(even on the free plan)
How to stretch your free usage & know when to switch tools.
This is one of the interactive tools I build for members of the Aligned AI Lab, my monthly membership for heart-led course creators & coaches who want AI to take real marketing work off their plate. If it reached you as a freebie, you're very welcome here, & you can come & see what's inside whenever you're ready.
If you've ever sat down to do something useful in Claude & hit the "come back in a few hours" wall, this is for you.
The free plan is REALLY useful. You can do proper work on it, the kind you'd actually charge for. Most people don't know how usage actually works (because nobody tells them), so they burn through it on things that don't matter & assume they need to upgrade.
Every message has a different weight, depending on what's in it. A one-liner barely registers. A question with a 50-page PDF attached costs a lot more.
That's the bit nobody tells you, & it changes everything about how you'd use it. Tap through the tabs above as you read.
Most people think of the free plan as "X messages per day". The reality is messier, & once you see it, you stop wasting your allowance on things that don't matter.
Claude measures usage in tokens (chunks of text). A short question barely registers. Attach a 30-page PDF to your question & that one "message" eats a huge slice of your allowance.
This is why some days you get 40 messages out of the free plan & other days you get 12. It tracks the WEIGHT of what you're sending, not how many you sent.
Every time you send a new message in the same chat, Claude re-reads the whole conversation up to that point. So a chat that's already 30 messages long costs more per new message than a fresh one, even if your new message is two words.
The free plan resets on a rolling 5-hour cycle. The window starts the moment you send your first message of a fresh session, so it's tied to when YOU show up, not to any fixed clock. Once you start, you've got roughly five hours before it resets.
Lots of users notice that limits feel tighter during certain hours, & it tracks with US business hours (since that's where most users are). In UK time that's roughly 1pm to 7pm. In Eastern, 8am to 2pm. Outside of those hours, you'll usually get more out of the free plan. Weekends feel fine all round.
Tip: You can start the clock deliberately, before you need it.
Once you know the 5-hour window starts when you send your first message, you can use it strategically. Send Claude a tiny "hi" at the right moment, & you control WHEN your reset lands.
The most useful pattern: time it so your reset lands DURING your deep work, not before or after.
Say you've got a four-hour sales page session starting at 1pm. If you send "hi" at 1pm too, you've burned a message for nothing & your reset doesn't land until 6pm, when you're done. But if you send "hi" at 10am, your reset lands at 3pm, right in the middle of your session. You'll feel your allowance refill halfway through, just when you'd usually be panicking.
Costs nothing. Means the wall lands AFTER your deep work, not during it.
Pick the time you'd like your reset to land (ideally the middle of your deep work session). I'll tell you when to send the "hi" that sets it up.
The small shifts that add up. None of them require a paid plan. Most people don't know about most of them.
Most token waste happens because we're figuring out what we want WHILE we're prompting.
Tap any habit to expand it.
This is the one thing I've changed in how I use Claude that's made the biggest difference. If you only take ONE habit from this guide, take this one.
Projects let you upload your reference material & set your standing context ONCE. Every chat inside that Project gets free access to it, so you don't have to drop the same files into every new conversation.
I built my brand kit last year by re-uploading the same source files into six different chats, paying full token cost six times. If I'd known about Projects then, I'd have saved myself the equivalent of probably half a day's allowance.
A few rules that make Projects work even better:
There's a prompt at the bottom of this tab to help you set up the chat once you're working inside your Project.
Write down what you actually want in one line before you start. Most of the token waste happens because we're figuring out what we want WHILE we're prompting.
You'll see me doing this all the time (I'm on the Max plan so I can afford to ramble a bit). But on free, by the time you've landed on the right ask, you've often sent six versions of the wrong one.
Try this: "Ask me 5 clarifying questions before you start." Then answer them.
You'll get a much better first draft, & one decent build beats six rebuilds.
There's a prompt at the bottom of this tab that does this for you, ready to paste in.
Claude defaults LONG. Most people don't override it & end up with a 2,000-word essay when they wanted a paragraph.
Saying "in five bullets" or "one paragraph max" costs a fraction of what you'd otherwise pay.
When the task changes, start a new chat. Don't keep piling unrelated stuff into one rolling conversation.
Each message in a long chat costs more than the last (because Claude re-reads everything before it), so the longer the chat, the more you're paying for the same kind of work.
If Claude's reply misses the mark, don't tell it "try again". Scroll up to your previous message, edit it, & resend.
Claude generates a fresh reply from that point without stacking another layer onto the chat. You usually get a better answer too, because you're sharpening your prompt instead of negotiating after the fact.
