Skip to main content
ChatStarter8 min read

The Mistakes That Waste Your Time

7 anti-patterns and how to fix them.

After watching hundreds of professionals use Claude, the same mistakes keep showing up. Here are the seven worst offenders and how to fix each one.

1. The Blank Slate Start

The mistake: Starting every conversation from scratch without any persistent context.

Why it wastes time: You spend 5-10 minutes re-briefing Claude on who you are, what your company does, and how you like things written. Every. Single. Time. Over a week, that's an hour of pure waste.

The fix: Set up a Project with your core context. Do it once, benefit forever. (See Projects, Memory & Skills.)

Every Monday morning
Hey Claude, I'm the VP of Marketing at Acme Corp. We're a B2B SaaS company that sells to mid-market e-commerce companies. Our tone is professional but conversational. I need you to...

2. The Vague Request

The mistake: "Write me a marketing email" or "Help me with my presentation."

Why it wastes time: Claude produces generic output. You spend three rounds of "that's not quite right" getting to something usable. Each round takes a minute or two, and the end result is still mediocre because Claude was guessing the whole time.

Vague
Write me a marketing email about our new feature.

The fix: Use the RCTFC framework from The Anatomy of a Great Prompt. Even adding just a Task and Constraints dramatically improves output.

3. The Everything Prompt

The mistake: Asking Claude to do 8 things in one message — research, analyze, draft, format, proofread, strategize, plan, and execute.

Why it wastes time: Claude tries to do everything at once and does none of it well. The output is long, unfocused, and mediocre across all dimensions. You get a 2,000-word response where every section is a B-minus.

The fix: Break complex tasks into steps. Use follow-up messages. Claude works best when each message has one clear objective.

Break it into steps

1

First message: Research and analyze

'Here's our Q3 data. Identify the three most important trends and explain why they matter.'

2

Second message: Strategize

'Based on those trends, what are our three best options for Q4? Pros and cons of each.'

3

Third message: Draft

'Let's go with option 2. Draft a one-page brief I can share with the leadership team.'

4

Fourth message: Refine

'Make the executive summary punchier and add a timeline to the recommendation section.'

Each step builds on the last. Claude uses everything from earlier in the conversation, so you're not losing context — you're focusing it.

4. Not Saying What's Wrong

The mistake: When Claude's output isn't right, saying "try again" or "that's not what I wanted" or "make it better."

Why it wastes time: Claude doesn't know what was wrong, so it guesses — and usually guesses wrong. You end up in a frustrating loop of vague feedback and random changes.

The fix: Be specific about what needs to change. Think about what a good editor would say.

Unhelpful feedback
This isn't right. Try again.

Pro Tip

A useful trick: instead of saying "make it better," ask yourself "what specifically would make this better?" and tell Claude that. The act of articulating the problem usually makes the fix obvious.

5. Ignoring the Conversation History

The mistake: Starting a new conversation for every related task instead of building on the existing one.

Why it wastes time: You lose all the context from the previous conversation. Claude has to re-learn everything. You re-explain things you already covered.

The fix: Keep related tasks in the same conversation. Claude remembers everything from earlier in the thread and builds on it.

Good reasons to start a new conversation:

  • Completely unrelated task
  • The conversation has gotten very long (100+ messages)
  • You want a fresh perspective without prior biases

Bad reasons to start a new conversation:

  • Habit
  • You think Claude "forgot" (it didn't — everything is in the context window)
  • The previous output wasn't perfect (give feedback instead)

Scenario

You asked Claude to help you write a blog post. The draft is good but the conclusion is weak. You're tempted to start a new conversation and try again.

6. Not Providing Examples

The mistake: Describing what you want instead of showing it.

Why it wastes time: Your description of "professional but conversational" might mean something very different from what Claude produces. Words like "concise," "engaging," and "high-quality" are subjective — Claude will interpret them, but not necessarily the way you would.

The fix: Paste an example. "Write something that sounds like this: [example]" is the fastest way to get the right tone and style.

Show, don't tell
Write a weekly update email to my team. Here's an example of my style from a previous email:\n\n[paste a real email you've written]\n\nMatch this tone and structure. This week's update should cover: [topics].

This works for any output type: emails, blog posts, Slack messages, reports, presentations. One good example is worth a page of description.

7. Using the Wrong Product

The mistake: Trying to do automation in Chat, or using Code for one-off writing tasks.

Why it wastes time: You fight the tool instead of leveraging it. Chat can't connect to APIs. Code is overkill for drafting an email. CoWork is unnecessary for a solo task.

The fix: Match the product to the task. (See Chat vs. CoWork vs. Code.)

Scenario

You've been using Claude Chat to manually pull your Stripe revenue numbers every week, paste them into a spreadsheet, and then ask Claude to analyze the trends. It takes about 30 minutes each time.

The Meta-Pattern

Notice something about all seven mistakes? They have the same root cause: treating Claude like a search engine instead of a person.

Search engines need keywords. People need context, examples, and specific instructions.

The underlying skill behind all of these fixes is the same: brief Claude the way you'd brief a brilliant person who's never worked with you before. Give them what they need to succeed, be specific about what you want, and give clear feedback when something isn't right.

Do that consistently, and you'll get consistently excellent results.

Quick check

You asked Claude to write a sales email and the result is too formal. What's the most effective next step?

Your Action Step

Pick the one mistake from this list that you make most often. Fix that one thing this week. Don't try to fix all seven at once — that's basically Mistake #3 (the Everything Prompt) applied to your own learning.

Most people should start with Mistake #1 (Blank Slate Start). Setting up a single Project takes 5 minutes and improves every conversation that follows.