In this article
A run without defined scope stops at a default
When a Mach playbook runs without a deliberate plan, it follows its defaults. A step gets up to three tries before the person has no way left. A person's sequence ends when they reply or opt out. Drafts are reviewed before they send, and the review goes to whoever owns the playbook.
Defaults are conservative choices that prevent a run from doing something clearly harmful. They say nothing about your specific campaign, your list, or your team's capacity to review drafts.
The four decisions below replace defaults with choices. They take about ten minutes to make before a run and prevent the most common reasons for a run producing the wrong output.
Define the list before the run
The list decision covers which companies are on it and why. A list built from a specific signal, for example companies that recently raised a round, produces different results from a list built from a keyword search.
Before the run, write one sentence describing what the companies on the list have in common and why that makes them the right targets now. If the sentence is vague, the list probably needs more filtering.
The list decision also includes size. A run against fifty companies is easier to review than a run against five hundred. If this is the first run of a new playbook, start smaller and verify the output before scaling.
Draw the steps before you press Run
A playbook is a set of drawn steps and arrows, and how many messages a person receives depends on the steps you draw. Each step gets up to three tries; after that the person has no way left in the playbook. The right number of steps depends on the channel, the signal strength, and the expected reply rate for your ICP.
A strong, specific signal justifies more steps than a general list. A prospect who posted about a specific problem last week is more likely to respond to a relevant message than a prospect on a market-segment list with no recent signal.
Draw the steps before the run and resist the temptation to add more after the first batch produces few replies. Low replies after two messages usually mean the message or the list needs work, and more messages will not help. Mach asks you to approve every step before it lets you press Run.
Know what ends a person's sequence
Mach ends a person's sequence on its own when they reply, opt out, book a meeting or bounce, or when their company has an open deal. Removing a person from the playbook also ends it. Beyond that, the stops are the ones you draw into the playbook.
A campaign tied to a specific event or window should end when that window closes. A run targeting companies before a conference should stop when the conference date passes. Pause the playbook then, rather than letting it run past the date.
Write the stop rules before the run starts. If you cannot name the conditions under which the run should stop early, the campaign does not have a clear goal.
Name who reviews the first batch
By default Mach drafts every message for review before it sends. You can change that per step. Who reviews the first batch determines how quickly the run delivers results and what quality problems are caught early.
Name the reviewer before the run starts, before the first drafts arrive. If no reviewer is named, the review goes to the playbook's owner, who may not be the person best placed to judge it.
For a new playbook, the first batch of ten drafts is the most important review. It establishes whether the message template, the research quality, and the ICP definition are producing the output you want. Issues found in the first batch are cheap to fix, and issues found in batch fifty are expensive.
Fill these four before clicking Run
Before pressing Run, write the four decisions on a single note: list description, the steps and their tries, what ends a sequence early, reviewer name. If any of the four is blank, fill it before starting.
After the first batch, review the decisions against the output. Did the list produce the kind of companies you expected? Did the steps feel right for the signal strength? Did sequences end where you thought they would?
The four decisions are per-run choices, so the next run can use different values. Keeping a record of the decisions and the results across runs is the fastest way to learn what works for your ICP.
