Phase 1: prove we can put 10,000 emails a day into the primary inbox, and produce the unit economics you have never had.
So we are working from the same picture before we talk about money.
How many sends it takes to produce one reply. You said it plainly: "We don't know that. That's the thing."
Figures above are as stated on the September 8 call and are used here as your working assumptions, not as measured results. Phase 1 replaces them with measured ones.
This is not a copy problem and it is not a data problem. It is a placement problem, and placement is a function of capacity.
"When we land in primary inbox, our reply rate is quite good." The message works when it arrives. That is the only part of this that is already solved.
Resend, SES and your own servers are transactional infrastructure. Sending cold volume through them risks the ban and, in your words, is already running into thousands of dollars.
You have 20 warmed domains. That is the whole story. Everything else downstream is a symptom of it.
Volume is not a setting you turn up. It is a quantity of warmed domains and mailboxes, and it has a price.
Domains go bad. Mailboxes get flagged. That is not a failure of the plan, it is the operating condition of cold email at volume. I provision 50% above what the daily number requires so that degrading infrastructure can be rotated out without your send rate ever dropping. Treat it as insurance you are buying on purpose, not overhead.
Deliberately narrow. One variable at a time, so the result means something.
| Audience | Work emails only, from your existing 6M. Personal inboxes are a different deliverability problem and they wait for Phase 2. |
|---|---|
| Volume | 10,000 per day, reached by ramp, not from a standing start. |
| Copy | Yours, unchanged. You have said it works. Phase 1 finds out whether it holds at this volume, and we do not confound that test by rewriting it at the same time. |
| Platform | EmailBison, on my instance, under an account I create for this engagement. |
| Replies | I handle reply management on the sending side. |
| Tracking | Built together. I own send-side data, you own trial and conversion data. Neither of us can compute cost per trial alone. |
Priced at cost of infrastructure. There is no margin built into Phase 1, by design.
| Line | Basis | Amount |
|---|---|---|
| Monthly | ||
| Sending infrastructure: mailboxes across 120 domains at 10,000/day capacity | Monthly | $1,200.00 |
| Inbox placement testing (seed testing against real providers) | Monthly | $200.00 |
| EmailBison account on my instance | Monthly | $100.00 |
| Contingency for scope and unbudgeted items | Monthly | $500.00 |
| Monthly total | $2,000.00 | |
| One time | ||
| 120 domains at $10.88 each, registered and configured | One time | $1,305.60 |
| Month 1 total | $3,305.60 | |
| Month 2 onward | $2,000.00 | |
Billed upfront by Stripe invoice at the start of each month. The domain charge is one time and does not repeat. Base capacity is $800 per month for 80 domains; the figures above carry the 50% rotation buffer described on the previous slide.
| Days | Phase | Activity | Deliverable |
|---|---|---|---|
| 0 to 14 | Build and warm | Domains registered, DNS and authentication configured, mailboxes created, warmup running. Data segmented to work emails. Copy loaded into sequences. Tracking agreed with your developer. | Infrastructure live and warming, tracking spec agreed |
| 15 to 21 | Ramp | Sending begins and climbs toward 10,000 a day. First inbox placement test. Early reply signal read. | First replies, first placement reading |
| 22 to 45 | Sustained send | 10,000 a day held. Degrading infrastructure rotated out. Replies handled. Trial data reconciled against sends weekly. | 30 days of clean data at volume |
| 45 | Read | Full read: sends per reply, sends per trial, cost per trial, and cost per paying customer once your conversion data lands. | The unit economics |
| 45 to 60 | Decide | Scale, restructure to a performance model, or stop. Infrastructure is already paid through this window, so the decision is made on data rather than on a deadline. | Phase 2 decision |
Your bar, from the call: "I want to see the 2% and I want to see a consistent reply rate." Made specific.
| Criterion | Target | Why this one |
|---|---|---|
| Human reply rate on work emails at volume | 2% or better, held for 30 days | Your stated bar. Consistency matters more than any single good week. |
| Daily send held without deliverability collapse | 10,000/day for 30 days | This is the thing that has failed before. It is pass or fail. |
| Human replies per day | ~200 at a 2% rate | Follows arithmetically from the two above. |
| Trials per day | ~40 at your stated 5 to 1 | Indicative, and dependent on your funnel, not on the send. |
| Sends per reply, sends per trial, cost per trial | Measured, with a real denominator | The number you have never had. Everything after Phase 1 is priced off it. |
One honest caveat. The 2% figure is what my other campaigns produce, and those are B2B campaigns to business buyers. This is a consumer-facing offer landing in work inboxes, which is a different population. I expect it to hold and I am willing to be measured on it, but the first 30 days is the first real measurement of that specific claim, not a repeat of a proven one.
You said the real money should be made when you make money. I agree, and I would rather structure it off measured numbers than guessed ones. Four shapes worth talking through.
Infrastructure fee drops, a fee attaches to each trial signup attributed to the campaign. Simplest to measure. Your objection: trials are not revenue.
Answers that objection directly. Needs an agreed attribution window from first email to paid, since a job seeker does not convert the same week.
The 5 to 10% you floated. Needs a written baseline, otherwise "incremental" is undefined and it goes wrong later for both of us.
Fixed fee steps with daily volume, the share steps with revenue, which is the shape you described wanting as the numbers grow.
Sketched so the direction is clear. None of it is priced here, and none of it is committed until Phase 1 returns numbers.
Scale toward 25,000 to 50,000 a day. Add personal inboxes as a separate, separately measured channel. Split active from passive seekers and treat them as different audiences.
The 50,000 to 100,000 a day you originally asked for, on infrastructure that has been proven at each step rather than jumped to. Sequence and offer testing on top of a stable base.
The enterprise arm is a genuine B2B motion and a different engine. Worth its own conversation once the consumer side is producing.