Show clients what your outbound tests found.

The most urgent row, first.

When you add or import leads, each one with a company or a title opens a row in the queue, up to a capped number in each batch, and the rest stay in your pipeline. The queue sorts by priority, then by the soonest to expire. You see who to reach and why the row is there, then write or edit the message before you approve it. No model scores live rows today.

a approve · r reject · e edit · u undo

Specimen · seeded demo tenant

Every send carries a name.

Write or edit the message, then approve or reject it. The record shows who made the decision and when.

wine red marks a human decision; machine output never uses it

An example of the client ledger.

The example ledger shows each angle’s evidence state separately from its decision. Evidence may remain unresolved; continuing to collect is a valid outcome.

rungs: reply → positive reply → meeting

The client readout

It shows what was tested, what the evidence supports so far (settled or not), what was decided, and what changes next. The design-partner pilot includes a client readout, first delivered in week 6. It is ready to forward to your client, and every number in it can be recomputed. Experiment enrollment and the proof engine are built but switched off in production today, so no customer program is being graded yet.

See a specimen readout · a seeded example of three programs, each read against its own holdout

Do we have enough volume for this to work?

The pilot operating threshold is 5,000 sends a month per program, across two or more programs. That is a commercial bar, separate from the published sample requirement below. Before any invoice, we run your real volumes through the minimum-n table, which gives the rows needed to detect 1.25×, 1.5×, or 2× lifts on a 2% positive-reply base, and we put the timeline in writing on the sizing call. If honest detection would take longer than a pilot, we say so, and the Proof Readiness Audit included in every paid plan re-runs the sizing each quarter, with a written memo.

Why isn’t pricing tied to the lift you find?

If we took a percentage of the lift we report, we would earn more by reporting more lift, and you would have no reason to trust the number. The design-partner pilot is $7,500 flat for 90 days. One fee covers a book of up to fifteen programs, and it is credited in full against year one if you continue. Larger portfolios are scoped on the call. Every self-serve price is flat too, and none of them moves with the result. We charge for the audit, never for its outcome.

What if an angle does not clear the example’s gate?

The seeded example shows every row whether or not it clears its preselected gate. The ledger keeps evidence separate from the decision: what the data has established, and what the example’s preselected rule decided. At n=1,102, an angle can be retired because its band rules out the gate while the evidence remains undecided. That decision is not proof of no effect.

Why does a human approve every send?

Two reasons. Deliverability: 2026 complaint limits leave no room for unattended sending. Accountability: when a prospect gets an email, someone’s name is on that decision. The queue makes each approval take seconds (a approve, r reject), so review takes minutes a day, not meetings.

Where does our data go?

Into your tenant, and nowhere you haven’t signed for. The application keeps each tenant’s data separate, and database row-level security adds a second check on the tables that carry an isolation policy. A decision cannot be saved without its record of who made it and when, and both are written in the same tenant-scoped write. Processing runs on the subprocessors named in the DPA. Cross-client benchmarks are opt-in, aggregate-only, and never reported below a five-tenant floor. The consent clause is in the DPA from day one, not added after a book is signed.

(Exhibit B is the subprocessor list, as printed in the colophon, all under DPA: Clerk · Supabase · Stripe · SendGrid · Fly.io · Vercel · Anthropic · OpenAI · Upstash · Sentry · Grafana.)

Are you SOC 2?

Where do replies live? Do you have an inbox?

In your sending tool, where they already are. Smartlead’s Master Inbox, Instantly’s Unibox, and the equivalent in whichever platform you run already centralize every reply, and that is where your team keeps reading and answering it. RevenueOS is send-side: it routes each send to a person to approve and keeps the record of that decision. Where your sending tool reports a reply, RevenueOS records it as a typed outcome (reply, positive reply, or meeting). We read the outcome, not the inbox, and we never ask you to move a conversation to us.