AI tools are changing how creators analyze subscriber data, optimize newsletters, and automate email marketing workflows. Using the Kit MCP, creator Dan Cumberland uncovered hidden audience segments, broken automations, engagement trends, and new revenue opportunities inside his 12,000-subscriber email list, all in a single day.
When Dan Cumberland sat down to test the Kit MCP, he didn’t expect to learn anything new about his own email list. He’d been on Kit for years. He’d built his own analytics scripts. He’d even started writing his own MCP wrapper before Kit’s official one shipped.
By the end of the day, he’d found 815 dormant subscribers he hadn’t known he had, a newsletter that was secretly doubling as a booking funnel, two broken automations, and a list more engaged than he’d ever quantified.
Most of what I found in a day, I’d been sitting on for months. The questions just weren’t askable before.
This is how he did it, what he found, and what every Kit creator can learn from his approach.
The challenge: a power user running out of room
Dan runs an audience and consulting business at Dan Cumberland Labs. His consulting work centers on AI roadmapping for professional service and AEC firms (architecture, engineering, and construction) using his proprietary AI maturity model, to help firm leaders understand where they actually stand with AI and what to do next.
He’d long since outgrown what most creators do with Kit. Before the MCP, he’d built two parallel sets of workarounds to push his email operation further:
- A library of Python analytics scripts that pulled data out of Kit and computed his own leaderboards offline: subject-line patterns, send-time buckets, lead-source attribution.
- A set of browser-automation scripts that drove the Kit UI directly to handle workflows the API couldn’t reach.
He’d even started building his own MCP wrapper, a proof-of-concept to give AI tools access to his Kit data. Then Kit announced the official one.
I came in with a strong baseline and a strong incentive to test the MCP against the actual workflows it would replace.
The real test wasn’t whether the MCP could do what Dan was already doing. It was whether it could do something he couldn’t.
The strategy: five new AI workflows in a single day
Dan ran his test in ascending risk: read-only inventory first, then mutations on a single sandbox subscriber, then full workflow chains using real content from his drafts folder. Once he was confident in the foundation, he moved on to what he was really there for: AI workflows the Kit MCP could unlock that weren’t possible before.
He ran five. You can run the same ones in your own account, regardless of your list size or technical background. You don’t need coding experience or a large list to run AI workflows like these. If you can talk to ChatGPT or Claude, you can do this.
AI workflow #1: Find inactive but previously engaged subscribers
The question Dan asked his account:
“Who in my list used to be one of my most engaged subscribers and has since gone completely dark?”
Through the Kit MCP and Claude, Dan ran three queries against his subscriber base, each combining a date filter, an inactivity filter, and a tightening threshold for historical engagement. In minutes, he had three cohorts:
- 9,907 subscribers in the loosest tier: subscribed before November 2025, zero opens since March 2026, with at least three opens prior.
- 3,333 subscribers in the medium tier: same filters plus at least 10 opens prior.
- 815 subscribers in the tight tier: at least 30 opens prior, then completely dark.
That tight cohort (815 people who had once been deeply engaged and then vanished) is exactly the kind of audience worth a dedicated re-engagement campaign. Before the MCP, Dan couldn’t build it. Kit’s segment builder doesn’t support the combined “engagement count greater than N and recent activity equals zero” query.
This isn’t a faster path to an existing capability. It’s a brand new category of question I can ask my list.
Sample AI prompt to copy
“Look at my Kit account. Find subscribers who opened 30+ broadcasts before March 2026 but have zero opens since. Rank them by how engaged they used to be, and tag the top 100 as ‘loyal-then-gone’ for a re-engagement sequence.”
Why this matters
Every dormant subscriber represents email you’ve already paid to acquire and engagement you’ve already earned. A re-engagement sequence targeting just 100 of Dan’s 815 dormant subscribers, with a typical 10% reactivation rate, would bring 10 of his most loyal historical readers back into the fold. Compounded across product launches, that’s revenue you’re leaving on the table when you don’t ask this question. The Kit MCP makes asking it a 5-minute AI workflow instead of an unbuildable query.
AI workflow #2: Analyze ideal customer profile penetration
The question:
“What percentage of my 12,000-subscriber list matches my ideal customer profile?”
Dan sells consulting to creators and operators in the architecture, engineering, and construction (AEC) space. He’d wanted a hard number on how much of his list matched that profile for years, but the answer had always required a CSV export, manual classification, and roughly three-to-four hours per pass.
Using the Kit MCP, he sampled 600 active subscribers across six paginated calls, then cross-referenced three signals from each subscriber’s record: an industry custom field, title keywords (engineer, architect, structural, civil, project manager, minus false positives like software engineer), and company-name keywords.
Result: roughly 4.5% of his list (about 570 subscribers) match his AEC ideal customer profile, spanning decision-makers, managers, and senior individual contributors.
