Customer Experience design and User Experience design get used interchangeably in a lot of organisations, and the confusion over what separates them is what brought Debbie Levitt onto the Smashing Podcast. She is a CX and UX strategist, researcher, designer and trainer, author of Customers Know You Suck: Actionable CX Strategies To Better Understand, Attract and Retain Customers, and host of the Delta CX YouTube channel, where she has recorded more than 175 episodes and over 600 hours of video. Vitaly Friedman asked her to define the terms and explain why the field still struggles for standing inside companies.
Where CX and UX meet
Levitt treats CX as the end-to-end experience: everything a company offers or does, in any form, touches customers. Products, services, digital and physical channels all count. UX, by contrast, is often assumed to cover digital screens alone and the research that supports them.
Her own position is that when both are done properly, they are the same thing. Plenty of UX practitioners already want to look past the interface at the whole journey — how someone interacts with a bank branch, customer support or a hotel reception desk. She recalls contracting at Macy's, where her team pushed to get involved with the stores and were turned down with the response that they were "just the people who make the screens." The goal she describes is a holistic link between digital experiences (website, mobile web, app) and what happens in the store. Different names persist, but she is not prepared to spend the day arguing about them.
Speed over quality, and the leaders who enable it
Asked why, in 2023, CX and UX still have to justify their place, Levitt points at two causes. The first is a company-wide preference for asking what the least amount of work is that can pass as good enough — something she attributes in part to what she calls fake agile. The original Agile manifesto cared about customer satisfaction and good design, but agility as practised became about making engineering teams faster and more efficient: release faster, release more. Her response is that speed and volume mean little if nobody is attracted, satisfied or retained. Lean gets the same treatment: it is meant to surface defects, risk and waste early so that quality prevents rework later, not to make people move faster.
That habit, she argues, produces mediocrity. Everyone knows what it feels like to hate a badly built app, website, hotel or airline, yet the same people return to work and declare their own output probably good enough.
The second cause is weak UX leadership. She describes leaders who obsess over keeping stakeholders happy and over visual design, who claim UX can be agile and Lean too by slicing the work down to almost nothing — a one-day research effort, a survey, a few seconds of design — and who sign off on impossible timeframes. Others waste their credibility arguing that designers are artists rather than explaining how CX and UX work mitigates risk, saves time and money, raises satisfaction and encourages loyalty. In her view the profession has been selling the value of the job instead of the value of the outcome: the crappy project that had to be redone, the coupons handed out to apologise, the multi-front disaster. A few extra weeks, she says, would have avoided the expense, the marketing fallout, the customer-support load and the voice-of-the-customer problems.
She also rejects building from assumptions. Agile, Lean, Scrum and the Nielsen Norman group all oppose it, she notes, and yet she found herself debating someone on LinkedIn who defended assumption-based methodologies as valid. Her answer: valid and good are separated by a canyon.
Meanwhile, data leaves out the why
Enterprises are not short of measurement, Levitt acknowledges: analytics, surveys, A/B tests, NPS. What they lack is the why and the how. An NPS of negative 30 will trigger a brainstorming session in which everyone guesses at the cause and then guesses at a fix — and nobody is surprised when that fails.
What she wants instead is genuine customer intelligence, which she distinguishes from market research aimed at demographics and discounts. Finding people who fit the target profile and advertising harder serves adoption, but without a better product, service or experience there will be no satisfaction, retention or loyalty. Generative qualitative research, conducted early, is how a team can claim it built what customers actually need — a standard she puts alongside the "killer app" thinking that seems to have faded. Today, she says, the emphasis is on checking stakeholders' ideas off a list, and many designs still fall short of using consumer expectations as the benchmark.
[Bots] are not yet a thinking robot. It doesn't have my ability to be strategic.
Reconsidering the MVP
Levitt admits to spending years arguing against the MVP and against Lean Startup, often against commenters on her Medium articles. Her objection is context. Those ideas came from books written over a decade ago for startups — remember the late 2000s and 2010, when a persistent startup failure was promising technology, features and systems that never got built, sometimes known as vapourware. Releasing something early was a sensible answer to that. That advice suited two founders racing out a first version in 2010; it does not transfer to large companies with large projects, large customer bases and a real dependence on retention.
The MVP can be reframed rather than discarded. A solid, realistic UX prototype is perfectly able to serve as an MVP. In her book, in chapter 21, she interviews product manager Steve Johnson, who says Eric Ries admitted a mistake: he meant minimum viable prototype, not minimum viable product, and never intended a pseudo-beta to be something sold or paid for.
