How do you improve a campaign you never tracked?

Your CRM has stages and reply rates for every lead. The reasoning that produced each message is gone the moment it sends, so when a campaign underperforms most of us change the opener and run it again.

Groundtruth is the system of record for your outbound. Everything the agent did, why it did it, what came back, and a way to judge what that adds up to.

The journey, the analyst, and the loop

A campaign console. The operator types: copy version c8 didn't move replies, what should c9 change. The analyst answers that line 18 of the copy guide requires a specific detail in every opener, that research found one for 312 of 1,520, and that the rule fired on all 1,520. The centre pane opens the copy guide at line 18 and shows the two cohorts: 6.1 percent where research found a specific, 1.7 percent where the rule made the agent invent one, against 2.9 percent for plain openers. — Asking a campaign what to change. The answer comes back with the turns behind it, and the record opens at the line that caused them.

The journey

Every action on a lead becomes one row: what was known, what was chosen and why, what came back, and which copy version was live. Joined by person they are that lead's whole story. Human touches go in too, and nothing is summarized.

The analyst

Point a question at the record in plain language. It reads every journey at once, and every claim comes back with the turns it came from, so you can open them and disagree.

The loop

You decide what to change and edit the guidance yourself. The version bumps, the record starts filling again, and the next read grades that change on replies and meetings.

A campaign you keep pointing at gets better at telling you what it is doing wrong.

One lead, start to finish

Most of what is worth knowing only appears at the length of one lead's whole story. Here is one, all of him.

Journey · RevOps Leaders — Q3 · person p_1182

Jul 08researchcopy c6 · v2enrichedno specific found KnewVP RevOps at [Company] · 340 people · stack Salesforce, Outreach · no post history, no funding event

Jul 08email_1copy c6 · v2enriched → sentopened 2×, no reply WhyResearch found no specific. messaging.md:18 requires one in every opener, so anchoring on the firmographic growth signal instead. “Saw [Company] is scaling the GTM team. How are you thinking about it?”

Jul 11email_2copy c6 · v2sentno reply

Jul 15escalatecopy c7 · v2sentreply detection unclear, held for a human

Jul 16li_connectcopy c7 · v3sent → invitedinvite sent

Jul 18—inboundinvited → connectedaccepted

Jul 18dm_opencopy c7 · v3connected → openedsent

Jul 19replycopy c7 · v3opened → repliedsent He wrote“honestly we're mid-migration off Salesforce so nothing's clean right now. why do you ask?” “Curious how you're routing leads while that's in flight. Everyone I've asked has a different stopgap for the gap.”

Jul 19email_3copy c7 · v3repliedsent on schedulethe email cadence had not seen the LinkedIn reply

Jul 23ask_callcopy c8 · v3repliedasked

Aug 01—cadencereplied → parkedno answer since

The reasoning behind every choice is recorded next to it, which is the only reason you can argue with any of it.

I'll point it at one of your campaigns

If you run outbound and want to see what your own record says, get in touch. I'll wire it up and read the first one with you.

Email me LinkedIn Book 15 minutes

Isn't this just logging?

Mechanically, yes. The difference is what gets recorded and at what grain. Logs answer "what did the system do". These records answer "what did it know, what did it choose, why, and what came back", keyed to a person so a whole journey reconstructs, and stamped with the strategy and copy version in force so a change is attributable to a cohort.

The test I use: could an analyst who was not there reconstruct the campaign and disagree with me about it? On my own campaign the answer was yes, twice, and both times the analyst was right.