He also caught a lesson worth sharing with anyone running a similar audience analytics audit: filtering by email domain alone produced a 0.6% match, seven times lower than the real number. False positives like instructure.com matched the substring struct, while real AEC employees disproportionately used Gmail and Hotmail addresses. The takeaway: combine domain + industry + title. Domain alone undercounts.
What had been a three-to-four hour exercise is now a 10-minute one Dan can re-run any time.
Sample AI prompt to copy
“Sample 600 active subscribers from my Kit list. For each one, check the industry custom field, the job-title keyword, and the company-name keyword against my ICP profile (AEC: architecture, engineering, construction). Tell me what percentage of my list matches, broken out by decision-maker, manager, and IC.”
Why this matters
Knowing your ICP penetration tells you whether your problem is list size or list quality. If 4.5% of a 12,000-subscriber list matches Dan’s ICP, his real addressable audience is 570 people, not 12,000. That number anchors every downstream decision: how much to spend acquiring more subscribers in that segment, which forms are pulling the right people in, and which sequences should target which subsets of the list. The Kit MCP turns an annual research project into a subscriber tagging exercise you can run quarterly.
AI workflow #3: Attribute lead magnets to revenue
The question:
“Which of my lead magnets actually produce paying customers?”
Dan pulled his full purchases list through the Kit MCP, sampled high-value buyers, and cross-referenced each buyer’s custom fields to trace the path back to the original lead-magnet form that brought them in.
The attribution worked through Kit’s kit_first_form custom field. One $2,000 AI Mastery Cohort buyer, for example, traced cleanly back to Dan’s AI Agentic Automation Navigator lead magnet, the first form they ever filled out on his list.
Before the MCP, this kind of attribution pass took roughly 10 minutes per buyer through the Kit UI. Across 50 buyers in a quarterly review, that’s an eight-hour project nobody wants to do. With the MCP, the full pass takes about 30 minutes.
Sample AI prompt to copy
“Pull my last 90 days of purchases. For each buyer, look up their kit_first_form custom field to trace which lead magnet they originally opted in through. Rank my lead magnets by total revenue and average buyer LTV. Tell me which one is producing the most paying customers.”
Why this matters
Most creators have multiple lead magnets running in parallel and no way to compare them by revenue. They optimize the one they last touched, or the one with the highest opt-in rate, instead of the one that actually pays. With AI-powered attribution, you can finally see which lead magnet is producing your top-tier buyers and which is just collecting email addresses. That’s the difference between a welcome sequence you optimize on volume and one you optimize on revenue.
AI workflow #4: Map your email automation architecture
The question:
“What does my full nurture system actually look like, in one diagram, today?”
Three parallel MCP calls (sequences, forms, tags) plus a walk of the exclusion rules on each sequence, rendered as a Mermaid flowchart. Total time: 15 minutes.
The diagram revealed Dan’s architecture in a way the Kit UI never had: a single nurture opener splits four ways via mutually-exclusive role tags (executive, technical, knowledge worker, ops). Each role-specific sequence excludes the other three role tags. All four converge at a closer sequence downstream.
The map is reproducible. The next time Dan adds a sequence or changes a role tag, he can re-run the workflow and get an updated diagram in 15 minutes, version-controlled like code, shareable as text, no static screenshots to maintain.
It also surfaced a cleanup target Dan didn’t know he had: roughly 25 retired sequences from a previous brand era still marked active in Kit.
Sample AI prompt to copy
“Pull my full list of sequences, forms, and tags from Kit. For each sequence, walk the exclude_subscriber_sources field to determine the role-routing logic. Render the full nurture architecture as a Mermaid flowchart, showing how subscribers move through the system from form opt-in to closer sequence.”
Why this matters
Most creators have built their email automations over years, one piece at a time. Nobody has the full picture in their head, and the Kit UI doesn’t surface it. That gap is where orphan sequences live, where misrouted subscribers get stuck, and where silent contradictions between automations break the experience for new subscribers. The Kit MCP makes the full picture visible in a single AI workflow. You stop guessing at what’s running and start auditing.
AI workflow #5: Train AI on your best-performing subject lines
The question:
“Which of my historical subject lines actually open well, and what patterns can I generate new variants from?”
Dan pulled the 30 most recent broadcast subjects with their open rates. He hand-derived a scoring rubric from the top performers: first-person confession openers, specific dollar figures, counter-intuitive directives, time-anchor framings, insight cuts. Then he scored five generated variants of a placeholder draft headline against the rubric.
Top scorers: “I stopped selling AI tools to engineering firms” and “What $80K in recovered proposal hours looks like.” Both matched the highest-opening patterns from Dan’s actual history.
What had been 20 to 30 minutes of inconclusive subject-line brainstorming per newsletter is now a three-minute exercise. Compounded across the 50+ newsletters Dan sends per year, it adds up fast.
