How I Use AI to Automate My Email Marketing Stats (So My Team Doesn’t Have To)

Every Monday morning, something runs in the background without me thinking about it. Claude pulls every email broadcast I’ve sent in the last 30 days from Kit, matches them to records in my Airtable Content Hub, updates the open rates, click rates, and unsubscribe numbers, and creates new records for any emails I may have sent on the fly. It also looks at my active marketing campaigns and links the emails to them automatically.

That used to live on Montana’s list. Now it doesn’t.

This is a real example of what I mean when I say AI should free your team up for work that actually requires a human — not give you more things to manage.

What This Actually Does

The sync connects two tools I was already using: Kit (where I send emails) and Airtable (where I track content, campaigns, and performance data). The problem was that the data lived in two places and somebody had to move it. That somebody is now a scheduled prompt in Claude Cowork running every Monday morning.

Here’s what it does, in order. 

• It pulls all broadcasts sent in the last 30 days from Kit — subject lines, send dates, open rates, click rates, unsubscribe rates, recipient counts, and the full email body. 
• Then it looks at my existing email records in Airtable and matches each Kit broadcast to the right record based on subject line and send date. 
• If a match exists, it updates the stats. 
• If there’s no match, it creates the record from scratch using everything it already pulled from Kit.
• It then goes one layer deeper and tries to link each email to the right marketing campaign, based on whether the send date falls within that campaign’s active window. If the email content clearly references one of my core offers, it links that too. If anything is ambiguous, it flags it and leaves the field blank rather than guessing.

Why This Works When a Lot of AI Promises Don’t

The reason this holds up is because it runs against real infrastructure I already use and trust. This isn’t AI building something from scratch — it’s AI doing rule-based data work I’ve been doing manually. The judgment calls are already made, the fields are already named, the logic is already defined. Claude is just executing it for me, every week, without me having to remember to ask.

Most AI use cases that actually stick are the ones where you’ve already decided what the output should look like. The prompt isn’t asking Claude to figure out your strategy. It’s asking Claude to do a specific job you’ve described, with your data, in your structure.

How to Build This for Yourself

You don’t need my exact prompt — you need to understand the structure so you can build one that maps to your own setup. Open a Claude chat and tell it what you want to build. 

Start by describing the job. 

Tell Claude you want to build a weekly sync prompt between Kit and Airtable or [Insert Tool Name here] lol. Explain that the goal is to pull broadcast stats from Kit, match them to existing records in Airtable, update the stats fields, and create new records for any broadcasts that don’t have a match yet. 

Give it your Airtable base ID, your table ID, and the names of the fields you want it to write to. Claude will ask clarifying questions or you can front-load all of it — either way works.

Tell it how to match records. 

Be explicit: match on subject line first, use send date as a tiebreaker, and don’t match records that are more than 7 days apart. 

Tell it exactly how to format the data. 

If your open rate field in Airtable is a percentage field, Claude needs to know whether to write 41.88 or 0.4188. If it’s a number field, same question. 

Specify the format for every stat field you want updated so there’s no guessing on the write side.

Decide whether you want campaign and offer linking. (This is Optional)

If you track campaigns in a separate Airtable table, you can tell Claude to look up any campaign whose date range overlaps with each email’s send date and link it automatically. Same idea for offers — give it a list of your offers and their Airtable record IDs, and tell it to scan the subject line and body for a confident match. If it’s not confident, leave it blank. This step is optional and you can always add it later once the basic sync is running cleanly.

Add a confirmation check at the top. 

Tell Claude that before it does anything, it should confirm both MCP connectors are active and authenticated — Kit and Airtable. If either one fails, it should stop and report the error rather than proceeding with partial data. 

This saves you from a half-finished sync you don’t notice until you’re looking at your records wondering why half of them are wrong.

Ask it to report back when it’s done. Specify the format you want: how many records were updated, how many were created, what got linked, and what it skipped and why. That summary is how you catch issues without having to manually audit Airtable every week.

Once you have a prompt you’re happy with, paste it into a Claude Cowork scheduled task and set it to run weekly. Monday mornings work well if you want the previous week’s stats ready when you sit down to review.

Pro-tip: Once you have it in Claude Cowork, I would run it so that it can prompt you for permissions. You can decide if you want to always allow for scheduled runs or if you want it to prompt you each run. 

The Actual Win

Montana isn’t logging into Kit on Monday mornings to pull numbers and paste them into Airtable. She’s writing client check-in messages, handling support tickets, and doing work that better supports our clients and SI as a whole. That’s what I wanted for her, and this is what made it possible.

The email stats still get tracked. The campaigns still get linked. It just doesn’t require anyone’s attention to happen.

Talk about supportive systems 💅🏾

---

This is just one example of the kind of quick win we tackle inside a Systems Sprint. Take a peek and be nosey — see what else is possible.