​​In this final part of my three-episode series on accelerating sales and adoption in B2B analytics and AI products, I unpack a growing challenge in the age of generative AI: what to do when your product automates a major chunk of a user’s workflow only to reveal an entirely new problem right behind it.

Building on Part I and Part II, I look at how AI often collapses the “front half” of a process, pushing the more complex, value-heavy work directly to users. This raises critical questions about product scope, market readiness, competitive risks, and whether you should expand your solution to tackle these newly surfaced problems or stay focused and validate what buyers will actually pay for.

I also discuss why achieving customer delight—not mere satisfaction—is essential for earning trust, reducing churn, and creating the conditions where customers become engaged design partners. Finally, I highlight the common pitfalls of DIY product design and why intentional, validated UX work is so important, especially when AI is changing how work gets done faster than ever.

Highlights/ Skip to:

• Finishing the journey: staying focused, delighting users, and intentional UX (00:35)
• AI solves problems—and can create new ones for your customers—now what? (2:17)
• Do AI products have to solve your customers’ downstream “tomorrow” problems too before they’ll pay? (6:24)
• Questions that reveal whether buyers will pay for expanded scope (6:45)
• UX outcomes: moving customers from satisfied to delighted before tackling new problems  (8:11)
• How obtaining “delight” status in the customer’s mind creates trust, lock-in, and permission to build the next solution (9:54)
• Designing experiences with intention (not hope) as AI changes workflows (10:40)
• My “Ten Risks of DIY Product Design…” — why DIY UX often causes self-inflicted friction (11:46)

Links

• Listen to part I: Episode 182 and part two: Episode 183
• Read: “Ten Risks of DIY Product Design On Sales And Adoption Of B2B Data Products”
• Stop guessing what is blocking your own product’s adoption and sales:
  Schedule a Design-Eyes Assessment with me, and in 90 minutes, I'll diagnose whether you're facing a design problem, a product management gap, a positioning issue, or something else entirely. You'll walk away knowing exactly what's standing between your product and the traction you need—so you don't waste time and money on product design "improvements" that won't move your critical KPIs.

Transcript:

Brian:  You are now experiencing data with Brian O'Neill. Experiencing data explores how product managers, analytics leaders, data scientists, and executives are looking at design and user experience as a way to make their custom enterprise data products and analytics applications more useful, usable, and valuable.

And now here's your host, the founder, and principle of designing for analytics. Brian O'Neill.

Welcome back to Experiencing Data. This is Brian T. O’Neill. This is part three of a longer episode I’ve been doing, a three-part episode, about accelerating sales and adoption in B2B analytics and AI products by focusing on designing for users' workflows and tasks. So, part one, we got into the introduction of this topic.

Part two, in the most recent episode that we did—or that I did, I should say; this is a company of one after all—I talked about what exactly is a user experience outcome, and really got into how do you actually set these up and how do you think about transforming your customer along a scale from frustration to satisfaction to delight, which borrows on work from Jared Spool’s framework about user experience outcomes, and how I really think that’s critical if you want your product to feel intuitive and be so obvious to buyers about why they should buy it.

We also started to go into the topic of AI products and how sometimes not sometimes—AI is definitely changing how work gets done here, so this idea of disrupting the market with AI-based solutions, and we talked a little bit about the incentives around this and how, in my opinion, you don’t want to disrupt the market. Maybe the VCs want to hear that you’re doing that, but that’s not what I think customers want to hear. Nobody wants to have things disrupted; they just want an improvement in their life. They want to get a promotion, they want to look better to their boss, all the classic B2B buyer types of things. They want to look smarter. Disruption is not necessarily a benefit to them.