Sample AI prompt to copy
“Pull my 30 most recent broadcasts with their subject lines and open rates. Identify the patterns in my top 10 performers. Build a scoring rubric. Then take this draft subject line [paste] and generate 5 variants that match my historical patterns. Score each variant against the rubric and tell me which to send.”
Why this matters
A subject-line gain of even 2 percentage points on open rate compounds dramatically across 50+ newsletters per year. The Kit MCP doesn’t generate generic “best-practice” subject lines. It generates subject lines that match patterns your audience already responds to, drawn from your own historical data. That’s the difference between optimizing for a generic newsletter benchmark and optimizing for your specific list. The result is a newsletter optimization system that gets sharper every time you run it.
The results: eight things the MCP surfaced about Dan’s business
The five AI workflows above were what Dan came to test. The discoveries that fell out of them were the real story.
A 17-percentage-point engagement trend Dan hadn’t quantified. All-time average open rate on Dan’s major broadcasts: 36.76%. Last 90 days: 53.05%. His list is materially more engaged than it used to be, and until the MCP made the comparison trivial, he hadn’t put a number on it.
A newsletter that was secretly a booking funnel. Dan’s broadcast “Why I tell clients to ignore AI memory” had roughly 3,000 unique clicks each across four very different destinations (~12,000 in total). The newsletter wasn’t doing one job. It was a top-of-funnel content piece and a strategy-call booking funnel at the same time. Per-URL click data on a single broadcast made that visible for the first time.
4,142 subscribers in a pre-built warm-lead pool. That’s how many people have, at some point, clicked through to Dan’s AI Strategy Call booking link. The largest pre-built warm-lead pool in his account, visible in a single query.
A tag that should have been firing, but wasn’t. A newsletter-confirmed tag Dan created in February 2026 had zero members. Either the nurture flow that should apply it never fires, or the tag was created and never wired up.
A lead-magnet form with zero opt-ins. A specific form Dan had set up also had zero opt-ins. Either the promo never ran, or the funnel is broken upstream.
An automation graveyard. Roughly 25 retired sequences from an earlier era of Dan’s business are still marked active in Kit. Cleanup candidate.
A ranked list of his highest-converting CTAs. The MCP’s per-URL click data surfaced which calls-to-action in Dan’s multi-link newsletters actually pulled, useful for any future newsletter structure decisions.
A subject-line voice profile Dan can now keep current. The 30-broadcast subject-line corpus is now cached and ready to be refreshed on a schedule.
I didn’t ask for any of these. They fell out of the testing process.
Looking ahead
Dan’s plan for what he found:
- The 815 loyal-then-gone cohort becomes a dedicated re-engagement sequence.
- The ICP-penetration number anchors Q3 outreach and proposal strategy.
- The automation map gets re-run any time the structure changes, kept current like a piece of source code.
- The subject-line corpus is cached and refreshed monthly to keep new variants on-pattern.
- The warm-lead pool of 4,142 booking-link clickers becomes a high-intent segment for upcoming launches.
The broader plan is bigger.
I could run my entire Kit operation through Claude.
How to apply this to your business
Dan’s setup is more advanced than most. But the questions he asked apply to every Kit creator with a list big enough to lose track of. You don’t need coding experience or a 12,000-subscriber list to run AI workflows like these. If you can talk to ChatGPT or Claude, you can do this.
The five questions to start with
These five questions map to Dan’s five AI workflows. They work on any list size, from 500 subscribers to 500,000.
1. Ask your list a question you’ve never been able to answer before
Start with the question, not the tool. The most valuable use of the Kit MCP isn’t replacing things you already do. It’s running queries the segment builder couldn’t express. “My most-engaged subscribers who have gone dark.” “My highest-CTR newsletters in the last 90 days.” “My subscribers who match a specific job title pattern.” If you’ve ever wanted to know something about your list and couldn’t quite figure out how to find it, that’s the place to start.
2. Audit your funnels for the ones that aren’t firing
Every Kit account has tags that should have members but don’t, forms that should have opt-ins but don’t, and sequences from an earlier era that are still marked active. They’re invisible until you go looking. The Kit MCP makes going looking a 10-minute AI workflow.
3. Quantify the trends you’ve been feeling
If you suspect your list is more (or less) engaged than it used to be, measure it. Compare your all-time open rate to your last-90-days rate. The numbers will either confirm something you already knew or surface something you didn’t. Pair this with email deliverability best practices to keep the trend moving in the right direction.
4. Map your automations as code, not screenshots
Use the Kit MCP to generate a diagram of your sequence-and-tag architecture. Re-run it whenever you make a structural change. You’ll spot orphan tags, misrouted sequences, and silent contradictions that are easy to miss when you’re staring at the UI.
