The Optimizely Opal AI Benchmark Report

Introduction

According to McKinsey, 78% of organizations now use generative AI in at least one business function.

But as adoption (and general AI overwhelm) rises, we’re all asking ourselves three key questions:

• How is everyone else using AI?
• How can AI create value for my organization?
• Are we currently unlocking its full potential?

This report offers insight into how Optimizely Opal is being used across our customer base and what benefits those customers are getting. With three months of usage data from nearly 900 companies, insights from early adopters, and the latest industry research, we explore how AI is actually being used and what benefits adopters are getting from usage.

Find out what’s working, what’s evolving, and what best-in-class AI marketing practices look like today.

Data sources

Our insights are drawn from three primary data sources:

• Internal Optimizely Opal data: Analyzed usage data across 47,000 Optimizely Opal interactions from customers who adopted Optimizely Opal after its public launch, covering feature adoption patterns, user behavior, and measurable performance outcomes.
• Customer stories from early adopters: Firsthand insights from nearly 900 Optimizely Opal adopters, highlighting their implementation experiences, challenges overcome, and the tangible impacts achieved.
• Industry research: Third-party research and surveys from leading firms like McKinsey and Gartner, providing external context around broader AI adoption trends and transformations in marketing.

Methodology

To ensure statistical rigor and a clean, understandable output, this analysis employed a methodology with a few key components:

Difference-in-Differences

We used a Difference-in-Differences approach to accurately measure Optimizely Opal’s impact. This method compares changes in key performance metrics between two groups:

• Optimizely Opal users: Customers who began using Optimizely Opal on or after May 7, 2025.
• Non-Optimizely Opal users: A control group of all customers who never adopted Optimizely Opal during the same time frame.

Matched time windows

For each Optimizely Opal customer, we analyze two periods:

• Pre-Optimizely Opal period: 
  • Experimentation: The same calendar window in the year before adoption (e.g., June–August 2024 if adoption was in June 2025), to account for seasonal patterns in experimentation velocity.
  • Content Marketing Platform: The period immediately prior to adoption, matched in length to their post-adoption window (e.g., if adoption was May 15, 2025 with measurement through August 15, the pre-adoption window would be the prior 90 days).
• Post-Optimizely Opal period: The window starting on the customer’s first day using Optimizely Opal and running through the present day.

For non-Optimizely Opal users, performance is measured over identical pre- and post-periods matched to each Optimizely Opal customer’s timeline. In other words, each Optimizely Opal user is compared to a group of every non-Optimizely Opal user (the control group) in the period exactly matching the Optimizely Opal user’s usage, ensuring direct comparability.

Calculating the uplift

For each Optimizely Opal customer, we calculated:

Optimizely Opal uplift (net impact) = Change for Optimizely Opal user − Change for non- Optimizely Opal users

Subtracting the control group’s improvement removes background trends—such as product enhancements, seasonal effects, or broader market changes—yielding a conservative estimate of Optimizely Opal’s true contribution.

Weighted-Average DiD uplift

To produce the headline benchmark for each metric, we calculated a weighted average of all Optimizely Opal customers’ DiD uplifts. Each customer’s uplift was weighted by their pre- Optimizely Opal activity level (e.g., number of experiments in the pre-period). This prevents the “average of averages” problem and ensures that customers with more substantial baseline data have proportional influence on the final result.

This approach provides a robust, reliable measure of Optimizely Opal’s net impact.

Key takeaways

The state of AI

• AI delivers measurable gains, with companies reporting AI-driven improvements from 30% to 91%.
• Optimizely Opal is used across the content lifecycle, led by experimentation (58.7%) and content orchestration (26.6%).
• Adoption spans 50+ countries, led by the US (41.1%), UK (16.2%), and Australia (7.6%).
• Retail (17.4%), Software (11.8%), and Financial Services (9.6%) lead adoption, with growing use in Healthcare (3.2%) and Education (3.6%).
• Companies of all digital maturity levels use Optimizely Opal—high (51.2%), medium (37.1%), and low (11.6%).
• Usage extends beyond specialist teams, reaching senior leaders, marketing, brand, creative, product marketing, customer success, and events.
• The top 1% of users engage with Optimizely Opal nearly 200 times, most often for test ideas, experiment summaries, and AI-generated text and images.

