183 accounts, and on 42 percent of them nobody to write to
183 of our accounts had only ever been called, out of 316 with an active owner and an active status. On 42 percent of those 183 there was nobody named on the record at all.
That is the finding, and getting to it was the work, because five different causes could each explain the same account and several of ours had more than one.
Five causes, ranked so each account counts once
Counting how often each cause occurs double counts every account that has two, and every number comes out too big. The shares then add up to more than 100 percent, and nobody can prioritise from that.
So I assigned each account to the first applicable cause in a fixed order, and only that one.
| First applicable cause | Share of the 183 |
|---|---|
| No contact on the account at all | 42 percent |
| An email contact enrolled in a sequence, and still only calls | 25 percent |
| A single touch and nothing after it | 19 percent |
| An email contact never enrolled in anything | 13 percent |
| A contact with no email address | one percent |
42 percent had nobody named on the account. On those accounts calling was the only channel that works.
A phone number can be found for a company. A person has to already be on the record.
From six months inside a B2B GTM organization with 36,000 accounts in its CRM.
The same cause runs at one percent among the accounts in that set that were worked on more than one channel. That contrast is what makes the 42 percent a cause rather than a coincidence.
The second largest cause was one sequence that sent nothing
80 percent of the accounts in the second bucket hung on one single sequence of ours.
That sequence had 73 contacts enrolled, one open and zero replies, and it had been running for weeks. Every other sequence we ran sat between 18 and 60 percent open. That band is read off a lifetime field, so it flatters the others rather than this one, which makes the gap wider than it looks.
Why it produced nothing is still open, and I could not settle it from the data. The step structure of a sequence is not readable through the API, so either it contains no email steps at all or it contains email steps that never fire. Only opening it in the interface tells you which, and nobody had.
Two explanations I checked and dropped
Both were more comfortable than the answer, and neither survived.
Opt-outs. The theory was that these people had unsubscribed and calling was what was left. The cohort contained exactly one unsubscribed contact.
Enriched decision makers lying around unused. The theory was that good contacts existed and reps ignored them. On a narrow check that held for nine accounts, and one of those was a real case. The strong cases do exist. They sit on the accounts that had been worked hardest, which is a different cohort and a separate piece of work.
Is this an old backlog?
It is current work. 82 of our accounts carried exactly one logged touch, and 70 of them had been called within 11 days of the analysis.
That 82 is counted differently from the 19 percent in the table, and it is larger, because the cascade assigns each account to one cause only. An account with nobody on it and a single touch lands in the first bucket, because that is the first cause that applies.
The distinction matters for what you do with the finding. An old backlog is a cleanup job for whoever owns the data. Current work is a process question for whoever owns the team, and our answer is a check before the dialer.
What to do instead
- Write the order of the causes down before you count anything. A ranking answers the question. A list of everything that is wrong leaves the reader to guess which one to fix.
- Check the comparison group. The same cause among the accounts that were worked properly is what tells you whether it is a cause or a coincidence.
- Put a contact check in front of the dialer. Where the largest cause is no named person, the fix sits upstream of anything a rep does.
- Read the last send date on every sequence, not the enrollment count. A sequence with people in it and nothing going out looks healthy from every report.
- Test the comfortable explanations first and say that you did. Opt-outs and unused contacts were both worth checking, both were small, and naming them is what stops somebody raising them afterwards.
Where nobody is on the record at all, the upstream question is why, and that is companies scoring as a good fit with no contact on them. On the accounts that do have people on them, the cost shows up as decision makers nobody had contacted on the most heavily worked accounts.