5. Train your AI assistant on your own voice, not on best practices
Dan’s subject-line scorer worked because it was trained on Dan’s 30 most-recent subjects and Dan’s actual open rates. Generic “best practices” don’t match how your audience actually reads. Your own data does.
Where to start, based on your business
Different creator businesses get value from different AI workflows first. Pick the entry point that maps to how you make money.
If you’re just getting started (under 5,000 subscribers)
Don’t worry about the scale of Dan’s analysis. The same AI workflows work at any size, and on a smaller list, they’re often more useful because individual segments matter more. Start with AI workflow #4 (map your automations) and AI workflow #2 (find inactive subscribers). Both surface the kind of issues that are easy to miss on small lists, and both take less than 15 minutes to run.
If you sell digital products or courses
Lead with AI workflow #3 (attribute lead magnets to revenue). Knowing which lead magnet actually produces buyers is the single highest-leverage question a digital product creator can ask. Pair it with AI workflow #5 (subject-line patterns) for launch optimization. The Kit MCP turns a quarterly attribution review from an eight-hour project into a 30-minute AI workflow.
If you run consulting, coaching, or a service business
Start where Dan did: AI workflow #2 (analyze ICP penetration). Knowing what percentage of your list actually matches your ideal client tells you whether your problem is list size or list quality. Dan uses this to calibrate outreach for his AI roadmapping and advisory engagements, knowing his AEC-matched audience is ~570 people tells him exactly where to focus, and what list-building has to happen before that number moves.Then run AI workflow #1 to find dormant prospects worth re-engaging. Together, these two workflows give you a re-targetable pipeline you didn’t know you had.
If you monetize through sponsorships or newsletter ads
AI workflows #1, #4, and #5 matter most. You’re optimizing for engagement metrics that affect your sponsor rate cards: open rates, click rates, list health. The Kit MCP turns engagement monitoring from an annual project into a monthly habit, and gives you the kind of email analytics data sponsors increasingly want to see.
Frequently asked questions
What is the Kit MCP?
The Kit MCP is the way to connect your favorite AI tool, like Claude, ChatGPT, or Cursor, directly to your Kit account. It lets your AI assistant analyze your email list and take action: drafting broadcasts, building sequences, tagging subscribers, and pulling reports, all from a single conversation. The Kit MCP covers Kit’s full public API.
Can AI analyze my email list?
Yes. Through the Kit MCP, AI tools like Claude and ChatGPT can analyze every part of your email list: engagement patterns, subscriber sources, content performance, deliverability signals, and more. Dan’s case study above shows what’s possible in a single day of analysis on a 12,000-subscriber list, and the same AI workflows work on lists of any size.
How can AI improve email marketing?
AI helps creators move faster on the parts of email marketing that used to require hours of manual work. With the Kit MCP, you can audit list health, segment subscribers, draft newsletters in your voice, build welcome sequences, attribute revenue to lead magnets, and surface patterns across hundreds of broadcasts. The result: insight to action in a single AI conversation.
Can ChatGPT analyze newsletter performance?
Yes. Once you connect the Kit MCP to ChatGPT, you can ask ChatGPT to rank your broadcasts by open rate, click rate, or any engagement signal. ChatGPT can also surface patterns across your top performers and recommend new subject lines or content angles based on what actually works in your account.
How do creators use AI for email segmentation?
Creators use AI to build segments that the Kit segment builder can’t express on its own. Common examples include: subscribers who used to be engaged but recently went dark, subscribers who match a specific job title or industry profile, subscribers who clicked one CTA but not another, and warm-lead segments based on engagement thresholds. The Kit MCP turns segment building into a conversation.
Can AI identify inactive subscribers?
Yes. With the Kit MCP, AI can identify inactive subscribers across multiple criteria: time since last open, engagement count trends, signup-source behavior, and historical engagement. Dan’s case study above identified 815 “loyal-then-gone” subscribers (people who had opened 30+ broadcasts then went completely dark) in a single afternoon.
Can AI optimize subject lines?
Yes. The Kit MCP lets AI analyze your historical subject lines against open rates, identify the patterns in your top performers, and generate new variants that match those patterns. Because the analysis runs on your actual data, not generic “best practices,” the results match your audience.
Can AI audit email automations?
Yes. AI can map your full sequence-and-tag architecture, surface orphan tags, identify sequences that are no longer being used, and flag tags or forms that should be firing but aren’t. Dan’s case study used this approach to surface 25 retired sequences from a previous era of his business that were still marked active.
Start asking your list better questions
The Kit MCP is available to all Kit creators starting today. Connect your AI tool of choice, whether that’s Claude, ChatGPT, Cursor, or any MCP-compatible client, and start asking the kind of questions Dan asked.
You don’t need 12,000 subscribers to find something worth knowing. You just need to ask.