The benefit of AI

• Optimizely Opal Experimentation users achieve more impact, running 78.7% more experiments, launching 24.1% more personalization campaigns, lifting win rates by 9.3% and conclusive rate by 1.4%.
• 10.9% of experiments start with Optimizely Opal-generated ideas, 6.8% are summarized by Optimizely Opal, and 19.5% are follow-ups from Optimizely Opal insights.
• Optimizely Opal users accelerate results, boosting campaign volume 17.1% and task volume by 12.8%. It also cuts campaign completion time by 53.7% and task completion time by 15.4%, while improving content engagement by 7.4%.
• Optimizely Opal-generated content sees high adoption—52.6% of images and 89.5% of text are accepted by users.
• Experiment velocity and content marketing productivity scale linearly with days of usage and credits consumed.

Building a culture of AI

• AI leaders excel by mastering prompt engineering, embedding AI into workflows, optimizing for GEO, and tracking AI success metrics.
• The best programs use clean data, clear brand guardrails, human review, early legal engagement, and team enablement applied within an intentional governance model.
• The future of AI is in multi-agent orchestration ecosystems and generative experience optimization.

To understand Optimizely Opal's true impact, we analyzed performance data from early adopters using a Difference-in-Differences methodology. The results show clear performance improvements that scale with both usage frequency and depth of engagement across optimization and content marketing functions.

Digital optimization

Experimentation programs benefit fundamentally from either more or better experiments. The DiD analysis revealed improvements across both vectors in the headline benchmarks:

+ 78.66% created experiments
+ 2.38% concluded experiments
+ 24.05% created personalization campaigns
+ 11.97% concluded personalization campaigns
+ 9.26% win rate
+ 1.38% conclusive rate

The gap between creation and concluded uplift may stem from the limited reporting window or from early adopters using Optimizely Opal mainly for idea generation, though further analysis suggests that greater engagement leads to more concluded experiments.

Win rate and conclusive rate also rise, signaling modest quality gains. Average uplift dips slightly—a natural effect of higher velocity—but is far outweighed by the surge in volume.

Analysis of usage and funnel data shows that the best users of experimentation use Optimizely Opal throughout the entire experimentation lifecycle:

• 10.95% of experiments are created from ideas by Optimizely Opal
• 6.80% of experiments are summarized using Optimizely Opal
• 19.54% of experiments are Optimizely Opal-driven follow-ups

• The more teams use Optimizely Opal—measured by both active days of usage and credits consumed—the more experiments they create and conclude, scaling linearly with increased engagement.
• The steeper uplift slope for concluded experiments compared to created experiments suggests that increased Optimizely Opal usage is not only driving more experimentation activity, but also improving teams’ ability to bring experiments to a meaningful outcome.

Users can calculate the exact projected uplift of their programs using the equations below, where ‘y’ (the dependent variable) is the expected uplift and ‘x’ (the independent variable) is active days of usage or credit consumed.

Customer story: Unlocking experimentation at scale

One Fortune 500 financial services company dramatically accelerated their experimentation program after adopting Optimizely Opal. In the year before adoption, their experimentation volume was modest due to switching systems. But in the following year— with access to Optimizely’s advanced experimentation platform and the added power of Optimizely Opal AI—their team increased experiment velocity by nearly 80x.

This rapid ramp-up showcases how combining robust experimentation infrastructure with AI-driven acceleration unlocks scale and impact that simply wasn’t possible before.

Content orchestration

Similar to experimentation programs, content marketers create more value by either producing more or better revenue-generating content. Companies can also realize value by time or labor savings when marketers become more efficient. The DiD analysis shows benefit across all three of those value drivers:

• + 17.05% campaigns created
• + 12.82% tasks created
• - 53.73% time per campaign
• - 15.41% time per task
• + 7.41% engagement time

The sharp reduction in campaign and task completion times—paired with higher output volumes—shows that Optimizely's Content Marketing Platform (CMP) users are achieving genuine workflow acceleration, not just working faster. At the same time, increased engagement time suggests teams are using Optimizely Opal to produce higher-quality content that holds visitors’ attention and delivers greater value.

