Orbit Media asked 1,042 content marketers what’s working. 92.4% now use AI somewhere in their blogging process, which the study calls the fastest adoption curve in the history of digital marketing. Over the same stretch, the share reporting strong results fell to 13.9%. For twelve years that number lived between 20 and 30%.
The tempting conclusion is that AI broke the blog. However, the reality is that AI use correlates with nothing in the results data, good or bad. What it tells us is that the average post now takes 3 hours and 20 minutes to produce, down from 4 hours and 10 minutes in 2022. Teams got fifty minutes back and pointed them at more posts.
OpenAI’s own engineering numbers say the same thing. Tomasz Tunguz walked through the disclosure that OpenAI researchers now log roughly 3.1 agent-workdays for every 8-hour human workday. That reads like a 3x engineer until you get to the part where more than half of the successful long-running tasks over the past six months still needed a human to step in. Tunguz describes it as one engineer supervising three shifts of machine runtime while awake for one of them.
The same survey did find two things that still move results. Posts with a human editor perform nearly twice as well as AI-only editing, and original research improves the odds of strong results by about 50%.
My recommendation is to take the fifty minutes AI saves per piece and spend it on an editor who’ll kill the weak ones. Publish six good posts instead of twelve adequate ones.
Worth Your Time
- The ChatGPT ad that never sends anyone to your site (Digiday): OpenAI’s new format opens a branded chat inside ChatGPT instead of loading your landing page, and Wayfair is testing it now under what OpenAI calls a Sponsored Agent pilot.
- Short queries are losing the auction (Search Engine Land): Google Ads impression share for 1-2 word queries fell from 42% to 24% between January 2025 and August 2026, while 7+ word queries quadrupled their conversion share from 1% to 4%. Your head-term list was built for a search behavior that’s evaporating.
- Free product is an acquisition channel, so make it pass the same test (Elena Verna): She holds giveaways to a 3-month payback threshold and notes partner-channel offers convert at 40-60% against a typical 5-10% free-to-paid baseline. Worth reading for the psychological point alone: “there is a big psychological difference between ‘try this’ and ‘buy this, but slightly cheaper.’”
- Harvey staffs about 180 former practicing lawyers across its deployments (SaaStr): Those legal engineers carry $220,000 to $320,000 OTE on a 75/25 split, which tells you Harvey books adoption as a revenue function rather than support. If your product needs a domain expert present to get used properly, that person belongs on the comp plan.
- AI can write the article, it can’t stand behind it (Nielsen Norman Group): NN/g uses AI for clarity, formatting, and critique, then keeps every publish-or-kill decision with a human and won’t name AI as an author. Steal the standard for your own content policy: a person decides whether the company is willing to put its name behind the thing.
- Stop hunting the one perfect case study (Product Marketing Alliance): Benifex CMO Sarah Reynolds argues the perfect customer story can’t exist, because sales, CS, and product each need a different one. Build a portfolio spanning roles, company profiles, and outcomes instead: you already have six half-stories that beat the one you keep failing to finish.
What did your team do with the hours AI gave back? I’m betting most of it went to volume. Tell me if you spent yours somewhere better, and on what.
Wondering where your growth engine is leaking? The Growth Gap Diagnostic is where I’d start. Or just hit reply and tell me what you’re stuck on.
P.S. Curious how AI search engines see your brand? That’s what Smird is for.