Zevari
Zevari

Outbound Academy · Guide

Test before you scale

You find the message that works on a few hundred leads, then spend the rest of your list on the winner, instead of finding out at lead 2,000 that your angle never landed.

Prove it by hand first

Before you automate anything, run the sequence by hand on about 10 people. Send the invite yourself, write the first message yourself, and see what comes back.

Ten conversations will teach you more than any template: which question gets answered, which offer gets a "sure, send it," and which line makes people go quiet. Once a sequence works by hand, automating it multiplies something that already works.

Change one thing per campaign

When you compare two campaigns, they should differ in exactly one way. Otherwise you can't tell what made the difference.

Good things to test, one at a time:

  • The angle: the problem you lead with.
  • The offer: a guide vs an audit vs a case study.
  • The persona: founders vs heads of sales, for example.
  • The region.
  • The goal: a reply, a resource request, an event signup.
  • Note or no note on the connection request.

If your angle test also changes the persona, and the persona test also changes the offer, you'll end up with numbers you can't use.

Give each test enough leads

Small samples lie. Five replies from 40 leads might be luck. A useful benchmark:

  • Run about 150 leads a week through each test campaign.
  • Wait until a persona or a copy approach has reached 100 to 200 leads before you judge it.

The weekly testing rhythm

A simple routine that works:

  1. Start a handful of campaigns, each testing one variable, each with its own list.
  2. Feed each one about 150 leads a week, at a steady daily rate (see Pacing).
  3. After a few weeks, rank them by results.
  4. Drop the bottom half and rerun the top half. With six campaigns, that's the bottom three out and the top three running again.
  5. Repeat for around six weeks. By then you'll know your best angle for each persona.

What to measure

Each campaign's page shows Acceptance and Reply rate. Use them, but look past them:

  • Acceptance tells you whether your targeting and your invite work.
  • Reply rate tells you whether your messages work.
  • Positive replies and calls booked tell you whether the campaign works. A campaign with fewer replies but more booked calls is the better campaign.

If you offer a resource, measure calls booked per resource, not downloads. A guide that 200 people download and nobody acts on isn't working.

From reviewing to Autopilot

Testing is also when you train your AI team. While you train them, every connection request, message, comment and post waits for your approval. Review and edit drafts until approvals become rubber stamps, because the drafts already sound like you and go to the right people. Then switch that specialist to Autopilot on the AI team page, and review results weekly instead of message by message. You can switch it back to Reviews with you at any time.

How to set up a test in Zevari

  1. Build your list and score it (see Why ICP scoring comes first).
  2. Split the list into equal, similar groups, one per variation.
  3. Create one campaign per variation from the same template, changing only the variable you're testing.
  4. Name each campaign after its variable, like "Ops leaders: hidden-cost angle", so the results are easy to read.
  5. Add people to each campaign at the same daily rate.

Templates that use this

Any template can be tested this way. Good places to start:

  • Connect and say hi: the simplest shape for testing your first message
  • Connect and four follow-ups: test the angle in message 2
  • Three cold emails: test subject lines and triggers
  • LinkedIn-active core: test note vs no note on the invite
Next: When they replyEvery reply turns into the right next step, whether that's a booked call, a follow-up at the right time, or a clean stop that protects your reputation.