Qualtrics X4 - the harder questions about agentic AI and experience context

Qualtrics CEO Jason Maynard's keynote at X4 this week was confident and certainly got attendees thinking about the art of the possible, when it comes to experience context driving actions and outcomes. However, some of the more interesting details came away from the main stage, during conversations I had with Brad Anderson, President of Product and Engineering at Qualtrics, and Mark Hammond, SVP of Core AI at Qualtrics. Both provided me with very candid insights into the opportunity ahead for the vendors, as well as the challenges facing itself and its customers. 

Qualtrics clearly knows where it wants to go, but it is also honest about the complexity of getting there. 

The underlying idea is that Qualtrics is moving from experience measurement to experience intervention - insight to action. The insight gap is closed, as Maynard has been saying since he took the helm. The new gap - between understanding and outcomes - is where the opportunity sits. But what is the reality of getting organizations that have been in ‘listening mode’ for years, shifting towards ‘action’? 

The education problem is the hardest problem

Brad Anderson wasn’t naive about the education ahead. Qualtrics has a really dedicated customer base that has been using the vendor’s platform - largely via surveys - to understand what employees and customers think and feel. Listening, with data mostly being analyzed in a rear-view mirror, is very different to hearing what a customer has to say and using agentic AI to act on it in real time. It might sound like a subtle technology difference, but as we will discuss, it has implications across a number of areas that include ROI, go to market, and changing operating models. When I asked Anderson how Qualtrics is managing the shift internally and with customers, he said:

It's our hardest challenge. The change management is everything, from whether our field understands it all the way down to whether users of Qualtrics inside a customer account understand how to quantify the financial and business outcomes.

The technology, as Anderson put it, already supports what Qualtrics needs to do. The hard part is human - and it runs in two simultaneous directions.

Inside Qualtrics, the go-to-market function needs to have a fundamentally different conversation than the one it has been having. Selling an experience management subscription that generates insight is one thing. Selling a system that intervenes in real time, that prices agentic interactions at a significantly higher rate - Anderson acknowledged the pricing model is "still being iterated on" - and that needs to demonstrate measurable business outcomes to justify the spend, is another proposition. And that’s a big shift.

With customers, the challenge is even more present. Mark Hammond was candid that Qualtrics has historically been high-touch. He said:

We've helped customers progress with people more than with product.

The ambition now is to bake progression guidance into the product itself - self-guided maturity advancement without hand-holding. Hammond added:

We're increasingly baking that guidance into the product itself. Part of what we hear when people ask what they want is more and more demand for turnkey, self-guided progression - people want to advance without needing their hand held through it. So we're investing heavily in those areas, though it's a more recent area of focus.

That's the right direction, but it's not embedded yet. In the meantime, there is an enormous distribution of customer maturity sitting underneath the keynote narrative. From organizations still trying to get basic survey adoption, to TruGreen, which deployed Experience Agents to address 51% of customer concerns in the first week and reduce escalations by more than 30% (although, Qualtrics too was very hands on with this project). Getting from one end of that distribution to the other is not a product problem alone.

The integration reality

Qualtrics' central argument - that experience context is the missing ingredient for effective agentic AI - only holds if that context can actually flow into the systems where decisions get made. This is another challenging area for Qualtrics. Understanding that a customer is frustrated is useful. Understanding that frustration inside a ServiceNow ticket workflow, a Salesforce interaction, or a healthcare navigation agent is what the "system of decision" ambition actually requires. 

Anderson told me Qualtrics customers currently connect to around 25 systems of record on average, with over 300 out-of-the-box connectors available. That's a solid foundation, but the move to agentic architectures needs more than API connections. Anderson did share with diginomica that the company's current belief is that MCP will be the protocol agentic interactions standardize on, and Qualtrics is releasing its first MCP server next week, putting an MCP interface across its full capability set. That's a meaningful technical step forward and one that I’m sure will be well received by customers.

But the barrier I picked up from speaking with customers wasn't technical. When Qualtrics agents identify something requiring action in an operational system, the question of who owns that system - and whether they'll permit it - becomes the blocker. Anderson does understand this and said: 

Human nature is to protect what you have control over. You have to offer a carrot, not a stick. The question to ask is: if we do this, here are the kinds of insights, findings, and recommendations I can bring back to you.

He outlined the barriers as three speed bumps he sees in every early agentic implementation: whether the data is prepared to train agents; whether the necessary integrations exist; and what internal approvals are required. Start with the scenario that minimizes all three, he said - demonstrate value, then build. That's coming from actual implementation experience on Anderson’s part, and it's the kind of practical detail that doesn't always make it into keynote presentations.

The ROI conversation is changing too

Shifting from insight to outcomes doesn't just change the product - it changes what success looks like, and who has to sign off on it. I asked Anderson how Qualtrics is structuring that conversation with customers now that an NPS score improvement is no longer the primary currency. For instance, I heard from customers this week that are using experience agents to drive revenue and also improve efficiency - in turn changing their operating models as a result (I’ll write up case studies on this next week).