Her insistence is that engineering does not need to build something before user-centred design can judge direction. She is a heavy Axure user (not a sponsored endorsement) and wants prototypes highly realistic in usability flow — typing in a field should really put text in the field — though not in visual design. Bring that artefact to usability testing and other research, and the team learns early whether the real problem is being solved well, partially, or not at all. That testing cycle should repeat. She regrets that readers of Lean Startup latched onto a few Agile-sounding phrases and concluded they cannot know their direction without engineering spending sprints, weeks or months building, testing, merging, releasing, then waiting for support complaints, a flawed A/B test or vague feedback — after which the item goes to the backlog or the icebox.
Related to that, she draws a line on measurement. She is not interested in usability testing to demonstrate delight. Usability testing is for confirming that the correct problem was identified and that a good concept was executed well.
Business language and honest metrics
Her practical advice for getting a shift under way is to stop using the vocabulary of design practice in conversations with managers and start using the vocabulary of business. Executives do not care about delight or empathy, she says. They care about attracting customers, keeping them happy so they stay — attraction or adoption, satisfaction, and loyalty or retention. Risk, wasted time and waste money are the terms that land, which is why she raises Six Sigma's cost of poor quality, covered early in her book, around chapter two, with a list of the ways a company loses both because it failed to build something better for users.
That emphasis, she says, drove her to an MBA rather than a master's in UX, HCI or human factors. It signals to a company that she cares how the business runs and makes money, ethically and with customers and DEI and accessibility in mind for the long term, not via shortcuts or paid ratings. She offers herself as an example of well-earned loyalty: she has just renewed her subscription to Monday.com (again, not sponsored), a product she promotes unpaid and whose staff have no idea who she is. When she moved to Monday from a previous project-management system she had used for ten years, that vendor's CEO asked what could be done to keep her. Her answer was nothing — she would not stay if it were free or if they paid her, because Monday fits her tasks and needs and makes her more productive. Those are her standards.
Companies can also examine whether their metrics pull in the opposite direction from customers' interests. She points to an organisation with a KPI, or even a North Star metric, based on how many people apply for jobs, when job seekers demonstrably want to apply for as few as possible. Because firms are feature factories run by stakeholders, the demand to push the number up wins, whatever it takes. Traditional impact maps reinforce this by starting from what the business wants to make happen — more button clicks, which translates into more newsletters, more discounts, more content, more emails. Her chapter 18 dismantles impact maps and replaces them with a customer-focused version.
Vanity metrics come in for particular criticism. Levitt recounts signing into an online stock-investing account and passing through six pages before reaching the account itself. She maintains it could be done on a single screen: username and password, a text message confirming it is really you, then access. She is convinced one of those six separately loading pages — a giant screen announcing a code would be texted, followed by a continue button — proves somebody's target of raising page views was met. More page views improve nothing about the experience. The same scepticism applies to time on site. It may indicate that visitors like what they see, that they cannot find what they need, or that they are lost because search does not work; the number cannot tell you which. Metrics like the 4.6 App Store rating can also be used to avoid uncomfortable truths — in that case, one- and two-star reviews accumulating over recent months, complaints verified as legitimate. In the conversation she describes, the response was that overall the rating was 4.6 out of five, so what was the problem.
Changing course, she argues, requires more than one company-provided metric shift. Organisations need outside consultants or new specialists because the existing staff have grown used to going along with how things are done, avoiding waves and leaving the status quo unchallenged. New leaders must be willing to tell a successful company that it does many things right and still has room to improve — and the company has to be open to hearing it.
Adoption requires the top, and self-awareness
Asked directly whether customer-centricity can take hold without commitment from above, Levitt answers no. Plenty is available from the bottom: a person can raise the issue with their manager, who raises it with theirs. But actual change comes from the top. It is not necessarily the C-suite, who may be too distant from day-to-day matters — it is more likely directors, heads, VPs, the people who watch budgets, outcomes and KPIs and report upward. Those are the people to approach with the question of how adoption, satisfaction and retention are going at present. The usual answer is that things are not going well, and that nobody knows why. When a reason is offered — customers are disloyal, they are tire kickers, they are broke — she treats it as invented, likely true of only a fraction. Then comes the test: run the survey, and afterwards ask again what is known, what is still unknown, whether the survey supports action and whether the action is known. If the next step is another guessing workshop, nothing has been learned.
