Building for rapid growth in AI video creation

Captions has grown quickly into a leading player in AI video creation, helping creators generate and edit high-quality, short-form content with ease. In just a few years, the company has gone from startup to scale-up, pushing weekly app releases across iOS, Android, and web—all while staying lean.

That pace of innovation requires discipline. As their team scaled, Captions needed a system that enabled them to move fast without guessing—one that made every change measurable and every release reversible.

“We chose Statsig because we knew rapid iteration and data-backed decisions would be critical to building a great generative AI product. It gave us the infrastructure to move fast without second-guessing.”

Dwight Churchill

Co-founder, Captions

A culture of controlled launches and fast feedback loops

A couple years into adoption, Statsig has become deeply embedded in how Captions builds product. Every new feature—across all platforms—is gated or tested before launch.

From backend engineers to PMs and designers, every product team uses Statsig as part of their standard workflow. Controlled rollouts are the default: even features that seem like safe bets start with a 10% gate, scale to 50%, and are only launched to 100% after metrics are reviewed.

“We’re pretty strict about controllability—every code change goes behind a gate or experiment. We don’t have time for work that doesn’t move the needle.”

Meera Parat

Growth PM, Captions

Experiments come into play when comparing multiple variants or isolating overlapping surfaces—like paywalls and onboarding flows.

Across both feature flags and experiments, Statsig Analytics gives them a way to track impact and dig into the metrics that matter. Whether its measuring funnel drop-off or checking performance on a new feature, analytics can be super helpful to the team when evaluating launches on a deeper level.

From product intuition to measurable impact

As Captions scaled, Statsig became essential not just for de-risking launches, but for validating the features that actually drive revenue.

One standout experiment focused on their AI-powered video editor. Previously capped at one minute, the team tested increasing the limit to two and three minutes. Both variants led to a measurable lift in subscription rate, confirming strong user demand. That insight gave the team confidence to invest further in the feature—and later consider gating longer edits behind higher-tier plans.

Another experiment tackled watermarking, aimed at reducing users screen recording videos to avoid subscribing. The team tested two versions: one aggressive and full-screen, the other more subtle. The aggressive watermark significantly hurt subscriptions and exports, while the simpler variant held neutral. Within a week, the team rolled back the more disruptive version and began improving the lighter one—backed by data and user behavior.

“Statsig helped us quantify the impact of something we already sensed from customer feedback. Seeing real lift in subscriber rate gave us the confidence to build more around that feature.”

Meera Parat

Growth PM, Captions

Statsig Analytics plays a key role in validating these kinds of changes. In cases where backend events aren’t mirrored elsewhere, Statsig becomes the team’s source of truth. They often use Metrics Explorer to group metrics by gate or experiment—breaking down usage funnels, surfacing regressions, and spotting meaningful differences between test and control.

In one case, they rolled out a backend-driven quality improvement behind a gate and used feedback submission data to evaluate impact. By grouping the metric in Metrics Explorer, they saw no meaningful difference between treatment and control—even after adjusting for a 25/75 split. That clarity led to a clear rollback, saving the team from spending cycles on a feature that didn’t deliver.

Outcomes: confident growth, controlled risk

For Captions, the impact of Statsig is two fold: identifying and shipping features that lift key business metrics and protecting against shipping features that may cause unintended regression.

On the growth side, Statsig has helped identify high-leverage changes that directly grow subscriptions—whether it’s through smarter onboarding flows or paywall optimizations. It also plays a central role in the team’s weekly business reviews. Before each review, the growth team will filter for subscriber metrics in Statsig’s Metrics Catalog and use the Insights view to see which gates or experiments are moving the needle. That visibility helps shape roadmap decisions and ensures the team is investing in what’s working.

At the same time, Statsig provides a built-in safety net. The team relies on Statsig’s controlled rollouts to catch potential issues early—before they reach all users. Metrics are monitored for every launch, and features that fail to meet guardrails are paused or rolled back. The result: more product momentum, fewer mistakes, and tighter alignment to core business outcomes.

About Captions

Captions offers an AI-powered creative studio that transforms the video creation process — enabling businesses, marketers, and creators to generate fully-edited narrative videos with just a few taps. By empowering anyone to go from idea to video, Captions aims to bring enhanced fluency to subscribers around the world.