Retention is the entry point though, according to Anderson. He said:

Every customer has a very different leaky bucket problem. The easiest conversation is: if we can help you build a real connection with your customers, and you can prove your retention rate is improving, that's the fastest way to demonstrate ROI. You've already got these customers - you just want to retain them. You're not going out spending on customer acquisition costs to bring new customers in.

The second ROI conversation - converting pipeline, turning prospects into paying customers - is harder, Anderson acknowledged. That's a longer sales cycle, a more complex proof point, and a different buyer to convince.

What sits underneath both conversations is the pricing shift. Agentic interactions, as Anderson noted in this week’s media briefing, are significantly more expensive than non-agentic ones. The commercial infrastructure for this new model is still being built. That’s fair and I appreciate the honesty from a vendor in the midst of change, but it also means Qualtrics is asking customers to make a leap of faith on ROI at precisely the moment when the pricing model itself is still being worked out. Customers will - and should - certainly have an eye on that dynamic.

Different buyers, different businesses

Hammond raised something I think is under-examined in the broader Qualtrics story. As the vendor moves towards truly in-the-moment intelligence, he revealed that the vendor is exploring experience context embedded in an app, not only via survey fired after an interaction. This, by his own admission, is a different business. He said:

That's a different buyer. It's a different end user. You have a developer profile instead of an experience management professional profile. So a lot of those things end up being net new businesses.

Qualtrics is running incubation programmes to figure out the buying motion, channels, and value propositions for these scenarios. Hammond wasn't pretending the answers exist yet. He said:

We don't yet have a full understanding of who the buyer is, what the channels are, what all the value propositions and use cases are.

This connects to a question I raised in yesterday's piece - whether Qualtrics can get into the rooms where enterprise-wide agentic architectures are being designed, when its historical buyers may not carry the organizational weight to mandate that experience context be central. Hammond's framing was cautious. He'd rather say experience context "is an important aspect" than the most important one. Anderson, however, said that smart organizations are prioritizing it: 

You can actually tell a lot about the maturity of a company's culture by looking at whether the CIO is involved in customer experience or not. If things are just being stood up without any centralized governance process, the CIO is not involved, and you end up with programmes around the company that are silos and one-offs. Once a company has said, "We need one listening system," the CIO becomes involved - and their decision on what that listening system should be is based on capability, but also on what I call the "ilities": reliability, security, and so on.

So one of the things I always look for is: if the CIO is involved in the decision, I know the culture is one of standardization, governance, and consistency.

What I often say to CIOs is: listening systems are being stood up around your company without common governance or common security. If we can help you understand the impact on the business - through churn reduction or increasing retention - there's an opportunity for you to show your value to the company by helping it standardize on one listening system and one set of AI models able to consistently interpret all the signal coming in.

That's appropriate, but I’d urge Qualtrics to think this through more thoroughly - the horizontal platform vendors continue to assert their own context claims and they may currently have more sway in those discussions. Relying on mature organizations with savvy CIOs is great, but the messaging around experience context being crucial needs to be heard far and wide - well beyond individual CX or EX groups. 

Another opportunity? 

One other interesting exchange I had across both interviews had nothing to do with product announcements.

It occurred to me during the event that Qualtrics possibly has an opportunity in front of it that it isn’t taking advantage of. I put the same question to Anderson and Hammond: given that Qualtrics' core competency is understanding human experience, and given that AI is provoking something close to an identity-level reaction in workplaces - not just resistance, but something more existential - is there an opportunity to use experience data to actively manage the human side of AI transformation, rather than just deploying more AI into it? I’ve been putting this question to other key AI providers in the enterprise - how can their platforms help manage the change? - but none have had a good answer for me yet.

Neither had been asked it before. Anderson said:

The technology already supports it. It's something we really need to go and emphasise. You're on to something that we really need to go and emphasise.

Hammond called it "an intriguing lens" his teams hadn't explicitly explored. For the record, Anderson said I could have 1% of the revenue if Qualtrics decided to run with it (just getting that in black and white!). 

In all seriousness, every CIO in our network says the primary blocker to AI adoption is people, not technology. Most vendors respond by building better training programmes, or expecting adoption to follow deployment. None of them are using experience data to understand and shape the human response to transformation in real time. That's exactly what Qualtrics is built to do - and it's currently sitting unclaimed as a use case.

My take

I don’t want this piece to read as critical. I hope Qualtrics takes it as intended - as guidance and advice. I have a lot of faith in the vendor - its executives are thoughtful, the team is smart, and the technology is a different proposition to every system of record out there. I am a big proponent of centering experience in enterprise decision making and I hope that Qualtrics can offer a strong answer to these questions, as I think the agentic experience proposition is very compelling. 

These are hard questions about execution - getting a wide and varied install base to move with it, closing the integration gaps that stand between experience context and genuine operational impact, and earning a seat in conversations that have historically been owned by other platforms.

None of those are insurmountable. But they're the real test of the next phase of Qualtrics' story. The insight gap being closed is the starting point, as Maynard said. What comes after is considerably more complicated.