What's up, it's Zayd

Last week, Shubh (Head of Growth at Valley) and I spent an hour going back and forth on a problem we still haven't resolved. I'm sharing it here because I think it's one that a lot of early-stage SaaS teams hit (and because I genuinely want to know what you'd do).

Here's the context first.

Over the past year, we cut churn at Valley by more than 50%. We did it by getting more selective about who we bring on.

If your ICP isn't active on LinkedIn, not a fit. If you've never done outbound before, not a fit. If you don't know exactly who your target buyer is, not a fit yet.

Even with all that filtering, some people still churn, and when we dug into why, it came down to two reasons.

Here's what makes this painful. We know that if someone launches a few campaigns and gets the messaging dialed in, they're hooked.

They stay. They expand. The product works…but only once you get past those two walls.

So, which wall would you tear down first?

Forwarded this email? Get Enabled in your inbox each week

My favorite finds of the week

This is my argument, so I'll steelman it.

There's no point getting people through the front door faster if they immediately hit a wall that makes the product feel broken. A bad first experience with the AI kills trust instantly. And once that trust is gone; once someone sends a campaign and the messages feel generic or off-brand; no amount of smooth onboarding can save you.

Think about it from the user's perspective.

They signed up because they believed Valley could write outreach in their voice. They go through setup, launch their first campaign, and the messages come back sounding like every other AI tool they've tried. That's a 'this product doesn't work' problem.

Fixing activation just gets more people to that wall faster. If the AI experience is broken, you're accelerating churn, not reducing it.

The counterargument to my own argument: fixing AI training is hard.

It requires months of engineering work, backend infrastructure, and research that doesn't have a quick payoff. We won't see the results for a long time.

This is Shubh's argument.

Right now, a significant chunk of users never reach the point where they'd encounter AI training issues. They're leaving before they launch a single campaign. We're losing people before the product has had any chance to prove itself.

If you fix activation (make the setup faster, clearer, more guided, etc.) some of those people will stick around long enough to experience the product. Some of them will love it even with the current AI training flow. You can't optimize an experience that people never reach.

His logic: fix the leaky bucket closest to the front door before you fix the one further in. Get more people past the first wall, learn from them, and then use those learnings to address the second wall with better data.

The counterargument to his argument: some of those newly activated users are just going to hit the AI wall instead. You've moved the churn event, you haven't eliminated it.

The B2B Growth and Sales Creator Handbook

Both arguments hold up..that's the problem.

If we fix activation first, we get faster feedback loops and quicker impact on the numbers, but we might be sending people to a worse experience.

If we fix AI training first, we solve the more permanent problem, but it takes longer and the impact won't show for months. And some of those months, we're watching people churn at the activation stage when we could have been learning from them instead.

There's also a resource constraint. We can't do both at full speed simultaneously. Whichever we pick for the next three months is the one that gets the team's full attention.

The honest answer is that we don't have enough data to know with certainty which problem is causing more total churn. We know both walls exist. We don't know exactly how many people hit each one.

When I get stuck on decisions like this, I try to invert the problem. Instead of asking 'what should we fix first?' I ask 'what would guarantee that we don't succeed?'

Answer one: people leave before experiencing the product. That's activation.

Answer two: people experience the product and decide it doesn't work. That's AI training.

Both answers are valid. But answer two feels more existential to me. A bad first experience with the core product promise (writing outreach in your voice) isn't a recoverable mistake. Activation friction is annoying. Broken core value is brand damage.

That's why I keep landing on AI training, butttt I still think Shubh might be right.

Here's the question I'm genuinely asking our readers: