Skip to content

Jim Christian

Thoughts on tech, tools, and life.

Menu
  • Home
  • About
  • Signal Over Noise Newsletter
  • Consulting
  • Notion Marketplace
Menu

Behind the Screens: Optimising My Newsletter With AI

Posted on August 27, 2025August 27, 2025 by Jim

How I spent an hour with Comet and Claude improving a year’s worth of newsletter strategy

I’m constantly tinkering with my systems and processes. It’s probably a character flaw at this point, but the “review, revise, release” cycle is baked into how I think about everything—from organizing the kids’ school stuff to writing my weekly newsletter.

Last week I worked with Claude to shift my newsletter timing from Friday to Wednesday morning based on my global audience distribution. This week, I decided to take a deeper look at the whole operation.

The Setup

After nearly a year of publishing Signal Over Noise weekly, I had a nagging feeling there were optimizations I was missing. Rather than just gut-checking it, I decided to systematically audit the whole thing using AI tools.

The plan was simple: use each tool for what it does best, then combine the insights.

Round 1: Comet Does the Heavy Lifting

Perplexity’s Comet is brilliant for this kind of analysis because it can work directly with what’s on screen. No uploads, no context switching—just point it at my newsletter archive and start asking questions.

I started with the obvious: “What insights can you give me about my newsletter?”

Watching Comet work through my past issues was fascinating. It systematically identified patterns I knew were there but hadn’t articulated:

  • My open rates are sitting between 45-70% (apparently that’s pretty good)
  • Almost nobody unsubscribes (which suggests I’m not annoying people)
  • Growth has been organic from about 20 to 70+ subscribers
  • The subject lines that work best use curiosity gaps and questions

Then I pushed it further: “Based on this information, give me key insights on how I can continue to grow and engage my audience—what am I doing right and what could I be doing more of?”

The analysis was solid enough that I saved it to a local file. Good thing, because it became crucial input for the next phase.

Round 2: Claude Gets Strategic

While browsing Reddit (as you do), I came across someone promoting a 7-email welcome sequence for newsletters. My immediate reaction was “that’s way too much for my audience,” but I figured I’d run it past Claude for a second opinion.

The proposed sequence was:

  • Email 1: The Warm Welcome (immediate)
  • Email 2: Your Origin Story (1-2 days)
  • Email 3: Social Proof & Community (3-4 days)
  • Email 4: Best Content/Framework (5-7 days)
  • Email 5: Objection Handling (8-10 days)
  • Email 6: Behind the Scenes (11-14 days)
  • Email 7: The Soft Pitch (15-17 days)

Here’s where Claude’s access to my full Obsidian vault through MCP became valuable. It could see a year’s worth of my newsletter evolution, our previous strategy discussions, and understand the community I’m trying to build.

Claude’s take? Essentially: “Your instinct is right. This might work for some newsletter models, but it doesn’t match your anti-barrage philosophy or community-building approach.”

But we did discover that my existing welcome email was outdated and needed refreshing. Quick win.

Round 3: System Updates

If my welcome email was stale, what about my project instructions? I use custom prompts in Claude to maintain consistency in my weekly newsletter writing, so keeping those current is pretty important.

“I’d like you to review the project instructions and look for any updates or enhancements, based on how this project has progressed and insights you’ve gleaned,” I asked.

Claude immediately spotted the obvious stuff:

  • All the “Friday newsletter” references needed changing to “Wednesday”
  • The editorial checklist question needed updating for mid-week attention spans
  • The voice guidelines could be more specific about the “anti-barrage philosophy”

But then I fed it the analysis from Comet, and something interesting happened.

The Compound Effect

When Claude combined our strategic discussion with the performance data from Comet, the recommendations became much more nuanced:

Performance context: The high open rates and low churn weren’t just nice stats—they indicated specific strategies worth doubling down on.

Community evolution: The organic growth pattern suggested I was ready to move from “newsletter writer” to “community builder” in how I approach content.

Cross-platform thinking: My content was already being adapted for LinkedIn and other platforms, so the instructions should reflect that multi-format mindset.

The updated project instructions now capture not just what I’m doing, but how it’s working and where it’s heading.

What Made This Work

Tool specialization: Comet excelled at rapid pattern recognition across my archive. Claude was better at strategic integration and understanding context.

Systematic approach: Performance analysis → strategic validation → implementation updates. Each phase built on the last.

Context preservation: Saving the Comet analysis and feeding it to Claude created compound insights neither tool would have generated alone.

MCP advantage: Claude’s access to my complete content history meant its recommendations were grounded in actual evolution, not generic best practices.

The Reality Check

Total time invested: about an hour.

Time to write this up: three times longer than doing the actual work.

That ratio tells you something about the efficiency of this approach. The analysis was comprehensive, the insights were actionable, and the implementation was immediate.

More importantly, I now have a methodology for regular strategic reviews that can keep pace with growth and evolution. Not just for newsletters—for any system that matters.

Sometimes the best productivity hack is simply knowing which questions to ask which tools.

Post navigation

← Outdoor Summer Cinema

Leave a Reply Cancel reply

Your email address will not be published. Required fields are marked *

Looking for my Prompting Frameworks? They've moved!

August 2025
M T W T F S S
 123
45678910
11121314151617
18192021222324
25262728293031
« Jul    
  • Behind the Screens: Optimising My Newsletter With AI
  • Outdoor Summer Cinema
  • Cold Brew With Lemon and Dalgona at Home: Simple Summer Coffee Experiments
  • Perplexity does video generation now
  • Thoughts on Claude
  • August 2025
  • July 2025
  • May 2025
  • April 2025
  • February 2025
  • January 2025
  • December 2024
  • June 2024
  • May 2018
  • March 2018
  • January 2018
  • December 2017
  • November 2017
  • October 2017
  • July 2017
  • May 2017
  • March 2017
  • December 2015
  • November 2014
  • October 2014
  • September 2014
  • August 2014
  • July 2014
  • June 2014
  • May 2014
  • April 2014
  • March 2014
  • February 2014
  • January 2014
  • December 2013
  • November 2013
  • April 2013
  • January 2013
  • June 2012
  • April 2012
  • March 2012
  • January 2012
  • December 2011
  • February 2011
  • November 2010
  • July 2009
  • January 2008

AI animation apple author tools automation bbc books camera gear Canon EOS M200 Claude coding content creation course creation education Elgato Prompter EOS Webcam Utility Pro Estonia Gaming garageband imovie Informatic AI ios iPad launch center pro lecture london mac mcedit minecraft newsletter os x personalization productivity raspberry pi Remote Resilience Hub restrictions scifi star wars tech technology textexpander URL schemes valencia video recording webcam software

  • Bluesky
  • Facebook
  • GitHub
  • Instagram
  • LinkedIn
  • Mail
  • Mastodon
  • Medium
  • YouTube
  • X
© 2025 Jim Christian | Powered by Minimalist Blog WordPress Theme