Xevolution · Free file 07

The Failure Runbook

Something is wrong and you want the one change that fixes it, not a list of twenty things to try. This is the order I work in when a send goes sideways, given here in full. It is file 07 of seven in The Outreach Operating System. There is no price on this page and nothing to sign up for.

Change one variable, then wait long enough to learn something. Most outreach operations are not broken — they are unread.

Something is wrong and you need to know what to change first. Work the steps in order. Do not change two things at once.

First: is it the mail or the message?

Check in this order, every time:

  1. Authentication passing? Send yourself a test to a Gmail address and open the original. If SPF, DKIM or DMARC shows fail, stop — nothing else matters until this is fixed.
  2. Any blocklist hit? Look your sending domain up on a blocklist checker. A listing explains everything downstream.
  3. Bounce rate this batch? Above 5 percent, the list is the problem.
  4. Delivered but unread? Open rate below 20 percent, the subject line is the problem.
  5. Opened but no replies? The body or the offer is the problem.

Only after those five do you get to "the market is tough". That answer is nearly always wrong and it stops you looking.

Symptom: open rates collapsed

Order of moves:

  1. Confirm authentication still passes (mailbox providers break DKIM silently when a DNS record changes).
  2. Check whether you changed anything about how you send — a new tool, a new IP, a new domain.
  3. Change subject lines only (prompt A5). Run two variants, 50/50, one week.
  4. If opens do not recover, the domain is tired. Rest it two weeks and send from a second one.

Do not: rewrite the body. Opens are not caused by the body.

Symptom: bounce rate above 5 percent

Order of moves:

  1. Stop sending to new rows today.
  2. Re-verify the entire list. Anything not `valid` comes off.
  3. Check the source of the bad rows — usually one directory that has rotted.
  4. Restart at half your previous daily volume for one week.

Why it matters: providers measure bounce rate as a reputation signal. Two bad days can cost you a month of warm-up.

Symptom: a spam trap fired / you appear on a blocklist

Order of moves:

  1. Identify the listed address, remove it, and record it as a trap.
  2. Find where it came from — almost always a purchased or harvested list, or a list older than a year.
  3. Submit a delisting request; be brief and truthful about what you changed.
  4. Cut volume by 50 percent for two weeks after delisting and monitor.
  5. If it was a purchased list, delete the whole source. Never "clean" it.

Symptom: a client threatens to leave

Order of moves:

  1. Reply the same day and ask one question: what number did not move that you expected to move?
  2. Show the data you have — sent, opened, replied, meetings — even if it is bad.
  3. Name the single change you would make and a date you would judge it by.
  4. Offer to pause billing for the two weeks while it is tested.

Do not: discount to keep them. A discount keeps a client who was already leaving and teaches them the price was never real.

Symptom: replies are hostile or complaints arrive

  1. Remove the person, same hour, permanently.
  2. Stop the batch. Diagnose the list before the next one.
  3. Read the last fifty sends for anything that reads as deceptive — a false "Re:", an invented referral, an unclear sender.
  4. If there were multiple complaints, cut the volume in half for a month and tighten the list standard.

Symptom: you have no time to run this

This is the real failure mode and it is not technical. Choose one, in this order:

  1. Cut the volume to 15 a day. A small consistent operation beats a large abandoned one.
  2. Cut the channels — one list, one sequence, no A/B testing until it is stable.
  3. Raise the price on the Run tier so the hours are paid for.
  4. Say no to the next client until the current one is stable.

The order matters: reducing scope before raising price is how an operator ends up busy and broke.

The one rule under all of this

Change one variable, then wait long enough to learn something. Most outreach operations are not broken — they are unread, because a dozen changes were made in a week and nothing can be attributed. Patience is a technical skill here.

File 07 of seven. Quoted in full from The Outreach Operating System by Xevolution. No tracking, no email capture, no expiry — save it, print it, keep it next to whatever you send from.

The other free files: the 90-Day Sending Calendar at /free, the Inbox Triage Decision Tree at /free/inbox-triage-decision-tree, the Objection Library at /free/objection-library, the Permission and Compliance Sheet at /free/permission-and-compliance-sheet, and the Target List Standard at /free/target-list-standard. Questions about your own setup: support@xevolution.biz