I built a QA tool that grades every campaign before it sends

Every sales leader has launched a campaign they regret. A broken merge field. A link that went nowhere. A sequence that went to 5,000 people and came back with one reply.

I have launched several. So I built a tool to stop me.

What it checks

Before any sequence goes live, the pre-flight check inside CORA reads it the way a sharp manager would, and returns one verdict: ready, fix first, or block.

• Copy made only of AI variables, with no human-written message underneath. That one is an automatic block.
• Custom fields that are missing on some of the contacts, so somebody gets "Hi ,".
• A link in the first email, which hurts delivery.
• Tracking switched on where it should not be.
• No senders assigned.
• Sequences that run too long, and punctuation habits that read like a machine wrote them.

Did it work?

We ran it backward over 66 campaigns that had already sent.

The 19 it would have marked ready averaged a 2.30% human reply rate and 1.17% opportunities. The 40 it would have sent back to fix averaged 0.66% and 0.12%. The 7 it would have blocked averaged 0.35%.

To be fair to the numbers, the rules were drawn from that same history, so this is not a clean experiment. But the gap is big enough to act on, and every new campaign now adds to an honest test.

One finding stopped me cold. Emails written entirely from AI variables replied at 0.35%. Emails with a real message replied at 1.29%. Personalization is not the same thing as a person. In Buyer Methodology terms, those emails fail the very first driver, Pattern Interrupt.

It keeps watching after launch

Passing pre-flight is not the end. A health monitor watches every live campaign and raises a flag when bounces climb above 3%, when there are no automatic replies at all after 500 sends, or when replies stay under 0.5% after two weeks and a thousand people.

That middle one sounds strange. It is my favorite. Out-of-office replies are proof your email arrived. When two of our campaigns reached more than 5,000 people with almost no auto replies and one human reply, the likely answer was not bad copy. It was emails that were blank or never delivered.

It checks what we say, too

The same tool audits copy before it ships anywhere. It blocks guarantees, superlatives, invented clients, press or case studies, unsourced statistics, prices we do not actually sell, and any sentence that pitches "leads." Only claims on an approved list get through.

It is not glamorous. It saves us from the most expensive kind of mistake, the confident one. The same lesson showed up in our signals data: the offer and the audience matter more than clever mechanics.

QA for people, not just campaigns

The last check is on us. When we audited the replies that showed real buying interest, we had answered 78% of the ones we could trace, with a median response time of about 14 hours. Good, not great. Then the gap: referrals. Of 56 people who pointed us to a colleague, 37 never got an answer.

Replies like these are exactly what reply triage is built to route, and the audit showed where it still needed work.

No campaign tool would have found that. A QA mindset did. That is the habit I bring to a sales floor: check the work before it ships, and check the follow-through after.