---
title: "A/B test and optimize campaigns"
description: "A simple method to A/B test audiences, invite notes, first messages and calls to action in LinkedIn and email outreach, and keep the winner."
url: https://prosyo.com/docs/campaigns/ab-testing/
section: "Campaigns"
updated: 2026-11-06
---

# A/B test and optimize campaigns

Small changes to the invite note or first message can make a big difference to reply rates. Here's how to test one thing at a time in Prosyo.

## The method

1. **Pick one variable** to test (see below).
2. **Duplicate the campaign:** **Campaigns → New campaign → Copy a campaign**. This copies the sequence only.
3. **Change only that one thing** in the copy.
4. **Split the audience:** import two similar lists, or select half of one list in **Prospects** and use **Add to list** to create list B. Enroll list A in one campaign and list B in the other.
5. **Run both at the same limits and working hours.** Set the copy's **Sending & safety** tab to match the original, including the sending accounts.
6. **Wait for enough data**, ideally 100+ people per version, before deciding.
7. **Keep the winner**, archive the loser, and test the next variable.

## What to test, in order of impact

| # | Variable | Measured by | Example A vs B |
|---|---|---|---|
| 1 | **Audience slice** | Acceptance and reply rate | Founders vs VPs of Sales |
| 2 | **Invite note** | Acceptance rate | No note vs a short personal note |
| 3 | **First message** | Reply rate | A question about them vs a relevant insight |
| 4 | **Call to action** | Positive replies | "Worth a chat?" vs "Want me to send the guide?" |
| 5 | **Channel order** | Reply rate | LinkedIn first vs email first |
| 6 | **Timing** | Reply rate | 3-day vs 5-day delays |
| 7 | **Warm-up touches** | Acceptance rate | With profile visit + like vs without |

## Invite note ideas to test

| Style | Example |
|---|---|
| No note | *(blank)* |
| Shared context | `Hi {{first_name|there}}, we're both in the {{custom.industry|B2B SaaS}} space. Would be good to connect.` |
| Compliment | `Hi {{first_name|there}}, enjoyed your recent post. Would love to connect.` |
| Direct | `Hi {{first_name|there}}, I help {{title|sales leader}}s book more meetings from LinkedIn. Open to connecting?` |

## Read the results

Open **Reports** on both campaigns with the same date range:

- **Acceptance rate** → tests on the audience, invite note and warm-up touches
- **Response rate** → tests on the first message and follow-ups
- **Positive replies** and the **Positive rate** → tests on the call to action and offer

Ask [SalesGPT](https://prosyo.com/docs/ai/salesgpt.md): *"Compare reply rates for UK Founders A and UK Founders B."*

## Rules of thumb

- **One change at a time**, or you won't know what worked.
- **Same week, same limits.** Don't compare a campaign from January with one from March.
- **Don't stop too early.** 20 sends tell you almost nothing.
- **Winning copy goes into templates** so the whole team uses it. See [Templates](https://prosyo.com/docs/campaigns/templates.md).

> **Upgrade:** Starter runs 1 active campaign, so A/B tests have to run one after the other. **Multichannel** runs 3 at once, and **Agency / Team** runs 3 per seat. [See plans](https://prosyo.com/docs/billing/plans-and-pricing.md).