You're Wasting Claude's Tokens Without Knowing It: 5 Ways to Cut Usage by 90%

You're wasting Claude's tokens without knowing it.
Here's how to cut your usage 90%:
 
1. Convert files before uploading.
A PDF page costs up to 3,000 tokens. 
I paste the text into a Google Doc, download as .md. Done. 90% cheaper.
 
2. Plan in Chat. Build in Cowork.
Chat is cheap. Cowork is expensive. 
I figure out what I want in Chat first. 
Then I paste the plan into Cowork + Opus 4.6.
 
Full guide here: https://lnkd.in/eyJ-SmrF
3. Type "ask me questions" instead of a long prompt. My prompt is 30 words max: 
 
"I want to [task] to [success criteria]. Read my folder. Ask me questions using AskUserQuestion before you start."
 
4. Edit your message. Never send a follow-up.
Every "No, I meant..." gets stacked on top of the conversation. Claude re-reads all of it. The Edit button replaces the old message. Cleaner and cheaper.
 
5. Match the model to the task.
Quick question → Chat with Haiku.
Writing a report based on files → Cowork with Opus.
Building a chart from data → Code with Sonnet.

PDF token cost / markdown conversion savings — confirmed as a real, commonly cited optimization. Tokens are the hidden cost of every Claude interaction, and converting PDFs and screenshots to markdown before uploading is a standard recommended step.

"Plan in Chat, build in Cowork" — confirmed as genuine best practice. Cowork is powerful but expensive since every session reads your project folder before starting any task, so thinking through structure in Chat first and moving to Cowork only when you know what you want built is directly recommended.

AskUserQuestion + short prompts — confirmed as a real, named Cowork feature, and your exact phrasing is nearly identical to published guidance: a 15-word prompt using AskUserQuestion means clarifying questions are generated once and answers are short clicks, since clicking options costs almost nothing compared to typing paragraphs.

Edit button replacing stacked follow-ups — confirmed accurate. Using the Edit button on a previous message replaces the old exchange rather than stacking a new "No, I meant..." on top, which Claude would otherwise re-read in full.

Model-matching strategy — confirmed as standard guidance (Haiku for quick tasks, Opus for heavy reasoning/reports, Sonnet for coding).