Most cold email advice is wrong.
The problem isn't your copy.
It's that you don't know what actually works yet

Better copy framework is not the solution. Better testing is.

People will tell you to focus on relevance, timing, and personalisation.

And they're right, but they still don't guarantee a positive reply.

What actually gets prospects interested is speaking their language.

And the only way to learn their language is to listen to:

- what they respond to
- who responds,
- what they say when they do.

The only reliable way I’ve found to learn this is structured A/B testing from day one.

Here's the system I use when launching a new outbound campaign:

1️⃣ Write your baseline email sequence copy

2️⃣ Pick ONE variable to test per campaign:

- Opening line
- CTA
- Message angle
- Tone
- Social proof

3️⃣ Write your copy variations:

Based on your chosen variables, write distinct versions of each email step.

4️⃣ I usually run these tests inside lemlist so I can track replies and variants side-by-side.

5️⃣ Test with a fraction of your TAM first:

It's tempting to go all-in when you've spent time building a quality list.

Start with 15–20% of your TAM and let the data tell you what to scale.

6️⃣ Launch and give campaigns enough runway:

I typically run campaigns for 7–14 days before concluding.

7️⃣ Set up your reporting tabs in lemlist before results come in:

Before launching, I always decide which metrics actually matter.

For me it's usually:
- Messages sent
- Reply rate by sequence
- Positive replies

Everything else is noise.

Having these set up in advance means you open a tab and see exactly what you need

Instead of setting up filters every time you want to check performance.

8️⃣ Analyse results properly:

Go beyond open and reply rates. Look at positive replies specifically.

Then read the actual messages. What did the ones that got positive replies say?

That's your next campaign's starting point.

With this, you get:

- Extended TAM usage: by the time you've run 2 campaigns, you've tested 4 strategies, not 2
- Faster signal: instead of waiting weeks between single tests, you're generating data in parallel
- Immediate wins: your best-performing variant drives replies while the others teach you what doesn't land
- ICP clarity: multiple data points reveal response patterns you'd never see from a single message
- Compounding knowledge: every campaign informs the next one

The fastest way to improve cold outreach isn’t writing better emails.

It’s running better experiments.

Curious, what’s the most interesting test you've run recently?