Read what Claude produced, then write ALL your changes in one go before hitting send.
I get this wrong all the time. I'll send "oh actually change paragraph two", then notice three needs work & send another, then the tone is off & send a third. That's three full message-costs instead of one, & the chat gets bloated with my back-and-forth.
Artifacts (the document or code panel that opens on the right) cost more tokens than a normal reply. If you're just going to copy the text into Kit or Google Docs anyway, ask for it as plain text.
Save the artifacts for when you actually want a designed deliverable you'll keep.
Quick check before you paste in a 50-page PDF: do I actually need Claude to see all of it?
Sometimes yes. Often, no, & copy-pasting the relevant section would do the job. Worth asking yourself every time.
If you've got flexibility in WHEN you do your deep Claude work, do it outside the peak window (UK afternoons, Eastern mornings). Even shifting one big task per day to a quieter window can change how often you hit the wall over a week.
If your day IS locked into peak hours & you can't move things, try the "say hi" timing trick from tab 1 to make your reset land mid-session.
These cover the most useful habits in this tab at once. Copy & paste, fill in the gaps, hit send. Tap the copy button on each.
Paste this at the top of any new chat where you're about to do something substantial. It does the work of three habits at once: clarifying questions, planning first, & controlling output length.
Paste this at the top of a chat when you want Claude to keep things tight & not eat through your allowance. It operationalises three of the habits above (concise replies, no unnecessary artifacts, no padding).
These are the things that eat your usage without you noticing. Tick the ones that are true for you right now.
No judgement. I do (or did) most of these too, & they're not WRONG, they're just expensive on a free plan. If you're on a paid plan with more room, most of these don't matter as much.
Tap each one that's true for you. You'll get a score at the bottom.
Heads up before you start
You do NOT need to use four tools to make this work. Claude on its own is genuinely enough for most things.
This tab is here for two reasons. One, the moments when you're about to do something that isn't really Claude's strength (image generation being the obvious one, but also things like mass brainstorming or research synthesis).
Two, & this is the bigger one, the moments when you can do PART of your work in another tool to save your Claude allowance for the bit only Claude does well. Brainstorm in ChatGPT, polish in Claude. Synthesise research in NotebookLM, write the final piece in Claude. That kind of thing.
If you only ever use Claude, that's fine. The habits in tab 2 will get you most of the way. Treat this tab as "good to know when I need it", not "another thing to learn right now".
Claude isn't the right answer for everything. & the moment you start using the other free tools alongside it, the pressure on your Claude allowance drops dramatically.
What are you trying to do?
Save your tokens for sales pages, launch emails, content that has to sound like you. This is where Claude properly shines.
Generous free tier, fast, less precious. Use it for "give me 50 options I can react to", or for first drafts you'll polish elsewhere.
Upload PDFs, transcripts, recordings. Ask it questions. Answers stay grounded in what you uploaded. Brilliant for course research, podcast prep, going through member feedback.
Both have free image generation. Claude can't make images, so don't try.
This is the workflow I wish someone had handed me when I started building in Claude. Once you've got the habits in tab 2 down, this is the next layer.
It's especially useful when you're building something out of your own material (a skill, a brand kit, a complete launch sequence), anything where you've got reference docs you'll come back to.
The workflow I wish someone had handed me when I started building in Claude.
Upload everything Claude might need: your existing writing, your offer descriptions, your voice samples, the past work you want it to draw on.
Rules that make this stage cheaper:
Doing this upfront sets you up for cheap, fast chats for the rest of the project.
When Claude produces something, read it. Then reply with all your corrections IN ONE MESSAGE.
I get this wrong all the time. I'll send "oh actually change paragraph two", then notice three needs work & send another, then the tone is off & send a third. It's not WRONG (& on a paid plan it barely matters), but on free, that's three full message-costs instead of one.
Better: read the whole thing, write all the changes in one go, hit send once.
Other things that help here:
The moment Claude produces something you're happy with (a skill, a piece of copy, a workflow doc), download it or save it out of the chat.
If you need to revise it later, open it in a text editor (or Notion, or Google Docs) & edit it there. Don't go back to Claude & ask it to regenerate from scratch. That's the most expensive thing you can do.
Save first. Edit elsewhere. Only return to Claude if you need it to think about something new.
When you've built something with Claude (a skill, an artifact, a workflow), test it in a NEW chat. Two reasons:
If this was useful, this is the kind of thing I teach inside the Aligned AI Lab.
It's a monthly membership for heart-led course creators & coaches who want to use AI to reclaim time & brain space by taking over real marketing tasks (like pages & offers), without becoming tech experts or losing their voice.
Explore the Aligned AI Lab