- AI Business Insights
- Posts
- 21 ways to skip the $100 Claude upgrade
21 ways to skip the $100 Claude upgrade
Most upgrades aren't a Claude problem
Last week I almost paid $100 to fix a $20 problem
Claude's "you've hit your limit" pop-up was sitting there. The Max plan button right next to it. $100 a month, one click away. The obvious answer.
Then I stopped. Because the real problem wasn't the plan. It was how I was using it.
Around the same time, an AI pro I follow dropped a post listing 21 ways people quietly burn through Claude tokens. Twenty-one tiny habits that turn a $20 plan into a "not enough" plan. Once they're written down, half of them are embarrassingly obvious.
The whole frame was the part that stuck with me: most upgrades aren't a Claude problem. They're a workflow problem. Fix the workflow and the $20 plan stretches further than the marketing makes it look.
Here's the full list, grouped so you can patch the leaks bleeding your account dry without trying to remember 21 separate things.
In partnership with
Blu Dot surpasses 2,000% ROAS with self-serve CTV ads
Home furniture brand Blu Dot blew up on CTV with help from Roku Ads Manager. Here’s how:
After a test campaign reached 211,000 households and achieved 1,010% ROAS, the brand went all in to promote its annual sales event. It removed age and income constraints to expand reach and shifted budget to custom audiences and retargeting, where intent was strongest.
The results speak for themselves. As Blu Dot increased their investment by 10x, ROAS jumped to 2,308% and more page-view conversions surpassed 50,000.
“For CTV campaigns, Roku has been a top performer,” said Claire Folkestad, Paid Media Strategist, Blu Dot. “Comping to our other platforms, we have seen really strong ROAS… and highly efficient CPMs, lower than any other CTV partner we've worked with.”
Using Roku Ads Manager, the campaign moved from a pilot to a permanent performance engine for the brand.
Group 1: slim your inputs
The biggest leaks are in what you send Claude, not what Claude does with it.
Stop uploading raw PDFs. One page eats around 3,000 tokens. Paste the text into a Google Doc, download as .md, and you drop to under 200 tokens for the same content.
Trim your about-me file. A 22,000-word context file is wild. Cut it to under 2,000 words. End sessions with: "Write a session-notes .md."
Don't dump 50 files just in case. Include only what this task actually needs. Quick email draft? Zero folders.
Use Projects for recurring files. Upload once. Every chat inside references it without re-burning tokens.
Group 2: edit, don't redo
Most token waste in long chats comes from stacking bad history on top of good history.
Kill the 500-word prompts. Try 29 words instead: "I want to [task] to [goal]. Ask me questions using AskUserQuestion." Let Claude pull the rest out of you.
Stop saying "redo the whole thing." Say: "Only redo section 3. Keep everything else. No commentary. Just the output." Surgical edits, not full rewrites.
Use Edit, not "no, I meant." Click Edit on your original message, fix it, regenerate. You don't stack the bad history on top of the good.
One message, multiple tasks. Instead of three separate messages, batch them: "Summarize this, list the points, suggest a headline."
Keep a prompt library. Same structure, swap the variable. You stop reinventing the wheel every time you open a chat.
Speak your prompts. Lazy text prompts like "make it better" give lazy output. Voice prompts via tools like wispr.ai pack richer context into one shot.
Are you tracking agent views on your docs?
AI agents already outnumber human visitors to your docs — now you can track them.
*Ad
Group 3: match the model to the work
The wrong model for the task is the most expensive habit on the list.
Sonnet for quick stuff. Save Opus plus Extended Thinking for deep work. Grammar checks don't need a flagship.
Don't build inside Cowork too early. Plan in Chat first. Move to Cowork only when you know exactly what you're trying to make.
Don't let Claude Code explore your whole repo. Be specific: "Build a bar chart from this CSV. Save as chart.png."
Know what Claude can't do. Images? Gemini. Real-time search? Grok. Right tool for the job.
Group 4: one topic, clean sessions
Dead context is dead tokens.
One topic per chat. New topic equals new chat, always.
Restart from a clean point. When Cowork goes sideways, click "Restart the conversation from here" on an earlier message. Don't drag the mess forward.
Summarize before chats get long. Every 15 to 20 messages, summarize, copy the brief, start a fresh session.
Group 5: smart defaults
Set once, save tokens forever.
Set Personal Preferences once. Settings, Personal Preferences, set your tone and style. Stop re-explaining yourself every chat.
Turn off search and connectors by default. Flip them on per task, not per account.
Schedule recurring reports. Use /schedule. "Every Monday at 7am, create my weekly briefing." Stop running the same job by hand.
Respect the 5-hour rolling window. Claude doesn't reset at midnight. Split your usage so you don't burn the whole limit before lunch.
The three that pay back first
Twenty-one habits is too many to install at once. Three of them move the needle harder than the rest.
Hack 1: convert PDFs to markdown. On document-heavy days, this alone can drop your token usage by 90 percent. Single biggest leak in the entire list.
Hack 11: match the model to the task. Running Opus on every prompt is the difference between a $20 month and a "why did I run out on day 14" month. Sonnet handles most of your day. Save the flagship for the work that earns it.
Hack 12: use Projects. Most people I talk to forget Projects exists. It's the feature that quietly turns recurring context into a one-time upload instead of a tax you pay on every new chat.
If you only do three of the twenty-one, do these.
How to roll this out without burning a weekend
Don't try all 21 today. Pick three that match how you actually use Claude:
Heavy PDF user? Start with hack 1.
Live in long chats? Hacks 15, 16, and 17.
Burn through the limit before lunch? Hack 21.
Rebuild the same prompt every week? Hacks 9 and 20.
Tighten those, then add three more next week. The $20 plan goes a lot further than the marketing makes it look once the leaks are patched.
My honest take
The unlock here isn't really about saving money. It's about noticing that most of the friction you blame on the tool is friction you put there yourself.
Twenty-one habit shifts. Five minutes each. Most of them are settings you've never opened or buttons you've never clicked. The cumulative effect is bigger than the upgrade you almost bought to avoid them.
The upgrade button is the lazy fix. The real fix is the workflow. Pick three, run them this week, and watch what your $20 plan can actually do.
AI Alone Can’t Run Revenue
Finance doesn’t run on “mostly right.” It runs on math.
In The Architecture Behind AI-Native Revenue Automation, Tabs’s CTO breaks down why LLMs alone aren’t enough—and what it actually takes to build audit-ready, AI-driven contract-to-cash systems for modern B2B teams.
*Ad