07Section 7 of 8
Connect OutFound and find the winners
We reviewed 708,000 emails for Everstage through Claude Code and OutFound. One campaign had a 3.3% reply rate and booked two meetings. Another had a 0.6% reply rate and booked nine.
| Campaign | Sent | Reply rate | Meetings |
|---|---|---|---|
| A | 58,000 | 3.3% | 2 |
| B | 46,000 | 0.6% | 9 |
A needed 29,000 emails per meeting. B needed 5,111. Reply rate points to A. Meetings point to B. Open the breakdown.
What I use OutFound for
I use OutFound to keep the campaign history and connect it to what happened for the client. Then I can ask Claude which emails actually got us meetings, signups, and customers. Here's how I use it.
I give Claude three sets of information:
| Give Claude | What I want it to work out |
|---|---|
| What we sent | Who received each email, which version they got, and when. |
| What people did | Replies, bounces, opt-outs, and clicks. |
| What happened in the business | Signups, people using the product, meetings booked and attended, new customers, and revenue. |
You can start with exports from the sending tool, booking system, product, and CRM. Ask Claude to match them and show you which results it can't confidently connect. Once that works, have it keep the information up to date automatically. Watch the campaign-data walkthrough.
If someone gets the email and books without replying, I want that meeting in the report. I also want each customer counted once.
Ask what the winning emails have in common
We look at the subject, opener, CTA, value proposition, social proof, and tone. I also want to know which roles, industries, company sizes, and offers worked.
I ask Claude to group the results by those choices and keep the actual email next to the finding. Then I can see what it means when it says an opener or offer did better.
The numbers I check
| Number | What it tells me |
|---|---|
| People who replied | Whether the message started conversations. |
| Emails per booked meeting | How much sending it took to get a booking. |
| Booked versus attended meetings | Whether those bookings became actual conversations. |
| Emails per new customer | How much sending it took to get someone to buy. |
| Total campaign cost per new customer | Data, sending, and operating cost divided by new paying customers. |
I keep the actual counts beside every rate and give campaigns the same amount of time to convert before comparing them. No customers means we got no customers. Calling that a $0 acquisition cost doesn't make sense. Missing revenue should show as missing.
Give Claude this weekly review
Review [client/campaign] using emails sent from [start] to [end]
and results available through [date]. Give each campaign the
same amount of time for people to respond or buy.
Use the sending records, exact email copy, replies, booking
records, product results, customers, revenue, and actual costs.
Check for duplicates and count each customer once.
Show results you can't match clearly as needing review.
Show sends, replies, positive replies, signups, people using
the product, meetings booked and attended, paying customers,
revenue, and cost per result. Keep these separate.
Rank the campaigns by [cost per new customer / attended meeting].
Include people who booked or signed up without replying.
Explain which offers, messages, and types of buyers worked best.
Show me three findings with the numbers behind them, what you'd
test next, and where you'd put more volume. Don't change or launch
campaigns. Bring me the recommendation first.Once I've found something worth scaling, I use 70% proven campaigns, 20% improvements, 10% new ideas. At 10,000 a day, that's 7,000 / 2,000 / 1,000.
I let Claude do the sorting and math. Someone on the team still decides what to do, writes the final copy, and owns the next test.