A map of the four leaks in comment-to-DM automation

1. The right question: where it leaks, not whether it works

Most people running a comment-to-DM automation check one thing: how many comments came in. That is the wrong question. Entry can work perfectly and half the people still never get the guide.

• Perfect entry measures intake, not delivery.
• The number that matters is how many got the guide in hand, not how many commented.
• Between the comment and the delivery there are several steps, and every step can leak.
• If you are not counting every step separately, you are blind to the leak.

2. How the funnel is built

Before checking for a leak, you need to draw the funnel. A comment-to-DM automation has four steps, and each one has its own point of failure.

Comment with code word
   |
   v
Opening DM message
   |
   v
Follow gate
   |
   v
Guide sent

3. The real example: 98 entered, 47 got it

I checked my system against Meta data over ten days. Entry was perfect: 105 comments with the code word, zero missed. But when I counted delivery, the picture flipped: out of 98 who commented, only 47 got the guide in hand. 51 leaked out after entry had already worked.

• Entry: 105 comments, 0 missed. Everyone who commented entered the funnel.
• Delivery: 47 out of 98 got the guide.
• The gap: 51 people disappeared between entry and delivery.
• Without counting by step, I would have celebrated "zero missed" and never seen the hole.

4. Leak point 1: the message falls into Message Requests

The biggest leak: 39 people never answered the opening message. Not because they ignored it, but because for anyone who does not follow you, the DM lands in the Message Requests folder, and most people never open that folder. They simply never saw the message.

• 39 out of 51 leaks happened here. One point responsible for most of the damage.
• The reason: a DM from someone you do not follow lands in Requests, not the main inbox.
• The check: how many of your audience do not follow you at all when they comment.
• The fix: design the DM flow around people who do not follow you, not around people who do.

5. Leak points 2, 3, 4: a disabled gate, free text, a glitch

The three remaining leaks are smaller but each is real. Together they account for 12 people.

• Disabled follow gate (7 people): asked to follow, and nobody checked them again after they followed because the gate was disabled.
• Free text (4 people): replied in their own words that did not match the condition, so the send never triggered.
• API glitch (one person): a one-time technical failure in the send.
• Every one of these is completely invisible if you only count entries.

6. The fix: turn the follow gate back on

The second-biggest leak, the seven who were asked to follow, was closed by turning the follow gate back on. The gate does one thing: once someone actually follows, it sends them the guide once. When it is off, people follow and nobody ever checks them again. One action closed an entire step.

• Follow gate = a repeat check after following, then one automatic send.
• Off, it leaks. On, it closes the step.
• Not every fix needs a rebuild. Sometimes it is turning back on a step someone disabled.
• After every fix, repeat the check and count delivery again.

7. How to check your own funnel

This is the process I run. It works on any comment-to-DM automation, not just mine.

1. Pull every comment with the code word for a date range.
2. Count entry: how many commented, how many entered the funnel.
3. Count delivery: how many actually got the guide in hand.
4. Break the gap down by step:
   - How many never answered the opening message?
   - How many got stuck at the follow gate?
   - How many replied with text that did not match?
   - How many fell to a technical glitch?
5. Close the step with the biggest leak first.
6. Run it again and confirm delivery went up.

8. Three rules to remember

• Measure delivery, not entry. "Zero missed" only covers the door.
• Close every step separately. One big leak is usually responsible for most of the damage.
• Check who never answered. That is where the leak nobody sees is hiding.