We found these same trends reflected in the funnel data. Content created by Optimizely Opal is getting used by marketers more often than not:

• Optimizely Opal-generated content (both images and text) is being actively adopted by users, not just tested or used for brainstorming.
• Text generation is particularly successful, with almost all generated outputs being used.
• Image generation adoption is strong but with room for further improvement as LLMs ability to create images catches up to their ability to create high-quality text.

Like with experimentation, the productivity benefits of Optimizely Opal usage in content marketing scale with usage:

• The more teams use Optimizely Opal—measured by both active days of usage and credits consumed—the less time they take to complete tasks and campaigns, with time savings scaling consistently as engagement increases.
• The steeper negative slope for time per campaign compared to time per task suggests that Optimizely Opal usage is driving especially large efficiency gains in campaign delivery, freeing up resources for additional creative and strategic work.

Users can calculate the exact projected uplift of their programs using the equations below, where ‘y’ (the dependent variable) is the expected uplift and ‘x’ (the independent variable) is active days of usage or credit consumed.

Customer story: Accelerating content performance at scale

A leading financial services organization rapidly accelerated its content operations after adopting Optimizely’s Content Marketing Platform (CMP) with Optimizely Opal AI. In the period before adoption, campaign creation and completion were modest. But in the following months—powered by AI-driven workflow acceleration—the team achieved a 43.22% uplift in campaigns created and an over 3× increase in campaigns concluded.

This surge demonstrates how combining a unified content platform with AI can unlock new levels of scale and efficiency, enabling teams to deliver more high-quality campaigns in less time.

Without clear frameworks, teams face inconsistent outputs, brand risks, and compliance gaps that can quickly derail AI initiatives.

Here’s what to establish up front, along with a governance framework you can follow:

Essential building blocks

A little groundwork upfront can go a long way to improving the long-term outcomes of AI usage. Here are some of the recommended first steps:

• Start with clean data: Audit content repositories and databases before scaling AI deployment. Clean, current data is the foundation of reliable AI output—garbage in, garbage out. Providing a single source of truth—such as an approved dataset or integrated live data—ensures your AI outputs remain accurate and valuable.
• Establish brand guardrails: Use custom fine-tuning to embed brand voice and content guidelines directly into AI behavior. Define acceptable parameters upfront to prevent off-brand outputs.
• Keep humans in the loop: Implement review workflows where experienced marketers approve AI-generated content, especially for customer-facing communications.
• Partner with legal early: Establish clear guidelines around AI content usage, IP rights, and data privacy. Stay ahead of evolving regulations to protect your brand.
• Build team capabilities: 62% of marketers lack formal AI training. Bridge this gap with practical prompt engineering education and develop “AI champions” to drive adoption.

Designing your governance model

There’s no one-size-fits-all AI governance model. Here are key factors to consider when designing the optimal governance model for your business:

• Control: Who can deploy AI tools and approve outputs? Ensure teams have mastered AI fundamentals—prompt engineering, quality assessment, brand alignment—before expanding access across the organization.
• Capabilities: Do you have the right tools and resources for proper oversight? Consider your team's technical expertise, available budget for AI platforms, and bandwidth for training. Start simple and scale capabilities as your team matures.
• Risk tolerance: How much risk can you accept from AI errors or off-brand messaging? High-stakes customer communications require stricter approval workflows, while internal efficiency tools can operate with lighter oversight.
• Connection to business priorities: Is your AI governance aligned with core business objectives? Avoid creating AI governance in isolation. Ensure your framework supports broader marketing goals and can adapt as business priorities shift.
• Data sensitivity: What sensitive data will your AI systems access? Build governance controls that scale with data sensitivity. Customer data requires stricter access controls than general marketing content.

Key Definitions