Approved Isn't Ready: The Metric That Lies
August 12, 2026 · sent
Approved Isn't Ready: The Metric That Lies
Happy Tuesday — Brian here. One piece worth your time up top, then a couple of shorter things I've been chewing on. Hit reply anytime; I read everything.
What did the dashboard say?
It reported approved leads. Ten of them, sitting in the queue, ready for the morning drip. That number was correct. Every one of those ten had been reviewed and approved by a human.
What it did not report was that none of them had a drafted message body. The send step discovers a missing body at send time, one lead at a time, and fails each one permanently. So the morning would have produced ten terminal failures and zero emails, while the readout said we were armed and loaded.
The two numbers had always been different. On the one day it mattered, they differed by the entire batch.
Why do the numbers drift apart?
Because the easy number is upstream of the work and the honest number is downstream of it. Approved is a decision someone made. Sendable is a state the whole pipeline has to agree on: a contact, a body, no suppression, under the daily cap.
Every step you add between the decision and the outcome is another place they can diverge. Nobody notices, because the upstream number keeps rising and looks like progress.
- Approved leads, drafted emails, booked demos — decisions, and easy to count
- Delivered, replied, closed — outcomes, and harder to reach
- The gap between them is your actual failure rate, and it is usually invisible
What should you count instead?
Ask what could complete right now, with no further human input. Not what is in the pipeline — what would move if you did nothing today.
For a sales list: how many contacts have an address, a written message, and no suppression flag. For a content queue: how many pieces are finished, not how many ideas exist. For a hiring pipeline: how many candidates have a scheduled next step on a real calendar.
In our case the fix was a function that returns three numbers instead of one: approved, ready, and blocked with a reason attached. The readout now says "0 ready to send of 10 approved — 10 no drafted body." Same data, and the second version is impossible to misread.
How do you know which number you are looking at?
Try this on whatever dashboard you check most. Pick the headline number and ask: if every human went on holiday tomorrow, would this number still turn into revenue?
If yes, it is an outcome and you can trust it. If no, it is a decision someone made, and there is a second number underneath it that nobody is showing you.
The number that made us comfortable was the number that let us skip the check. That is the pattern worth watching for. A metric that only reports good news is not measuring anything — it is a mirror pointed at your intentions.
What does this cost when you get it wrong?
An entire batch, and the day you find out is the day it was supposed to work. We caught ours while loading the first real cohort, before a single send. The alternative was ten permanently failed leads and a report that showed green.
Count the thing that would happen without you. Everything else is a plan.
Also this week
AI Agents vs ChatGPT (Don't Connect Your Apps)
A chat window with access to your calendar is not an employee. It is a well-read intern who forgets everything at the end of each conversation and never once picks up the phone while you are asleep. When you connect your business apps to a general chat assistant, you are buying a better way to ask questions about your business. That has value. It is not the same as buying someone who does the work. The difference shows up in one word: initiative. An assistant waits to be asked. An employee notices the lead came in at 9:04pm and books it before you wake up. Everything about how you would judge a hire — did it act, did it follow up, did it finish — assumes the thing acts on its own.
Why Automations Fail Silently (Check the Trigger)
A recurring job that needs someone awake to start it is not automated. It is a chore with a calendar invite, and the difference only shows up on the morning nobody opens the laptop. The work itself is rarely the problem. The problem is the step before it — the one where a person has to be present for anything to begin. That step never appears in the diagram. Nobody draws a box labelled "Dave opens his laptop," so nobody notices when it quietly becomes the thing everything else depends on. Picture the review requests you send after every job. The template is written. The list is accurate. The message is good. And it goes out when somebody remembers to send it.
---
Ready to put it to work? I'll set up Mavrick inside your business — the AI coworker that answers every call, calls new leads back in under 60 seconds, qualifies, and books the job.