Ryan Tamminga, Chief Customer Officer, Alchemer – Interview Series

Ryan Tamminga is a customer experience and enterprise transformation executive with nearly two decades of leadership experience spanning customer success, professional services, consulting, and product strategy. Since joining Alchemer in 2019, he has advanced through a series of leadership roles, including Vice President of Customer Success and Senior Vice President of Product & Services, before being appointed Chief Customer Officer in 2026. In his current role, he oversees customer engagement, success, and service delivery while helping align customer needs with product innovation and business strategy. Prior to Alchemer, Tamminga held senior leadership positions at ReedGroup, Deloitte, and Accenture, where he led large-scale transformation initiatives focused on business operations, analytics, automation, and enterprise performance improvement.

Alchemer is a customer feedback and experience management platform that helps organizations capture, analyze, and act on insights from customers, employees, and stakeholders. The company’s software combines survey creation, feedback collection, workflow automation, and AI-powered analytics to help businesses move beyond simply gathering data and instead drive measurable action from it. Serving organizations across a wide range of industries, Alchemer enables teams to centralize feedback from multiple channels, uncover trends and sentiment, and use those insights to improve customer experiences, inform product decisions, and strengthen business performance. Through its growing suite of AI-driven tools, the company aims to help organizations transform feedback into a continuous source of operational and strategic intelligence.

You’ve spent more than a decade in consulting at firms like Accenture and Deloitte before leading customer success, product, and services at Alchemer. How has that mix of process transformation and hands-on customer experience shaped your approach to deploying AI in enterprise environments?

Working in consulting provided very valuable experience that informs my current role at Alchemer. For example, I discovered that technology is rarely the most difficult part. I spent years helping Fortune 100 companies implement enterprise systems, and the projects that struggled weren’t undone by bad software. They had change management problems, unclear ownership, and often were deployed before the underlying processes were ready to absorb them.

That experience shapes how I think about AI deployment today. When I talk to customers who are struggling to get value from AI, the issue almost never turns out to be the technology      itself. It’s that they haven’t defined what they’re trying to accomplish, who owns the output, and what changes when the AI surfaces something actionable. The technology is ahead of the organizational readiness in most cases.

The hands-on customer work is the other half of that equation. You learn quickly what value  actually means to the people doing the job every day and how to translate that to what it means for board metrics. Those are often very different things. The combination of process rigor from consulting and customer empathy from working alongside teams is what I aim to bring to the strategy of how we build and deploy AI at Alchemer.

You’ve pointed to a significant AI maturity gap across organizations. What are the biggest misconceptions leaders have about their readiness, and where do implementations typically fail first?

The most common misconception is that buying the AI tool is the same as being ready to use it. Leaders see a demo, they see the output, and they assume, often incorrectly, that adoption follows automatically. It is important to ensure that the technology, whether AI or not, starts with the business problem that needs to be solved. The successful adoption of the technology is almost always tied to how well you can communicate how it will solve that problem for your teams and customers.

Where implementations fail first is almost always at the handoff point. If there’s no defined workflow for what happens next (e.g., who sees it, who acts on it, what system it goes into) then      the insight isn’t acted upon. Organizations need to be structured properly to act on insights.

Recognizing this gap has shaped how we build solutions at Alchemer. We start by identifying the      problems we are looking to solve with the technology capabilities we release. AI capabilities, workflows, and user controls are only part of what we deliver. Helping teams build the organizational muscle to act on feedback is just as important, and it’s where real outcomes are realized.

Many companies are investing heavily in AI but struggling to demonstrate ROI. What metrics or frameworks should they be using to evaluate whether AI in customer experience is actually working?

Start with what changed in behavior, not what the AI produced. The right question isn’t, “Did AI generate a summary?” Instead, ask questions like, “What metric is the key indicator of success?”, “Did the team act differently because of it?”, and “Are the aligned business metrics shifting in the right direction?” AI can clearly identify where a problem or opportunity exists and provide new insights. However, if companies don’t act on those insights, customer feedback may be ignored or not addressed in a timely manner.

Time-to-insight is a good metric to start with. How long did it take to go from collecting feedback to having something actionable in front of a decision maker? Feedback analysis that used to take six months is significantly reduced with AI. We have a customer who reduced that cycle from six months to hours, demonstrating a real, measurable shift towards solving time to insight and action.

