Most agencies describe a process and show you a dashboard at the end of month one. Here is every artifact a Miles engagement produces, in its real format, before you sign anything.
Book a callThe targeting contract. It defines who counts as qualified, which makes it the document the guarantee is enforced against.
Signal stack: an account enters a campaign only when the ICP gate plus two of these agree:
Estimated addressable universe: 2,140 accounts. At observed trigger frequency, roughly 55–80 accounts enter a qualifying signal state per month, which is the real constraint on volume, and the reason we don't promise 4,000 sends a month.
Changes to this brief reset targeting and are agreed in writing. Version history is kept.
Why it matters: "qualified meeting" is where most agency relationships break down. Ours is a signed document with an anti-ICP list, not an adjective in a proposal.
Sent the day before every booked call. You walk in knowing why they replied, so a cold meeting behaves like a warm one.
Series A · 62 employees · San Francisco · sells ML observability into enterprise data teams
Do not pitch the platform in the first ten minutes. She replied to a pipeline-math argument, not a product argument.
Why it matters: the difference between a booked meeting and a useful one is whether the person walking in knows what the prospect already told us. This is the artifact that makes the meeting count, and the reason clients stop describing our meetings as "cold."
Full funnel, including the parts that went badly. If we are behind guarantee pace, it says so at the top. You should never learn that from a renewal conversation.
| Metric | This period | Last period | Running |
|---|---|---|---|
| Prospects added | 612 | 598 | 1,842 |
| Emails delivered | 1,431 | 1,388 | 3,977 |
| Delivery rate | 98.4% | 98.9% | 98.6% |
| Replies | 51 (3.6%) | 44 (3.2%) | 128 (3.2%) |
| Positive replies | 14 (27%) | 11 (25%) | 34 (27%) |
| Meetings booked | 4 | 3 | 8 |
| Meetings held | 3 | 3 | 7 |
What worked. The "new VP of Data" trigger segment is carrying the period: 9 of 14 positive replies came from accounts where a data leader started in the last 90 days, at a 6.1% reply rate against a 3.6% blended average. We are reallocating roughly 40% of next period's volume into that segment.
What didn't. The Series-B-and-later funding segment is underperforming badly: 340 sends, 4 replies, 0 meetings. Our read is that they already have incumbent tooling and the funding trigger alone isn't enough. We're pausing it rather than iterating. The message isn't the problem, the segment is.
One miss on our side. A reply on 18 Jun sat 31 hours before it was answered, against our same-day standard. The prospect had gone quiet by then. That's on us; the alert routing has been changed.
Needed from you: two of the held meetings scored 3/5 on fit. Fifteen minutes on what made them weak would sharpen the ICP filter for the back half.
Why it matters: the misses are in the document. An agency that only reports wins is an agency you'll be surprised by in month three.
You approve copy before anything sends. Every email arrives with the argument for why it's written that way, and what we're testing.
EMAIL 1 · DAY 0 · SUBJECT A/B: "your first 90" / "data reliability at {{company}}"
Why this way: the trigger is in line one and it's verifiable, so it can't read as mail-merge. No link (step 1 links suppress placement). The peer reference is segment-level, not fabricated. CTA is time-boxed, not "book a demo."
EMAIL 3 · DAY 7 · NEW THREAD
Why this way: new thread resets the open rate. Single link, on our tracking domain. The explicit out at the end is deliberate. It raises negative-reply rate, which protects the domain and gives us cleaner segment data than silence does.
Currently testing: subject line A vs B, one variable, minimum 200 sends per arm before any verdict. Everything else is held constant. Last test, peer reference versus metric-led opener, went to the peer reference at 4.1% vs 2.8% reply.
Why it matters: you keep every sequence, subject line and test result. If you take outbound in-house next year, this is the asset you inherit.
What fired in the last 24 hours, what cleared both gates, and what's already in an inbox. This is the 48-hour clock, visible.
| Account | Signals stacked | Detected | Status |
|---|---|---|---|
| Halden Systems | New VP Data (day 6) + 3 platform roles open | 18h ago | in inbox · 31h |
| Corvid Analytics | Series B ($30M) + warehouse migration post | 22h ago | in inbox · 40h |
| Pellham Retail Group | New CTO + 2 data roles open | 7h ago | drafted · awaiting approval |
| Brightmoor Health | Funding only, second signal not found | 11h ago | held at gate 2 |
| Sable Freight | Hiring spike, but 340 employees + logistics | 14h ago | failed ICP gate · dropped |
Yesterday: 34 signals scanned · 11 cleared the ICP gate · 4 stacked to two or more signals · 3 entered sequence within 48 hours. Two accounts were dropped on purpose: single-signal accounts are roughly 20% true positive and aren't worth a warm domain.
Why it matters: you can see what we chose not to contact. Any agency can show you volume; the discipline is in the drops.
Every one of these lands in your inbox
in the first three weeks.