How to Analyze Newsletter Performance With Claude and Flodesk

Last month I got a message from the partnerships team at Flodesk asking if I wanted early access to something they were about to launch. I said yes before I knew what it was.

Here's what it is: Flodesk now connects to your favorite AI tools, Claude and ChatGPT. You ask your AI a question about your email marketing in plain English, and it pulls a real answer from your actual Flodesk account. Your favorite AI, now fluent in Flodesk.

I've been a marketer for 15+ years. I've used almost every email service provider under the sun. None of them have made me stop and say "wait, this changes everything" the way this one did.

This is one of the first times an email platform has shipped this kind of AI integration for creators and small business owners. And I think it changes the way marketers will use Flodesk going forward.

Stop digging. Start asking.

For years, this is what every marketer's workflow looked like when they wanted to understand their email performance:

1. Open your ESP
2. Navigate to the reports tab
3. Click into a specific campaign
4. Read through six charts
5. Export to CSV
6. Open the CSV in a spreadsheet
7. Manually pull a story together
8. Close everything and forget half of it by the time the next campaign goes out

I'd spend 15 to 20 minutes on this dance. Every week. For one campaign. That's not analysis, that's archaeology.

Now? I open Claude. I ask, "How did my last email perform?" And in about 8 seconds I get a real answer pulled from my actual Flodesk account. Ask anything. Your email data is one prompt away.

That's the unlock.

What it does today

Right now the integration is live for email analytics. You can ask Claude or ChatGPT things like:

• How did my last email perform?
• What was my best performing email this month?
• Which links got the most clicks in my last campaign?
• Compare my last two emails side by side.
• Are my open rates trending up or down?
• Make me a dashboard with my email analytics.

The AI hits your live Flodesk account, pulls the actual numbers, and gives you back a real answer in plain language. No digging through dashboards. No exporting CSVs. No copy paste into a spreadsheet to make sense of anything.

This is Phase 1. Audience segmentation and subscriber management are on the roadmap for Phase 2, which means you'll eventually be able to ask things like "show me everyone who opened my last 3 campaigns but hasn't bought anything" and have your AI segment that list for you. Phase 2 hadn't shipped at the time of my first look, so check Flodesk's release notes before you rely on it. It's the part I'm most excited about.

My actual test (real numbers)

I wanted to push past a generic demo and use this on my own data, so I took Flodesk up on their offer of a free account. I uploaded 500 subscribers from my actual Inbox Gold newsletter list and sent a test campaign with a punchy subject line:

Then I opened Claude and asked, "How did my last email campaign perform on Flodesk?"

Here's what came back:

• 490 sends
• 20.9% open rate (101 opens)
• 6.6% click rate (32 clicks)
• 0.83% unsub rate

But the numbers weren't the surprise. The analysis was.

The strategic read I didn't ask for

I expected Claude to just hand me the numbers. Instead it walked me through what was working, what wasn't, and what to do about it. Pasted here, slightly tightened:

Your click rate is the standout. 6.6% is strong, well above the typical 2 to 3% most lists see. The people who opened were genuinely engaged and your CTA pulled. The content and offer landed.

Your open rate at 20.9% is the soft spot. It's below where it should be for an engaged list, and on a list this small (490) that's the lever with the most upside. The subject line is the suspect. That all caps hook is on brand and scroll stopping in a social feed, but in an inbox it can read as shouty or promotional and get filtered. The very thing that wins on social can hurt in email, where curiosity and personal tone usually beat volume.

The 0.83% unsub is healthy. Nobody's running for the door.

Then it suggested two next moves: test a lower key, curiosity driven subject line on the next send, and pull a micro segment of the 32 clickers to send a focused follow up. "That's where the revenue hides on a list this size."

That's the entire job of an email strategist, delivered in roughly 12 seconds, sourced from my actual send data. I've paid five figures for less actionable advice.

Then it went deeper

I followed up with: "Are my open rates trending up or down?" Here's what came back:

What it surfaced was harder to hear, and more valuable, than the first analysis:

Looking at the trend data, the May 30 campaign is the sole email in the entire period. Everything else registers as zero. Here's the real story: that May 30 send is your only broadcast in the last seven months. November through April: zero emails. One send in May. Nothing else.