So, one of the things I said I would get into though on this AI topic is what to do when your B2B product that uses AI heavily, particularly these language models, but not necessarily only those, what do you do when it solves one thing but immediately creates a new customer problem? So again, as I’ve done this on the previous episodes, as a good data person, I’m going to qualify my use of the word AI here. So, in this episode, I’m primarily talking about generative AI here, and more specifically, language model-driven interfaces and products that are leveraging structured and unstructured data. So, many B2B companies are obviously experimenting with this, so I’m just adopting the language. I’m trying to reflect the language of what the market and the way people talk about this, even though I know, for a lot of data experts, that’s a giant bucket term that has almost no clear definition, it means all these different possible technologies. But anyhow, reality is, that’s how it’s being seen, so I’m just letting the reality drive my own language here a little bit.

The promise of AI is big and sometimes significant disruption is possible when a machine can do tasks or work at a scale that no human has been able to do before. So, let’s look at an example here, like using AI to automate away the busy work and some workflow that your target customer has. So, let’s assume your product’s AI capabilities mean that you can erase, like, 50% of the job and the work that your target buyer does. I’m going to say your buyer here, thinking about the company. You know, the actual users may be downstream from the fiscal buyer.

But let’s just say, you know, half the work that they do right now can be eliminated. So, all this busy work they do, 20 hours a week, it’s literally going to go away in an instant. You know, not an instant, but within minutes it’ll get done. That’s, like, the promise of your product, right? Well, the promise of this type of disruptive capability is that workers will now spend more time on higher-value work, so work will begin to change. Like, this is what we often hear.

And so, if you’re analytical, you might be thinking, well, what’s going to happen is that we’re just—if the first 20 hour of the weeks, you know, takes—let’s say the problem is from step A to step G and if we can get rid of 20 hours a week, going from the A, B, and C part, that’s going to take minutes now instead of half of the week, and the rest of the week will be focused on D, E, F, G, right? So, you’re just going to end up bringing problem D sooner into their face on a more regular basis, they’re going to be seeing that problem a lot more because A, B and C are kind of gone now. So, let’s assume, like in this hypothetical job, like, half the job is analyze some cases, some support tickets or, you know, customer service requests or something, and decide what order to solve them in, and then the other half of the user’s job is to actually go and solve the cases that were prioritized as being important. So, let’s assume your magic AI-powered B2B product can do 95% of the first part of the job. So, you’ve basically eliminated half of the user’s job here and theoretically, you can compute a fiscal value of your product to your customer. And you know, MBAs will probably love you for thinking this way, right?

A product person might also be thinking the second half of the person solve the use cases, well, that second half is now in their face, as I was saying earlier. That’s now the new first half for the customer. The second half of the job, the D, E, F, G part of the workflow, this is now the first half of the problem. So, the AI has effectively removed the ‘prioritize the tickets,’ part of the work from the plate of the user.

So, should your product be solving the second half as well, the ‘solve the support tickets and cases’ part or should it just be doing the prioritization part and getting the things ready part? The answer is maybe [laugh]. Famous consultant—insultant—speak, right? ‘It depends.’ Man, we love that kind of answer. Look, the way to answer this is probably to understand the market and what the market is willing to pay for, so I would ask some questions.

Is your solution for the first half easily replicated by a competitor? Like, is that space just going to get flooded really fast? Is the market interested in adopting that much change all at once? In other words, solving both parts of this problem, like the A, B, C, D and the E, F, G—or the A, B, C and the D, E, F, G part, or are they really just, right, now, willing to spend money on the A, B, and C part? Or are they even willing to spend any money, let alone do they want to only spend it on that part?

When we make the scope of the product even wider, what risk are we taking on? Do we need a lot of new engineering and data pipelines or domain problem understanding, all this kind of stuff? And if we do all that and we keep broadening the scope of what our AI-powered solution can do, what does this mean for our sales and marketing teams in terms of how they’re going to go to market and sell this? Because now, especially with AI products, a lot of them can get why—they’re supposed to be flexible and have context and the use cases tend to be less procedural and have a wider gamut of possibility about what they can do. So, I think you need to have a clear long-term strategy, but you need to execute it with small validated product bets.

