The gate checklist
31 leads passed a gate they should not have passed.
The dashboard was all green. We found it in the replies, three days later.
From our own operation, with the mistake described the way it happened rather than the way it reads well.
What happened
A gate in our qualification checks whether a lead fits the ideal customer profile. Company size, industry, role. Leads that pass move on into a sequence.
31 leads from LinkedIn engagement passed that gate without anyone ever checking whether the topic they engaged with was relevant. 30 of them were already enrolled in a sequence by the time we noticed.
The gate checked "qualified" and never checked "topic relevant". The two sound like the same thing. They are not.
A managing director at a perfectly matching company who commented under a post about parental leave is a flawless ICP match and an entirely wrong reason to reach out.
Why a dashboard will not show you
Every number was correct. Leads qualified: 31. Errors: 0. Sequence running. There is no metric for "checked the right criterion": a gate reports that it decided, not what it decided about.
We noticed it in the replies. People answered politely and visibly had no idea why we were writing to them. That is the most expensive way to find a configuration error: through your reputation.
The class of failure
The gate was not broken. It did exactly what it said. The error sat one level up: between what we meant to check and what we wrote down.
This class of failure has three reliable traits:
Nothing breaks
No error, no red bar. The automation runs cleanly, just on the wrong basis.
The numbers look good
More leads pass the gate than expected. That reads as success, not as a warning.
It surfaces late
Only in the replies, days later, once the sequence is already running.
The checklist
These seven questions before any gate goes live. They cost ten minutes and would have prevented our case.
When it is already running
If you catch it once people are already enrolled, there are two honest paths. Both are defensible; one has to be chosen and written down.
Stops the damage immediately. Costs you the information about how the wrongly enrolled leads would have responded, which would have been useful for the next version of the gate.
Gives you real data on what the mistake actually cost. Costs reputation with people who should never have been contacted. Only defensible if the message is harmless on its own.
What does not work: leaving the decision open while the sequence quietly keeps running. Then you get both downsides and neither upside.
The lesson that stays
Gates check what is written, not what was meant. Ten minutes of review before going live is cheaper than three days of replies from people who do not know why you wrote to them.
Take your sharpest gate
The one with a send directly behind it. Run the seven questions, especially two and five.
Back to the topQualification that checks the reason too
We build your gates so fit and reason are checked separately, with a real "do not know" exit and a signal when the pass rate turns.
Book an intro call