Time-to-response is another metric to capture early on. Customers expect a response to their feedback, especially when it’s negative. Measure how long it takes to respond to a negative survey response or online review. It might take days if you are reviewing manually, assigning action via a support ticket, then responding to the customer. An online eyeglass retailer was able to go from nearly a month to minutes.

Response rate is the final metric to track. H&R Block Canada manages nearly 1,000 locations through tax season. Before AI, getting to 100% review response coverage was nearly      impossible. Now it’s a baseline. That’s measurable, and the downstream effects on search visibility and customer perception are trackable.

Get started by identifying the problem that’s costing you or frustrating your customers (e.g., slow analysis, low response rates, missed customer signals) and measure the delta before and after. Don’t try to measure everything. Measure one thing that matters and use that momentum to help build and deliver the business case.

From your perspective, what separates organizations that successfully operationalize AI from those that remain stuck in experimentation?

The organizations that succeed with AI treat it as infrastructure rather than as a project that           has a start date, an end date, and a team trying to justify the investment. Infrastructure becomes core to how work gets done. For example, think about the difference the introduction of CRM software made in the late 1990s and early 2000s.  The project was the implementation, but the systems became core operating infrastructure that were the foundation for how go-to-market teams have been operating ever since. That transition is what operationalizing actually means, and most organizations haven’t made it there yet when it comes to AI adoption.

The other differentiator is ownership. Successful customer experience (CX) deployments have someone whose job it is to make sure outputs are acted on. There needs to be someone accountable for what happens because when no one owns the outcome, insights loop back to a dashboard and stop there. This has always been true for CX programs — it’s amplified by AI because the outputs are coming faster, and expectations from customers are accelerating.

The third thing I watch for is whether the organization has had a pattern of consistency of success metrics. What gets measured, gets managed. Teams tracking trends over time, benchmarking performance quarter over quarter, or comparing results across geographies can’t afford AI solutions that produce different answers on different days. The organizations that operationalize successfully tend to demand reliability in addition to capability. They want AI they can build upon.

There’s a lot of excitement around general-purpose LLMs, yet you advocate for purpose-built AI in feedback workflows. Where do general models fall short in enterprise use cases?

Not every problem requires the same AI solution. Many vendors in the feedback and CX space have built their AI capabilities on top of commercially available general-purpose models like ChatGPT, Claude or Gemini. They are designed to do everything for everyone, and that generality can be problematic for organizations that demand a high level of reliability and consistency.

With that in mind, Alchemer has taken a different approach to support the challenges our customers are looking to solve. Alchemer uses the right AI solution for each specific task rather than routing every problem through commercial LLMs. The result of this purpose-built AI strategy is more accurate outputs, consistent results over time, and AI that is optimized for feedback data rather than adapted from a tool built for something else.

We’ve seen this play out directly with customers. Washburn & McGoldrick evaluated general-purpose AI tools before choosing Alchemer and found that the same dataset produced different categorizations on different days. You can’t build a benchmarking program on that.

What does it actually mean to embed AI directly into business workflows, and why is that approach more effective than treating AI as a standalone tool?

A standalone AI tool is something you open when you decide to analyze something. An embedded AI capability is already working before you’ve thought to ask.

Here’s the difference in practice: If a review comes in overnight that hits a risk threshold because of a safety concern, an embedded system triggers an alert, routes it to the right person, and initiates a response workflow automatically. No one has to remember to check the dashboard. The AI is already part of the process.

At Alchemer, we think about this across all feedback channels and the full feedback lifecycle. AI in our survey capabilities improves what comes in and can generate relevant follow-up questions in real time so a survey becomes a conversation. In review management, AI can draft on-brand, personalized responses and even post it. AI in our analytics layer surfaces what matters across all that feedback. And our workflow automation connects AI-triggered actions directly into the business systems where teams actually work. When those pieces are connected, the gap between insight and action shrinks from days to minutes. That’s what embedded actually means, wiring the actions stemming from customer feedback into the systems that our customers’ teams use every day to engage their customers.

Turning unstructured customer feedback into real-time, actionable intelligence sounds powerful, but difficult. What are the biggest technical and organizational challenges in making that work at scale?

On the technical side, the volume and variability of the data can genuinely be challenging. Customers write in different languages, with abbreviations, misspellings, and shorthand that general models frequently misread. Models also have to understand the language of the business. Different industries use different terminology and businesses apply their own nuances on top of that.

