3 Retention Mistakes Costing AI Companies Faster Growth

At Dialed, we believe the winners in AI won't necessarily be the companies that grow the fastest today. They'll be the companies that figure out how to retain that growth.

The amount of capital, attention and customer acquisition flowing into AI right now is incredible, but acquiring customers is only half of the equation. As these companies mature, the difference between a customer who stays for six months and one who stays for three years will have an enormous impact on how efficiently they can continue to grow.

The investors closest to AI have been pointing at this gap for a while. In Generative AI's Act Two, Sequoia's Sonya Huang and Pat Grady called out a major retention gap: generative AI apps had a median daily-to-monthly active user ratio of just 14%, compared with 60–65% for leading consumer companies.

More recently, a16z's Santiago Rodriguez and Alex Immerman described the wave of "AI tourists" in Retention Is All You Need: users who sign up out of curiosity and churn within a couple of months. It's become one of the defining patterns of this era of software.

I've spent more than a decade building lifecycle programs for high-growth companies, including unicorns.

Across that experience, I've found that the biggest lifecycle opportunities are rarely complicated. Most companies are leaving meaningful growth on the table in three places.

1.They aren't communicating with their customers enough.

Most companies dramatically underestimate how much communication it takes to keep a customer engaged over a long period of time.

This doesn't mean sending more generic newsletters or product announcements. It means creating enough relevant touchpoints throughout the customer journey that users continue discovering value, building habits and finding reasons to come back to the product.

A new customer who hasn't completed onboarding should be treated differently from one who has. A customer whose usage has declined over the last 30 days should be treated differently from a power user. Someone who hasn't adopted a feature strongly correlated with retention should receive different communication from someone who uses it every day.

Customers have jobs, meetings, families and dozens of other products competing for their attention. Even if they love a product, they're thinking about it significantly less than the company building it is.

The goal of lifecycle isn't to send more email for the sake of sending more email. It's to create enough useful, timely touchpoints to keep customers moving toward the behaviors that make them more likely to stay.

2.They're missing critical lifecycle journeys.

Most companies have some version of onboarding and a handful of automated emails, but there is a significant difference between having automated emails and having a real lifecycle program.

Those journeys also shouldn't operate in isolation. The best lifecycle programs become increasingly personalized based on what is known about the customer: who they are, why they signed up, what plan they're on, what they've done inside the product, which features they use, which features they haven't discovered and how their behavior is changing.

Most software companies already have an incredible amount of first-party data that can indicate what a customer needs next. The opportunity is connecting that data to the lifecycle program so communication changes with customer behavior.

This is where lifecycle starts becoming much more than email. It's a system that identifies important moments in the customer journey and uses email, in-app messaging, push and other channels to move the customer toward the next valuable action.

3.They don't have enough resources dedicated to lifecycle.

This is the problem we see most often, including at companies that already have strong lifecycle people internally.

The strategy isn't necessarily missing. The capacity to execute it is.

There might be one lifecycle manager responsible for strategy, segmentation, copy, implementation, reporting and testing. A larger company might have a strong lifecycle team, but their roadmap is six months long because every new journey requires coordination across creative, data and engineering.

Either way, lifecycle work gets pushed down the priority list because there is always something more urgent to ship.

The problem is that lifecycle improvements compound. If a company discovers a better onboarding journey this month, every new customer entering that journey can benefit from it. If they identify an intervention that reduces churn among an at-risk segment, that intervention can continue running across future cohorts. If they increase activation by a few percentage points, they're getting more value from the acquisition they're already paying for.

The difference between running five meaningful lifecycle experiments a year and fifty isn't just the performance of those individual tests. It's the accumulated impact of years of improvements being applied across an increasingly large customer base.

Retention is a growth strategy.

AI companies are currently experiencing an extraordinary period of customer acquisition, and understandably, most of the attention is going toward capturing as much of that demand as possible.

But eventually every high-growth company reaches the same reality: acquiring another customer becomes more expensive than keeping the one they already have.

a16z has gone further, describing retention as a core ingredient of "modelbusters," companies that grow faster and for longer than anyone predicted. As more AI companies see their retention curves start to smile, they suggest retention could eventually outpace anything we've seen in SaaS.

That's why we believe retention is going to become one of the defining growth advantages in AI.

Companies don't necessarily need a transformational improvement in churn to change the trajectory of the business. Small improvements in activation, engagement, retention and expansion compound across a growing customer base and make every dollar spent on acquisition more valuable.

The AI companies that win won't simply be the ones that acquire customers the fastest. They'll be the ones that get exceptionally good at keeping them.

That's what we're building Dialed to help companies do.

Lifecycle and retention for fast-growth companies is all we do.