We read every customer email your shop received over the last two years. Not a sample. Not a summary. Every single thread, analyzed by AI, one by one.
Each thread was read and classified individually by AI. Not keyword matching. Not filtering by subject line. Every email understood in context, the way a human would read it.
Search for "wholesale" and call it wholesale. Misses context, misses intent, miscategorizes constantly.
Would have told you wholesale was your #1 category. It's not even close.
Each thread read in full. Intent identified. Sentiment measured. Value estimated. Questions extracted. Automation potential scored.
Found 19 distinct categories and 2,018 real customer questions.
19 categories emerged. The top 5 account for 64% of all email. Here's the full breakdown:
551 threads. This is the heart of your business showing up in your inbox, and it's the category that benefits most from faster, more consistent responses.
Custom hat orders are your #1 category. But "custom hat" covers a lot of ground. Here's what people are actually asking when they reach out:
This isn't a one-and-done reply. Each custom hat order is a multi-message conversation, options, sizing, pricing, production updates, shipping. 20% of threads go 11+ messages deep. That's roughly 2 hours every week just on custom hat emails.
Options, sizing, pricing, and "ready to order" add up to 66% of custom hat emails. These are people trying to spend $300–$1,889. Every hour they wait is a risk to the sale.
95% are high-value ($300+). 15.6% are Yellowstone-driven.
With 7.7 messages per thread, AI isn't just answering the first email. It drafts the follow-up on sizing, the production timeline update, the shipping confirmation. Every message in the chain gets a head start.
Even emails that need hat-making expertise, AI handles the 80% that's process, pricing, and options. Your team adds the expert judgment.
Options and Sizing make up 51% of all custom hat emails. These are the highest-leverage categories for automation because the answers follow a pattern. Here's what the AI does with each one:
At 7.7 messages each, that's roughly 2,150 individual messages where AI provides a draft. This is the fastest way to take the load off your team while keeping the quality and warmth customers expect.
We extracted 2,018 real questions from these threads. Here's what people are actually asking:
Every quote above is a real email from a real customer, pulled directly from your shop's inbox.
Yellowstone isn't just a pop culture reference for your customers. It's a buying trigger. Nearly 1 in 10 emails references the show by name.
These customers arrive knowing what they want and ready to spend $300–$1,889. Every hour of delay risks the sale.
Over half of all email threads carry positive sentiment. Only 5% show frustration or negativity. People genuinely like doing business with you.
Out of 2,277 threads, only 16 were genuine complaints. That's a 0.7% complaint rate, which is excellent. But each one tells a story worth hearing:
The pattern: complaints cluster around custom hat quality, wrong items shipped, and appointment coordination. Faster triage catches these before they escalate.
Your customers love you, but they're waiting too long to hear back. More than half don't get a response within 4 hours.
The median is better than the average, meaning some emails get answered fast, but a long tail drags the number up. Those slow responses are where customers lose patience, especially on $300+ custom orders.
Emails don't arrive evenly. Here's when your inbox is busiest:
| Mon | Tue | Wed | Thu | Fri | Sat | Sun |
|---|
The hot zone. Nearly a third of all weekly email lands here. Monday morning is the single busiest slot as weekend backlog pours in.
Customers are still emailing (286 threads on weekends) but nobody's on the other end. These sit until Monday morning, joining the pile.
When nobody's responding, wait times explode. Here's the median response time by hour. Watch what happens at 3 PM:
Emails arriving before 2 PM get a median response of 1.7–2.8 hours. After 3 PM, that jumps to 16–18 hours. Nobody's on the other end.
Saturday after 1 PM through all day Sunday. Median wait: 26 hours.
Better, but still dragged up by the afternoon cliff. Morning emails (8–9 AM) get answered in under 2 hours.
The team handles mornings well. The gap is afternoons, evenings, and weekends. That's exactly where AI fills in: drafting responses 24/7 so your team has answers ready when they sit down, not a backlog to write from scratch.
AI touches every email that comes in, classifying, routing, and prioritizing before anyone sees it. Your team sends about 56 replies per week. Here's what the daily workload looks like at full speed:
AI nails it. Read, send. ~1 min each.
AI drafts 80%. Add hat knowledge. ~4 min each.
Complex builds, negotiations. ~8 min each.
Complaints, vintage assessments, one-of-a-kind. Rare.
3 min reviewing quick drafts + 16 min adding expertise to AI drafts + 28 min writing custom responses. Compare that to writing all 11 replies from scratch at 8 min each: ~90 minutes.
"Draft-only" means AI writes the response, but a human always reviews and sends. No email goes out without a person pressing send.
AI drafts everything it can. These are the ones where a human needs to take the wheel. They're specific, they're rare, and they're the emails that should have a person behind them:
Sizing questions, option lists, pricing ranges, shipping details, appointment scheduling, order status updates, reshaping process explanations, production timelines, follow-ups. AI handles them all. Your team just reviews and sends.
The system learns from every edit your team makes. As it absorbs your hat knowledge, sizing nuances, and customer patterns, the green bar grows and the gray shrinks. Here's the trajectory:
AI has learned standard sizing, common options, typical pricing. Half of yesterday's "add expertise" becomes today's "just send."
AI handles 90% of hat expertise. Knows your sizing by model, every material option, crown shapes, brim styles, production timelines. Only truly unique requests need scratch writing.
7.5 of those are read-and-send. 2 need a quick expertise add. Only 1 per day is written from scratch. The rest of the day goes back to the front desk, the customers walking in, the phone ringing.
The AI won't sound like a robot. We extracted your team's voice from 2,277 real shop emails and built rules around it:
"Thank you so much for reaching out! We can most definitely help you with that."
"Good Morning Sarah! I am happy to help!"
"Please let me know if you have any questions!"
Warm. Personal. First names. Exclamation points. Genuine enthusiasm.
"Dear valued customer, we appreciate your business."
"Per our policy, please be advised that..."
"Thank you for contacting Greeley Hat Works support."
No corporate speak. No generic signatures. No "valued customer."
The analysis is done and the rules are written. Here's what we need from your team to turn this into a working system:
Once this system is running, the AI learns from every edit your team makes, getting sharper with every email that flows through it.