Have a clear long-term strategy with a clear North Star, but execute it with small validated product bets. So, push for moving your user experience from basic satisfaction to delight early, if you’ve nailed the frustration part, before you broaden out and solve the next problem. There’s always going to be another downstream problem here, but I would push for get from that basic satisfaction thing we talked about in the last episode to delight before you broaden out and solve the next problem. If you’re getting lock-in, loyal customers early, like, locked in, you’re now solving for your current and sort of your future needs as well.

So, why do I think this? Because when customers can trust that your product can deliver and they are delighted with it, they may also be willing to start adopting or trialing new things that you want to do in your product, including these, like, “Leapfrog,” quote-quote features. And by leapfrog, I mean adopting completely new ways of solving the customer’s problem that they might not have asked for and haven’t seen before, but they’re willing to participate in designing it with you as design partners. And great companies are constantly talking to their top customers and they’re working together. They’re willing to try because your product is trustworthy, it’s delivering obvious value, and remember, they’re not just satisfied, they’re delighted at what they have now, and so they’re getting a sense that their voice is being considered here, and that you’re being careful with their problem and about the kinds of solutions that you’re going to build.

When we reach delight, we have users that will see our products as indispensable. Indispensable solutions that they or the team can’t live without because they would dread going back to the old way. It’s too hard, it’s too laborious, the switching cost is too much, it’s boring, it’s risky, it’s dangerous to their career, whatever the reasons are. And when you have that lock-in, of course, where they’re delightfully locked in, this leads to reduced churn, of course, and greater loyalty, and hopefully word-of-mouth praise. But it also allows you to buy that trust to work on the next generation solution that you think is really part of your North Star strategy, they just can’t see it yet. But you’ve got their trust.

A couple of final thoughts here, or just really one final thought here, I wanted to say about something to watch out for, as you’re DIYing your design around customer workflows. First of all, I’m a big believer that there’s a lot we can do on our own, particularly with AI tools and ramping up our knowledge about domains that are new to us. I think there’s a lot you can do without a staff designer, without a product designer and a product strategist to work with you if that’s not your cup of tea in what you do, but there are places things can go wrong, even if your intentions are good. And while no product leader or founder probably hopes to or intentionally designs their solution to work against the customer’s wishes or needs, it can happen accidentally, particularly if we’re not actively designing the experience, with intention, and we’re not validating our bets. We talked about bets in the previous episode, so again, make sure you listen to episode one and two first—I mean, even though we’re at the end of three now. It will make more sense if I’m using language you’re not familiar with.

But I’m going to link up an article I have called “Ten Risks of DIY Product Design On Sales and Adoption of B2B Data Products”. This article that, again, we’ll link up in the show notes, will explain some of the common impediments that I see when teams do DIY their own product’s user experience design, so being aware of these risks is the first step to avoiding them entirely because, as with any project or any technology, if you don’t even know what you don’t know, you have no idea the kinds of harm or risks that might be involved with that kind of project or product work that you’re doing. So, that’s really what I want you to get out of that is to just be aware of what some of the DIY risks are so that hopefully you can address those on your team or with your vendors or staff or whoever is doing your work, you can be aware of what those are and navigate around them.

I hope this three part episode on accelerating sales and B2B analytics and AI products by focusing on users, workflows and user experience outcomes, I’m hoping this was really helpful to you. If you’re not sure if you need all of this, and you have a sense that maybe something is off with your user experience or your product strategy and you’re not getting the sales that you want and the adoption that you want, but you can’t quite put a precise finger on what the cause is, I do have a solution for this. It’s my 90-Minute “Design Eyes” Strategy Assessment, and in a single call, I’ll diagnose with you one-on-one, whether you’re actually facing a product design problem or maybe a product management gap, a strategy gap, or maybe you just have a positioning and messaging issue, where the way you’re talking about it and presenting it isn’t coming across the way customers might be seeing it. The point is, you’ll walk away knowing exactly what’s standing between your product and the traction that you need. If you’re interested in my “Design Eyes” Assessment, you can book one over at designingforanalytics.com/assessment. And that’s about it. Until next time, thanks for checking out Experiencing Data with Brian T. O’Neill. Cheers.