The most common complaint about outbound agencies is not that they failed to book meetings. It is this, from a founder hiring for exactly this role:
"In the past we've worked with someone who booked meetings, but we never had visibility into what was working."
Meetings happened. Nobody could say why, which campaigns produced them, or what to do more of. When that person left, the knowledge left with them.
The fix is boring and completely effective: a weekly report with the right lines on it, agreed before the first send.
The report, line by line
| Line | Why it has to be there |
|---|---|
| Emails sent | The base everything else divides by |
| Delivered | Sent minus bounces. If this gap grows, infrastructure is failing |
| Bounce rate | Above 2% is a data problem eating your sender reputation |
| Replies | Everything that came back, including the negatives |
| Positive replies | The line most reports omit, and the one that matters |
| Meetings booked | What the pipeline is promised |
| Meetings held | What the pipeline actually is |
| Show rate | Booked versus held. Below 70% means reply handling is broken |
Weekly, per campaign, not blended. A blended number across campaigns is how a failing angle hides inside a working one for a month.
Why positive replies is the line that gets hidden
Reply rate counts "take me off this list" exactly the same as "yes, let's talk."
On one real campaign: 13,708 sent, 202 replies, 107 positive. Nearly half the replies were rejections and auto-responders. Report "202 replies" and the campaign sounds twice as good as it is. Forecast from it and your pipeline number is double reality.
So the question to ask any agency before signing: "what do you count as a positive reply?" If there is no crisp answer — interested, qualified, wants to talk — the reporting will inflate, not because anyone is lying but because the flattering definition is the default.
The two asks that keep everyone honest
Shared access to the sending tool and the lists. You can see what went out, to whom, and what came back. This costs an honest agency nothing. Resistance to it is information.
A definition of "meeting" in writing. Held, not booked. Right seniority, right company size, no-shows excluded. Almost every bad agency engagement I have seen went wrong on this definition rather than on price — the full list of questions is in how to choose a cold email agency.
What a good weekly note reads like
Numbers first, then three sentences: what worked, what got cut, what is being tested next. Something like — the stack angle is outperforming the price angle two to one on positives, so price rotates out Monday; bounce crept to 1.8% on one domain batch and those inboxes are resting this week.
That is the entire job of the report: making the campaign legible enough that you could hand it to someone new tomorrow and lose nothing but the person.
What you should not see
Open rates as a headline. Unreliable since Apple began pre-loading images, and tracking pixels hurt placement anyway. An agency leading with opens is leading with the number least connected to revenue.
Screenshots instead of access. A cropped dashboard shows you what someone chose to show you.
Monthly reporting. A month is long enough for a broken campaign to burn a domain fleet. Weekly is the minimum cadence at which problems are cheaper to fix than to discover.
If the reporting conversation makes an agency uncomfortable, you have learned the cheapest possible lesson about what the engagement would have been like.