What we learned sending cold emails from our own CRM.

We run our own outbound outreach the same way we ask clients to run theirs: nothing goes out without a human approving the send, and we audit our own CRM the same way we would audit a client's. This post is what that audit found — aggregate lessons only, no recipient names, companies, or domains.

Our CRM audit reviewed the full activity log: 135 event rows spanning three actual send campaigns — an initial 30-lead pilot batch, a roughly 30-lead India logistics campaign, and a roughly 11-lead US logistics campaign, on top of a much larger backlog of leads that were drafted but never approved to send. As of the audit's own reconciliation note, the mailbox showed 236 messages sent against 205 campaign recipients, with 1,116 leads still sitting as unsent drafts and CRM-side status counters (drafted/sent/bounced) that did not agree with each other across two different tabs of the same sheet.

The first lesson was mundane and important: two dashboards in the same spreadsheet reported different totals for "total leads" — about 1,325 in one tab, 1,149 in another. Neither was obviously wrong; they were just counting different things without saying so. When you cannot trust your own counts to agree, you cannot trust a reported bounce rate either. We now treat "the two totals disagree" as a blocking finding on its own, not a rounding curiosity.

The second lesson was about bounces. In the pilot batch of 30 sent emails, the activity log recorded 8 bounce events (one explicitly flagged in the log itself as a duplicate of another). That is a meaningfully high bounce rate for a list that had supposedly been reviewed lead-by-lead before sending, and it pointed at the same root cause seen in the India logistics batch: stale or guessed email addresses from public sources, not verified against a live mailbox before the send.

The third lesson was the biggest one, and it was not about deliverability at all: 1,116 leads were sitting as Gmail drafts, created but never sent, dwarfing the roughly 70 leads actually contacted across all three campaigns combined. A backlog that large sitting unreviewed is itself a risk — drafts age, contacts change roles, and a draft written for a stale situation sent late is worse than not sending at all.

What changed as a result

We now treat a CRM audit as a required, standalone step before any further send authorization — not something to spot-check after the fact. We verify that dashboard totals reconcile against the raw row count before trusting either number. And we treat a large drafted-but-unsent backlog as something to actively clear or archive, not something to leave accumulating, because an old draft is a liability the moment it gets sent without a fresh review.

None of this is a claim about anyone else's CRM hygiene. It is what a careful audit of our own outreach process found, and it is the same standard of evidence — read the actual log, don't average two dashboards that disagree, name what you didn't check — that we bring to a client's systems.