For example, the person who greets you at a business might be a receptionist at a doctor’s office, the host at a restaurant, and a barista at a coffee shop. Those similar roles can be named differently in different industries. While a general model might be ok for an initial review, the underlying models have to be purpose built for the nuance of feedback data and how customers talk about specific industries, products, brands, and services. The models also have to be consistent, because you’re almost always comparing against historical baselines.

While the technical challenges seem to be increasing, the organizational challenges, while significant, are becoming easier to solve. The first major challenge is knowing what to do with increased volume and richer insights. Most teams look at an AI-generated insight and say, “that’s interesting.” The best organizations build workflows that say, “This insight goes to this person/system, who does this thing, within this time window.” Fortunately building those workflows has never been easier. With some planning, and when done right, the acceleration of the general curiosity of the teams working to understand customer feedback becomes really exciting as it evolves.

Another major organizational challenge is trust. In Alchemer’s recent study only 29% of CX software buyers said they’re currently comfortable acting on AI-generated outputs without review. It reflects real experience with AI that’s been inconsistent or unexplainable. Building trust requires AI systems that are transparent about how they reach conclusions, with audit trails and configurable controls that let teams decide what the AI can and can’t do. At Alchemer, we treat trusted AI as a product, not a feature.

You’ve suggested that consistency can matter more than accuracy in AI-driven market research. Can you explain why consistency is so critical and why it’s often overlooked?

Accuracy tells you how well the AI understood a single response. Consistency tells you whether you can trust the comparison over time. For market research, the comparison is the point. It’s not meaningful to know what customers say today in isolation. But it is valuable to understand whether things are getting better or worse compared to last quarter, how one region compares to another, or whether the themes you’re seeing now were present six months ago. None of that is possible if the underlying classification shifts between runs.

For example, if you hired two different analysts to code the same open-text feedback six months apart, you’d have a comparability problem even if both were excellent. You wouldn’t know whether a shift in themes reflected a real change in customer sentiment or just a difference in how two people interpreted the same data. AI with inconsistent outputs can cause this same challenge.

Fine-tuned models that apply the same classification logic every time solve this. The model doesn’t produce different answers on different days. It makes open text answers reliable in the same way that longitudinal Net Promoter Score (NPS) programs are reliable. It’s what allows an analyst to tell the business something meaningful about where they’re headed, not only where they are right now.

As AI becomes more embedded in customer experience, how should organizations think about it as a workforce multiplier rather than a replacement, especially for non-technical teams?

The workforce multiplier framing shifts the question from “what will AI replace?” to “what can my team do now that they couldn’t before?” That’s a more productive frame, and in my experience, it’s also the more accurate one. It tracks historical patterns of large-scale advancements in technology adoption over the last twenty years. The concrete version: a customer insights analyst spending three days a week manually coding open-text feedback can now spend that time interpreting patterns, presenting findings, and working on the questions behind the questions. The AI didn’t replace the analyst. It gave the analyst’s skills more room to operate.

This matters even more for non-technical teams. When anyone can ask questions about their data in plain language without needing a data scientist or building a report from scratch, the people closest to the customer get to the insights faster. That changes the pace of decisions across the whole organization, not just the efficiency of a single task.

The multiplier effect only materializes if teams are prepared to use the capacity that AI creates. That’s an organizational design question as much as a technology question. It’s why we spend as much time on adoption and best practices with our customers as we do on the capabilities themselves.

Looking ahead, how do you see AI reshaping customer experience over the next few years, and what should enterprises be doing today to prepare for that shift?

The shift I’m watching most closely is the move from reactive to proactive. Most feedback programs today are reactive. For example, something happens, feedback comes in, teams analyze it, and decisions get made. The cycle is getting faster, but it’s still fundamentally backward-looking.

What AI makes possible is getting ahead of that curve. Identifying signals early enough to act before a problem shows up in your scores. Knowing which customer segments are at risk before they churn. Understanding why satisfaction is dropping in a specific region before it becomes a pattern. That’s where the combination of purpose-built AI and longitudinal feedback data becomes genuinely powerful.

Enterprises today should build the organizational infrastructure to absorb what AI makes possible. Consolidate your feedback data from reviews, surveys, social media, in-app, and more. Define who owns the analyses and what happens when something actionable surfaces. Build the workflows that connect insight to action before you need them. The companies that will pull ahead aren’t necessarily the ones with the most sophisticated AI technology, they’re the ones who act on AI insights consistently, quickly, and at scale.

Thank you for the great interview, readers who wish to learn more should visit Alchemer.