Last week Anthropic posted ‘good news’ on X.
Sounds great. My limits will go up from September 14th. “Until then, the current 50% increase will be in place.”
Wait, what current increase?
If you were early enough to the post, you either had to do the math yourself, or scroll down to see what’s going on.
If you were late, you were slapped with a community note from the get-go:
This means Anthropic was left in the worst position possible. For group A, they were giving bad news twice. Once by accident, once on purpose.
For the group B, they were straight up liars.
Funnily enough, this could have been all avoided easily. The hack is to name the bad news before someone else does.
“We’re reducing current Claude Code limits by 17% from September 14. Here’s why, and here’s what we’re building to make it worth it.”
That’s it. The backlash would have been a fraction of what it became.
When you name the bad news yourself, you are the one in control of the frame. When you bury it and someone else finds it, you lose the frame entirely and look like you were hiding something.
Every Claude user understands he’s getting the subscription credits at the fraction of the price.
“Claude subscriptions are roughly 20 to 100 times cheaper for regular, heavy usage than paying per call or token through the standard API.”
Remember - your users are not idiots. No one likes to get treated as one.
At least they thanked us after all that.
See you next time,
— Oliver
P.S. ✨ Hope you enjoyed this free post. Each week I share multiple proven growth concepts that will make your life much easier. I’m locking in, so there’s going to be 3 short form growth posts (similar to this one) and a longer one, every week.
How I can help
👨💻 Software
Need a top tech talent for your next product? Are you looking for your next gig? I’ve got you sorted.
📢 Advertising
Advertise in my newsletter to get in front of 15,828 founders.
🤿 Get more deep dives
Upgrade to a premium subscription to get an extra deep dive a week.
They could have done two more things to help themselves out:
- post the limit reduction news *after* the Fable 5.1 announcement so that it gets partially buried in launch hype, and it lands on top of a more efficient model story
- announce a date for a Luna/GLM-5.3-Flash/Qwen3.8-Flash-Next comptetive Haiku 5.1 model to help efficiency on the cheap volume side (another efficiency story)