How Anaconda Turns Experimentation Into a Growth Engine

Most marketing campaigns still run on a dangerous mix of gut instinct and “what’s worked before.” That’s fine—until budgets tighten, customer expectations evolve, and leadership starts asking for proof. Intuition alone doesn’t cut it anymore.

Marketers need a way to validate every decision, tie activity to outcomes, and show their impact on the business’s bottom line. The fastest path? Treat marketing like science.

At Activate Summit 2025, Anaconda proved just how powerful this shift can be. By adopting a structured, data-driven framework for experimentation, it replaced guesswork with repeatable wins backed by statistical confidence.

The Scientific Method for Smarter Marketing Decisions

In many organizations, marketing success is still judged by opens, clicks, and CTRs—metrics that may look promising on a dashboard but tell only part of the story. Without a direct connection to outcomes like revenue, activation, or retention, even the best-looking campaign reports can leave leadership asking: So what?

Anaconda, an open source platform for data science and AI distribution, tackled this gap by bringing the rigor of the scientific method into their marketing practice—a structured approach that ties every test to KPIs that actually move the business forward. Their “experimentation starter pack” breaks the process into four repeatable steps.

1. Define Your KPI Funnel

Start with the end in mind. Map marketing metrics, like email clicks, site visits, and engagement rates, directly to business-critical outcomes, such as activations, subscription upgrades, or revenue growth. This ensures that every experiment supports marketing priorities and company-wide goals.

2. Craft a Strong Hypothesis

Frame your test in the if/then/because format to keep it focused and measurable. For example: If users receive onboarding emails every two days, then they will activate faster because more frequent touchpoints keep them engaged.

3. Build an Experiment

Test one variable at a time, and always include a control group. AI tools like ChatGPT or Claude can help you generate fresh ideas, but the key is keeping each experiment focused on a single, clearly defined change—so you know exactly what’s driving the results.

4. Measure Your Impact

Analyze results against your defined KPIs, even if the lift isn’t statistically significant. Each iteration should produce actionable learning.

“If you’re strategic about the metrics you want to move and the hypothesis you want to test, you can learn and get better with each successive test … even if the metrics you’re trying to move don’t actually move from that specific experiment.”

~ Dean Iwaoka, Head of Growth at Anaconda

The Onboarding Experiment That Boosted Activations by 20%

The challenge: Anaconda’s data revealed a critical friction point: A significant share of new users never activated their accounts after signing up. Left unaddressed, this gap risked slowing adoption and weakening long-term retention.

The hypothesis: If users receive onboarding emails, then they will be more likely to activate their accounts faster, because more frequent touchpoints keep them engaged.

The experiment:

• Group 1: Standard onboarding series over 30 days (nine emails)
• Group 2: Accelerated onboarding over 14 days
• Control: No emails

The result: Both onboarding sequences delivered clear wins, driving an approximate 20% lift in activation rates over the control group. More importantly, the experiment proved that email wasn’t just generating surface-level engagement. It was moving a business-critical metric and reinforcing the channel’s role in driving product adoption.

“We showed approximately 20% lift in our activation rates…both versions of the email series were better than the control group or receiving nothing.”

~ Dean Iwaoka, Head of Growth at Anaconda

Powering Experiments with a Composable MarTech Stack

Running one test is easy. Scaling hundreds of them quickly across multiple channels requires infrastructure. That’s where a composable martech stack comes in.

Anaconda’s approach uses Hightouch as a composable customer data platform (CDP) and Iterable as the orchestration layer for cross-channel messaging.

Here’s how it works:

• Audience building: Hightouch connects directly to Snowflake, giving marketers self-serve access to audience creation and segmentation—no data scientist required.
• Experiment execution: Randomized cohorts are created in Hightouch and synced to Iterable for coordinated email and in-app messaging and to paid media platforms for campaign consistency.
• Measurement: All performance data flows back into Snowflake for a single source of truth and deeper analysis.

The result is a dramatic acceleration of the testing cycle. What once required a data scientist weeks to engineer through a manual pipeline, Anaconda’s team can now execute in minutes. Infrastructure that was once a bottleneck is now a growth enabler.

“When I was at Meta, we would need a data scientist or data engineer to actually build a manual data pipeline to power this kind of experimentation—that could take weeks to do. Now my team can do it hands-on-keyboard in a number of minutes, and it saves us a lot of time in all the rinse cycles that we’re doing on experimentation.”

~ Dean Iwaoka, Head of Growth at Anaconda

See how Anaconda is making it happen in the full session recording Activate 2025.

AI Decisioning: The Next Leap in Personalization

Traditional A/B tests at the cohort level deliver valuable insights, but they’re inherently limited. They reveal what works on average, not what works best for each individual customer. The next leap in experimentation is AI Decisioning: real-time, always-on personalization at the individual level.

Dean Iwaoka and Alec Haase pointed to a compelling example from Whoop, which achieved a 10% sales lift after AI decisioning uncovered a hidden behavioral pattern in its user base. When every interaction becomes a micro-experiment, unexpected revenue opportunities emerge.

This is the new era of experimentation, where AI agents uncover and deliver the most effective experience for each customer, at a scale no human team could match.

Why Experimentation Wins (and Keeps Winning)

Anaconda’s success proves the value of replacing one-off campaigns with ongoing, structured experimentation. Decisions get faster, insights sharper, and metrics directly tied to revenue, activation, and retention. Each test builds on the last…but scaling that impact takes more than mindset—it takes the right technology. 

With a composable stack powered by Hightouch and Iterable, marketers can move from idea to execution in minutes, run disciplined tests that uncover deeper insights, and deliver experiences so relevant they keep customers coming back.