Replacer Cheat Sheet - How to go from 0 to better than perfectly-your-voice content

Here's how the scoring actually works and the fastest path to 250.

Rich voice data is better than more voice data. Before you upload a bunch of content in sources, go through this process first.

How the Numbers Actually Break Down

The system adds two buckets:

Base Score (0-100 cap): Replacer evaluates 5 dimensions across 4 sub-dimensions on a 0-5 scale. Returns zero if your profile has no signal yet.

Bonus Tracks (0-850 cap): Seven separate scoring systems that stack on top of base.

The seven bonus tracks work like this:

• Edit Patterns (150 cap): 3 points per pattern + 10 points per category
• Anti-Voice (100 cap): 3 points per constraint + 10 points per category
• Context Profiles (200 cap): 15 points per format + 3 points per observation
• Feedback Rules (100 cap): 3 points per rule + 5 points per rule with notes
• Source Conversations (100 cap): First 5 conversations worth 10 points each, then 5, then 2
• Voice Consistency (100 cap): 10 points per no-edit approval from your last 10 reviews
• Voice Guide Depth (100 cap): 1 point per 50 characters, maxing at 5,000 characters

The fastest tracks per minute? Source Conversations for your first five, Context Profiles the moment you generate in a new format, and Voice Guide Depth which compounds automatically with every conversation.

The 250-Point Playbook (3 ish Hours)

Step 1: One Real 45 Minute Conversation (\~110 points)

Your first substantial conversation flips the voice signal flag from false to true, which unlocks base scoring completely. This jumps you from 0 to 60-80 points instantly.

You'll earn 10 points on Source Conversations and grow your voice guide to roughly 1,500-2,500 characters, adding 30-50 points on Guide Depth.

What to actually talk about: rotate through all five base dimensions in one sitting. Cover formality and humor for tone, signature phrases you use without thinking for vocabulary, how you structure arguments, a few contrarian opinions you hold, and your core expertise domains. Each dimension you touch meaningfully lifts base by 4-8 points.

End of Step 1: 100-130 points

Step 2: Two Shorter Conversations, 20 Minutes Each (+30 points)

Conversations 2 and 3 add 20 more points on Source Conversations. Use the in-app gap prompts here - they target whichever dimension Claude scored lowest in your first conversation.

Each conversation should add 10-15 base points and another 1,000-1,500 characters of guide depth.

End of Step 2: 150-170 points

Step 3: Generate Content in Five Formats, Then Edit Everything (+80 points)

Click "Generate" for blog post, social media post, email, pitch deck, and short-form content. Five distinct formats.

Each new format immediately adds 15 points to Context Profiles before you do anything else. That's 75 points just for clicking buttons.

Now edit every single draft visibly, even if they're close. Small rewrites populate your edit preferences. Getting 10 edits across 4 categories nets you roughly 70 points on Edit Patterns.

Hit thumbs up or down with written notes on each piece. This adds 20-25 points on Feedback Rules.

End of Step 3: 230-260 points

Step 4: Rate "Not My Style" on Voice Samples (+30 points)

Go to the voice page and reject 5-8 samples with written reasons. Each rejection adds anti-voice constraints across 2-3 categories, building 40-60 points on Anti-Voice.

End of Step 4: 250-290 points - you're past 250

What NOT to Waste Time On

Voice Consistency requires 10 no-edit approvals from your last 10 reviews. You can't game this early because it's rolling, and it directly conflicts with Step 3's "edit everything" strategy. Skip this until you're chasing 350+.

Source Conversations past number 5 drop to 5 points each, then 2 points. After five conversations, every minute is better spent on Context Profiles or editing existing content.

Trying to push base score past 85 has steep diminishing returns. Claude rarely awards perfect 5/5 scores without rich, multi-quote evidence from your conversations. Don't chase the last 15 base points when you can buy them cheaper from bonus tracks.

The Single Biggest Leverage Point

Generate-then-edit across 5+ formats. It's the only place where one click adds 15 points before you've typed a word, and the resulting edits feed three scoring tracks simultaneously: Edit Patterns, Feedback Rules, and Context Profiles.

Order of Operations Rule

Do conversations before content generation. The edit-derived points in Step 3 are worth more when there's already a solid base and voice guide for the generator to draw from. Otherwise the drafts come out too generic to edit meaningfully, and your edit preferences patterns end up noisy.

Start with the conversations. Build the foundation. Then exploit the format generation system for quick points. You can generate A/B samples in the Voice section to track the improvement over time.