed enough style guide that the editing criteria are explicit, not just felt.
The prioritization question is worth sitting with. Not all pieces need the same editorial investment. High-traffic evergreen content and anything customer-facing that drives conversion probably needs more attention than a social caption. Tiering the review process by stakes rather than applying the same standard to everything is how most teams I’ve seen manage this.
We’ve hit this exact ceiling with clients who scale content before scaling quality control. The honest answer is that editorial oversight is a cost you can delay but not eliminate. At some point the volume catches up. The businesses that handled it best built the review process before they needed it at full speed, rather than retrofitting it later.
There’s a version of this where the editing bottleneck is actually telling you something about the prompting and generation stage. If you’re consistently editing in the same ways, those patterns should feed back into how you brief the tool. Not all of the editing time is unavoidable. Some of it is fixing predictable problems that better prompts would have prevented.