Why this one
The example I would pick is not the largest product I have led or the one with the biggest revenue number. I would choose Acquire because it captures how I think as a product leader. I saw a shift happening in both the technology and customer behavior, and it led me to make a pretty unpopular call: we needed to stop adding more and fundamentally rethink the product strategy. That decision changed what we built, how we served enterprise customers, and how we thought about the role of AI in customer support.
The situation
When I joined Acquire, the company was trying to be everything to everyone. United was one of our largest customers, and over time we had built so much around their requirements that roughly 80% of the functionality was effectively customized for them.
That created a scalability problem. The broader customer base was not getting a sufficiently focused product, and every custom feature did not just cost us once. It created ongoing engineering, testing, maintenance, and support costs. Meanwhile several important customers were considering churning, and investor confidence in the product direction was declining.
The obvious response was to keep adding features faster. I came to the opposite conclusion.
V2 should do less
I went back to first principles and asked two questions. What are the capabilities that every important customer actually needs? And for everything that is unique, why do we have to build and own that complexity ourselves?
Looking across our key customers, I identified five capabilities that were truly universal. So I made a fairly unpopular recommendation: V2 should actually do less. We would focus the core product around those five capabilities and create an extensible workflow layer for everything else. Instead of our engineers building and maintaining every customer-specific workflow, we would give customers the building blocks to configure those experiences themselves.
That gave sophisticated customers flexibility without turning Acquire into a custom-development organization. It also made the product much easier to build, test, maintain, and support as we scaled.
Standardize what is universal. Give customers the tools to own what is unique.
The AI thesis
At the same time I saw another shift happening around AI. Customer support was one of the first areas where I thought LLMs had an unusually clear path to enterprise value: enormous volume of repetitive language-based work, years of existing conversational knowledge, and very measurable economics around resolution, agent productivity, and cost-to-serve.
But I disagreed with one of the assumptions dominating the industry at the time, that success meant maximizing deflection and eventually eliminating the human interaction. Customer behavior was not supporting that assumption. People were still calling. Years of trying to push customers from phone into chat had not eliminated their desire to speak with a person when the situation required one.
So I reframed the question from how do we eliminate the human to how do we use AI to make the human dramatically more productive.
We already had an important asset: the knowledge and answers accumulated through existing chat interactions. With LLMs we could turn that into much more natural conversational and voice experiences, and let AI handle the repetitive questions that did not require human judgment. When someone genuinely needed a person, we did not need to fight that behavior. We could redesign the interaction so a human agent was no longer constrained by the traditional phone model of serving one customer at a time. One agent could support roughly five customers rather than one.
The thesis moved from human replacement to human leverage.
Creating conviction
The hardest part was creating conviction around these decisions while the company was already under pressure.
Internally I had to align product, engineering, and business leaders around the idea that the answer to churn was not more features. It was greater focus and a stronger platform strategy.
Externally we had customers who were already considering leaving. We could not save them by promising to build every feature they wanted. Instead I showed them where we believed customer support was going: a scalable core, extensible workflows for their unique needs, and an AI strategy that automated repetitive interactions while making their human teams dramatically more productive.
That changed the conversation. Customers saw the new direction as a stronger strategic position for them. Even before everything was built, several at-risk customers were willing to stay and wait for V2 because they believed in where the platform was going. It also gave investors a much clearer story about how Acquire could scale beyond a collection of bespoke enterprise implementations.
Outcome
We narrowed V2 around the five core capabilities, shifted customer-specific complexity into the extensibility layer, and launched the scalable new version in roughly four months.
We secured $4M in accumulated contract value that was about to churn. Customers who had been preparing to leave stayed and renewed on the strength of the new direction, before the full platform had shipped.
What it taught me
Deflection was a metric, not the problem. The real problems were customer experience, agent productivity, and cost-to-serve. Once we went back to those fundamentals we found a much larger solution space.