How to wow in first 90 days as product and experimentation leader

The first 90 days in a new product or experimentation role feel like drinking from a fire hose while someone yells "What's the revenue impact?" in your ear.

However, this window is your greatest advantage. You have fresh eyes and full permission to ask hard questions. Your CEO, CPO, CTO, and engineering teams want to hear your perspective.

Here are five high-impact ways to wow early and build credibility fast.

1. Understand the business and assess your experimentation maturity

Before launching anything, ensure you understand everything. Your first 30 days should feel like detective work: "What are we testing, why, and does anyone actually know?"

Line up conversations with:

CEO: What's the growth strategy? What metrics matter most?
CFO: How do they measure ROI? What's the bar for "this experiment is worth it"?
CTO/VP Engineering: What's the technical foundation? What's slowing teams down?
Product leadership: What features are validated vs. hunches?
Data/Analytics: What can you measure reliably? Where are the blind spots?

What you're really trying to figure out:

Where does your team stand on velocity? The median company runs 34 experiments a year. Top performers run 200+. But more tests don't automatically mean better results. Impact per test peaks at 1-10 annual tests per engineer. Beyond 30, theexpected impact drops by 87%.

Cultural readiness: Do teams ship to validate or ship to release?

If you're not testing, there's a really good chance that most of what you're doing is a complete waste of time and money.

Also, can you access cross-channel data and tie experiments to revenue, retention, and customer lifetime value, or are you stuck celebrating engagement metrics?

2. Audit your infrastructure, metrics, and velocity blockers

Most teams measure what's easy, not what matters. They track clicks, page views, and form submissions because those are simple to instrument. However, these metrics often fail to accurately predict revenue.

Before you launch anything new, get clear on what you're actually working with. Here's what to look for:

Team and ownership:

Who owns experiment implementation? Is it centralized or spread across product teams?
Where are the skill gaps? (Statistics? Front-end dev? UX research?)
Who's the unofficial "testing hero" running everything manually (and quietly burning out)?

Processes and workflows:

How long does it take to ship an experiment from idea to live?
How many approvals does one A/B test require?
Can teams deploy experiments on their own, or are they stuck in approval purgatory?

Technical infrastructure:

Can you run server-side tests, feature flags, and edge experimentation?
Does your experiment data live in one tool while revenue data lives in the warehouse, requiring days or weeks to reconcile?
Are you tracking metrics that actually predict revenue, or just vanity metrics that are easy to instrument?

What you'll discover:

A test running for months with no owner
Different teams tracking "conversion" in different ways
No clear way to tie test results to actual business outcomes
A platform no one uses correctly (or at all)
A "temporary" workaround from 2 years ago that's now mission-critical

Next, most teams can't answer leadership's #1 question: "What's the revenue impact?"

This is your moment to set standards and build infrastructure that supports both speed and rigor. Simplify, consolidate, and introduce systems that let you connect experiments directly to business outcomes without waiting weeks for data teams to reconcile spreadsheets.

3. Deliver quick wins that prove experimentation drives outcomes

Quick wins build credibility and buy you runway for bigger changes.

Fix broken tracking: Nothing builds trust faster than showing teams they've been making decisions on unreliable data, then fixing it.

Connect experiments to revenue: Be the first person to answer, "What's the revenue impact?" Use warehouse-native analytics to tie test results directly to business metrics.

Run one high-impact test fast: Virgin Media went from 40 tests/year to 600 by building infrastructure that lets teams ship experiments on their own. Start with one fast win.

Kill a sacred cow: Find a "best practice" everyone assumes works and test it. When it loses, you've proven the value of evidence over opinion.

Build a real-time results dashboard: Stop making analysts build custom reports for every experiment. Self-service analytics means decisions happen in days, not weeks.

Use AI to generate test ideas: Show the team what's possible with AI experimentation. Analyze your site, user behavior, and past test results to surface high-impact test ideas in minutes.

4. Build alignment with engineering, data, and product

The best experimentation leaders build partnerships. When these relationships work, the impact of a testing program increases. When they don't, you're stuck in approval purgatory.

Experimentation + engineering:

Can we run server-side tests, feature flags, and edge experimentation?
Who owns test implementation? How fast can we ship?
Can teams deploy experiments on their own, or do they need approval?

Experimentation + product:

No feature ships without a hypothesis and success metrics
Define what "validated" means
Build a culture where "ship to learn" beats "ship to ship"

Experimentation + data/analytics:

One source of truth for metrics definitions
Build warehouse-native analytics so anyone can analyze experiments without waiting for data science resources
Traditional platforms force you to export data and wait days for business impact analysis. Warehouse-native solutions keep all your data in one place.

Experimentation + leadership:

Connect experiments to OKRs and revenue targets
Report on learnings, not just winners
Advocate for speed and autonomy over bureaucracy

5. Define success metrics, then build the experimentation engine

Stop measuring "number of tests run." Focus on outcomes.

Key metrics:

Experiment velocity: Tests shipped per quarter (quality over quantity)
Win rate: Only 1 out of 10 experiments wins. So, focus on uplifts per test.
Revenue influenced: How much revenue ties directly to validated experiments?
Time to insight: Days or weeks?

Build the infrastructure:

Use PIE (Potential + Importance + Ease) to score ideas. This prevents teams from wasting resources on low-impact tests.

Democratize experimentation: Tests with 4+ variations are 2.4x more likely to win and deliver 27.4% higher uplifts. But 77% of experiments test only two variations. Why? Creating variations requires developer resources. The solution: Options like Opal AI that let non-technical teams create and ship experiments.

Advanced testing methods: Move beyond basic A/B tests to multivariate testing, server-side testing, feature flags, contextual bandits, and CUPED.

Warehouse-native analytics: Track any business metric (revenue, retention, customer lifetime value) directly in experiments. No data exports. No reconciliation. No waiting weeks for data teams.

Your 30-60-90 roadmap

30 days: Map current state, align with leadership, run one quick win, identify velocity blockers
60 days: Fix tracking gaps, align teams on workflows, establish experimentation rituals, launch 3-5 meaningful experiments
90 days: Publish your strategy, set clear KPIs, build systems for high-velocity testing, launch something that makes executives ask, "How can we do more of this?"

Want help accelerating your first 90 days?

Optimizely One is an end-to-end platform built especially for product and experimentation leaders. What you get:

Autonomous experimentation (Optimizely Opal) that removes manual work while keeping leaders in control of what ships.
Unified experimentation, personalization, and analytics connected to a single, trusted data foundation.
Continuous learning that surfaces what matters most and guides what to improve next.

The result?

Customer experiences improve continuously. Programs scale without you losing control. And you can prove impact to leadership with one system of record.