So your "20.9% open rate" isn't a trend, it's a cold start single data point. And honestly, 20.9% on a list that hadn't heard from you in months is respectable. When a list goes quiet, inboxes forget you, spam filters get suspicious, and opens crater. You should've seen worse. The fact that you didn't tells me there's real underlying affinity here.

But here's the hard truth as your advisor: you're not running an email channel, you're sending the occasional flare. And that's the actual problem, not the subject line.

It then laid out a four part system to fix it:

1. A cadence you defend like a calendar appointment. Weekly or biweekly. Pick one. Even biweekly beats sporadic because the algorithm rewards predictability.
2. A list warmup before scaling sends. Don't blast 490 cold. Send to the most recently engaged segment first, prove deliverability, then widen.
3. A reengagement sequence for the silent majority. A short 2 to 3 email "are you still in" series either revives them or lets you prune dead weight so your rates reflect reality.
4. A repurposing engine so the newsletter isn't net new work. I'm already producing content across LinkedIn, Threads, and the BestSends SaaS builds. The newsletter should be the owned audience home for things I'm already saying.

I'd been calling myself the Editor in Chief of BestSends. A few minutes of conversation with Claude inside Flodesk turned me into a customer staring at his own bad habits. That's the moment I made the call to migrate the rest of my list off my current ESP and over to Flodesk for good.

Why this matters more than you think

Most ESPs are bolting "AI" onto existing dashboards as a checkbox. Subject line generators. AI rewrite buttons. Vague predictive analytics nobody trusts. Most of it is theater.

Flodesk is doing something different. Instead of asking you to learn another interface, they're meeting you inside the AI tool you're already using all day. Open Claude or ChatGPT, ask a question, get an answer about your email program. The interface IS the AI.

This is where the entire category is heading. The future of email marketing runs through the AI tools you already use every day. Whichever ESPs ship Claude and ChatGPT integrations first will set the expectation. Flodesk just planted that flag for creators and small business owners.

If you've been around long enough to remember email pre-Mailchimp, pre-mobile, pre-automation, this is one of those category shaping moments. The platform that wins "AI native ESP" over the next 12 months is going to be very hard to dislodge. (For how Flodesk stacks up against the incumbent today, read our Flodesk vs Mailchimp comparison.)

Your favorite AI is now fluent in Flodesk. That sentence will look quaint in five years. Right now it's the wedge.

A privacy-aware workflow for measuring your newsletter

Here's the repeatable version of what I did, written so you can run it every week without oversharing data or over-trusting the AI.

1. Pick one question per session. "How did my last campaign perform compared with the previous one?" works better than "analyze everything."
2. Ask for numbers first, opinions second. Get sends, opens, clicks and unsubscribes, then ask what they suggest.
3. Verify against the dashboard. Open the campaign in Flodesk and confirm the figures match before you act on them.
4. Treat opens as directional. Privacy features in some mail apps can inflate opens. Weight clicks, replies and conversions more heavily.
5. Watch sample size. On a small list, one send is a data point, not a trend. My first result was a cold start, and Claude flagged it.
6. Change one variable per send. Subject line, send time, or call to action. Not all three.
7. Log decisions outside the chat. Keep a simple sheet with the date, the change, the result, and what you'll test next.

What not to share

• Don't paste subscriber names, email addresses or exported lists into an AI chat. Ask about aggregate performance.
• Review your AI provider's data-retention and training settings before you connect business accounts.
• Disconnect the integration if you stop using it, and only grant the access you need.

Other platforms have their own AI and analytics features, but I only verified this workflow on Flodesk. To compare platforms, start with our AI email tools guide and the Flodesk review. For what to measure beyond opens, read our segmentation playbook.

How to get it

At my first look this was a waitlisted beta, with public access pending approval from Anthropic for the Claude connector and OpenAI for the ChatGPT connector. Check Flodesk for current availability. To join the early-access list: