Before Optimizely, we built the testing system ourselves.
I started running A/B tests in 2001. In those days that meant building most of the machinery ourselves: the variants, the tracking, the reporting and the process around it.
When browser-based platforms made that work accessible, the interesting part stopped being the software. It became the decisions.
- 2001A/B testing becomes daily work, built by hand
- 2015Browser platforms; Optimizely joins my toolkit
- 2016Optimizely becomes our standard for client testing
- 2019Co-delivery: onboarding Optimizely’s new customers
- 2022Optimizely MVP; our packaged Web onboarding offer
- 2025Building with Opal
- 2026Agent workflows, approval gates, MCP
Timeline
Seven chapters, not every year.
The platform grew up. So did the questions customers brought to it.
- 2001
A/B testing becomes daily work
At a subscription health business, I brought testing into acquisition and retention. Every test meant building the plumbing first.
Hand-built - 2013 to 2015
Browser testing enters my toolkit
Browser-based testing tools replace the hand-built setup. I’m working across several of them, and by 2015 Optimizely is part of my toolkit. The habit of testing is what carries over.
Early platform era - 2016+
Optimizely becomes the standard
By 2016 Optimizely was the tool our CRO practice used for client testing. The work shifts from single tests to programs.
Enterprise CRO - 2019
Co-delivery
I’m part of the small team that launches a partner co-delivery program with Optimizely, onboarding its new customers alongside Optimizely’s own team.
Partnership - 2021 to 2022
Enablement at scale, then the MVP program
I lead program enablement and onboarding for Web and Full Stack customers, and by 2022 I own our packaged Web onboarding offer. I’m named an Optimizely MVP in 2022.
Enablement - 2023 to 2025
From testing tool to connected platform
Programs reach into customer data, CMS, content workflow and identity. I spend as much time on how systems connect as on what to test. I start building with Opal in 2025, the same six questions, now with an agent that can do some of the typing.
Connected DXP - 2026
Agentic systems
Agent chains with human approval gates, CMS (SaaS) builds, MCP connections and a configured Virtual Teammate. I co-present at Opticon in New York.
Agents
What never changed
The platform kept changing. The question didn’t.
- What is the business trying to change?
- What does the customer need?
- What signal can we trust?
- What should the experience do?
- How will we know?
- What should we do next?
The answer to the last question starts the next round
That’s why I never thought of my job as running Optimizely. The platform just gave me better ways to answer those six questions, and every new capability, personalization, customer data, now agents, made one answer faster or more trustworthy.
It also explains where programs stall. Skip the signal question and you personalize on guesses. Skip the last one and you run tests that never change anything.
The partnership chapter
2019 is when I stopped only running programs and started helping other teams launch theirs.
- OptimizelyCustomer success, enablement and the platform itself
- Partner deliveryDiscovery, metrics, ideation, training, experiment design and program management
- Customer teamA program they can run on their own
Optimizely was adding partner-led onboarding next to its own team, and I was part of the small group that launched it. I did the early estimating, wrote the first discovery documents and sat in the strategy seat on early kickoffs.
The work changed shape. Instead of running one team’s program, I was helping set up program after program, and every kickoff started the same way: who owns this, and what counts as a win, before anyone picks a first test.
The mechanics of that model are on the Onboarding & Enablement page.
- 2019Co-delivery launch team
- 20+Onboarding engagementsNamed resource in the co-delivery program
- 2022Owned the packaged Web onboarding offer
From tool to system
Every step up added a new question, and a new place things could break.
One example from my own work at each stage.
- Testing tool2001: I built the variants and the tracking by hand before any platform did it for me.
- Experimentation program2019: onboarding new Optimizely customers meant deciding owners, metrics and a first test worth running.
- Personalization programPersonalization campaigns inside onboarding programs, and deciding when a segment deserves a different experience.
- Data and identityFor a professional association, I checked a campaign list against membership data before we designed anything for it.
- CMS, content and workflowCMS projects where content, forms and audiences had to line up with the tests.
- Connected platformTesting tied to identity, analytics and governance across a whole journey, not one page.
- Agent workflowsA chain of narrow Opal agents for experimentation, with a person approving before anything launched.
None of those stages replaced the one before it. Agents still need a trustworthy signal, a clear owner and a metric that actually controls the decision. That’s where the next chapter picks up.
Community and credentials
Speaking, recognition and the room where the platform gets explained.
I’ve spoken at Opticon in 2018, 2019, 2025 and 2026. The 2026 session was co-presented in New York.
In 2019 I was also interviewed for Nasdaq’s Opticon19 coverage, talking about what had changed in the field: change a website gradually, so each change proves it works before the next one.
I’ve been an Optimizely MVP since 2022. For me that recognition is mostly a reminder of where the useful conversations happen: in the questions customers ask after the demo is over. Opal doesn’t change that. It helps me and the teams I work with move faster through the work, but a person still decides what the test is for and what the result means.
That history eventually put me in the room when customers had to decide what to build.
Optimizely and related product names are trademarks of their owner. This section reflects my own professional experience with the platform. It is not affiliated with or endorsed by Optimizely, and client examples are described by industry, not by name.