Before you write a single ad, run the message through outbound. Send two opening lines to the same kind of list, count the replies, and keep the winner. That line is your first headline, and the segment that replied is your first audience.
We gave this advice to a team about to launch its first Meta and Google pilot. They wanted ads to tell them what to amplify. Our answer: find out in outbound first, where it costs a fraction of the ad budget and you choose every person who reads the message.
Why outbound is the better place to test a message
In an ad test, the platform decides who sees each version. You give Google keywords and give Meta signals, and their systems pick the people. On Meta, Advantage+ audience treats age, gender and detailed targeting as suggestions that don't always constrain who sees the ad. So when one version wins, you can't be sure if the message won or the audience it happened to reach.
In outbound you pick every recipient. If engineering leads reply and finance leads don't, you know which segment the message works for. Count the interested replies separately, since "not interested" is a reply too.
How to run the test
- Split the list by segment. One list of engineering people, one of finance, one of anyone else you suspect is a buyer. The same message goes to each, so the segment is the variable.
- Test one line at a time. In lemlist you can build a Variant B of a sequence and compare the two. Its help center recommends changing one major variable, such as the opening line, with the same list and sending window.
- Stop the sequence when someone replies. lemlist lets you set a reply to continue, pause or end the lead. Pause or end, then answer the reply yourself.
- Give it 10 working days. We don't send on weekends, so count working days, not calendar days. One of our own LinkedIn sequences, since paused, reached a 14% reply rate. Keep the line that clearly beats the others on replies.
Read the first batch carefully. About 50 people is a direction, not proof that the next 200 or 500 will reply the same way. If the opener pitches the product, see why your first DM shouldn't mention what you sell.
Move the winning line into your ads
A message written for one person rarely fits an ad as is. Google's responsive search ad headlines support up to 30 characters and descriptions up to 90, so keep the idea that earned the reply, cut it to a headline, and put the reason people replied in a description. Use the same line as the headline on the landing page, so the ad and the page say the same thing.
Then target the segment that replied. On Meta, entering it as a suggestion steers delivery but can reach beyond it. If reaching that segment matters more than the cheapest results, Meta's option to further limit your audience turns those settings into hard limits.
Short on time? Run outbound and ads side by side. The outbound results will still tell you which ad message to keep.
The takeaways
- Test the message in outbound before you test it in ads. You choose every recipient.
- Split the list by segment and change one line at a time.
- Set replies to pause or end the lead, and run at least 10 working days.
- Cut the winning opener to fit: 30 characters for a Google headline, 90 for a description.
- Target the segment that replied. On Meta, a suggestion steers delivery but doesn't limit it.
Common questions
Can I test ad copy with cold outreach before running ads?
Yes, for the message and the audience. Send two versions of the opening line to the same kind of list and count replies. The line that wins is a tested starting point for your ad headline, and the segment that replies tells you who to target. It won't tell you what an ad will cost per click or per lead.
How do I A/B test an opening line in lemlist?
lemlist lets you create a Variant B of a sequence, either by duplicating Variant A or starting from scratch, and compare the two. Its help center recommends changing one major variable at a time, such as the opening line, with the same lead list and sending window. Choosing a winner ends the test and can't be undone.
How long should an outbound message test run?
We give it at least 10 working days. We don't send on weekends, so calendar days overstate the sample. One first batch of about 50 people isn't proof that the next 200 or 500 will respond the same way, so treat early results as a direction.
Will a cold message fit in a Google Ads headline?
Usually not as written. Google's responsive search ad headlines support up to 30 characters and descriptions up to 90, and you can add up to 15 headlines and 4 descriptions. Keep the idea that earned the reply and cut it down to fit, then use the rest in a description.
Can I target only the segment that replied on Meta?
Not with suggestions alone. With Advantage+ audience, Meta says suggestions such as age, gender and detailed targeting don't always constrain your audience, and ads can reach other people when Meta's AI expects them to respond. Locations, minimum age, custom audience exclusions and languages are hard limits. To hold the audience to your segment, use Meta's option to further limit the audience.