On this point Levitt makes an offer that is hard to refuse. She will not overcharge, and says she is under-priced. A six-week project falls in the five-figure range — sixty, eighty, whatever thousand dollars — or beyond a hundred thousand if she has to add people to move faster. Set against money currently lost through loyalty and through sales failing to retain customers, she frames the spend as a question: is $150,000 over six weeks worth replacing guesses and assumptions with knowledge? Preliminary findings from generative observational or interview research can be available within four to six weeks, not months. Behaviour, perspective and task matter more to her than demographics. She describes herself as fully task-oriented.
Lessons from the field: allies, detractors, compelling reasons
Among the patterns she has drawn from working across companies, two stand out for anyone attempting change. Change-management courses advise finding allies; Levitt adds that detractors must also be identified and managed. Toxic leaders may actively want the effort to fail, because a successful change would raise the question of why they never made it — an accidental spotlight. Some may sabotage the newcomer. This is another argument for using consultants: a detractor can undercut an external adviser and look foolish for it, and that person leaves in six months regardless. Employees who lead the same effort risk demotion or dismissal, outcomes she says she has observed. The honest phrasing of the working arrangement is that she flies in, fixes as much as the company will allow, and flies away.
Second, a company needs a compelling narrative for why it should change. Even organisations that say they want to be customer-centric and want more journey maps often do not really want change: they are profitable, growing, and comfortable by their own chosen measures. Levitt describes working last year with a $1 billion European company competing with Indeed in job postings and applications, the market leader in Germany. When the position is so strong, the hard question is where change is even possible. She looked for areas that were not going as well as they could, and for allies, and found a UX team without a voice — treated as order takers. Improving how user experience work and workers are elevated became the way in.
The search is always for the place where someone will let the change happen. Has the company lost significant money recently? Has it damaged customer trust so badly that support is over-used? Fewer unhappy people calling in saves money. Her exercise for surfacing these arguments, featured in chapter 18 and run on some Mondays during her YouTube show, is the Delta CX version of an impact map. It inverts the usual starting point: begin with the customer's problem, then work out the root causes, then the impact on the business — the money, time, resources and environmental damage accumulating because the customer problem and its root causes went unaddressed. Building that artefact creates a ready-made argument: making this better saves the company the waste, risk, lost time and lost money attached to the status quo. Such arguments can be attempted at any level, and a junior staffer with access to some of the data can try, but they land best with managers and leaders. The task is to find the small open door.
One measure with immediate value is the presence of at least a few customer-centric metrics alongside business-centric ones. When Levitt examines a company's KPIs, or even its North Star metric, she frequently finds them set against what customers want for themselves. That mismatch drives an endless spiral: stakeholder says grow this number, so add newsletters, discounts, content and emails. Her dry aside is a show of hands for anyone who wants more emails.
Toxic leadership and the empowerment that isn't there
When Levitt is asked what tends to block this work inside organisations, she names the toxic leader: the narcissist or malignant narcissist who insists everything be done their way, manufactures fake facts and stories, spreads fear among subordinates, and puts people on performance review or performance improvement plans for questioning the status quo. Every company she enters has such a person, she says, and everyone there already knows exactly who it is — the fact that this person stands out means other hiring has been fine. [BLOCK_0]
If the leaders have grown used to that behaviour and are unwilling to measure success differently, and if CX and UX have no voice, then engineering is left pressing for faster delivery while product acts as an order-taking puppet for the stakeholder. Levitt notes the industry's claims about empowered teams and says the reality is nowhere close. She recalls a company that failed to live up to its stated values and responded by rewriting the values themselves into something meaningless.
Speaking up against the status quo is, in her account, the main lever — and it carries risk. An unhealthy company may retaliate, and many people understandably decline to take that chance. Hence the external consultant, who can say what needs saying without having a horse in the race, and who is gone in six months. Employees who can no longer tell the truth are a sign, for her, of a culture problem deeper than any KPI.
There is no single magic bullet. Levitt points out that a bullet that effective would have made her book ten pages long: three pages introducing it, five explaining it, two thanking readers. Her books run long because they walk through the many scenarios practitioners encounter — as consultants especially — with things to try on each piece. Staring at the mountain of problems in an organisation invites despair; working smaller pieces, and showing that the effort was worth it, does not.
Where customer intelligence starts, and what the future asks
On the specific metric types Levitt favours, she points to an example from a broader conversation about employee-application goals and a job board: an email-recycling operation that puts a retention metric followed by a note reporting revenue is unchanged. The problem with email volumes alone, she argues, is that extra emails often substitute for usefulness. On journeys and definitions, service design joins CX, UX and LEAN as concepts floating in the same ocean.
Looking ahead
Asked for parting words to a much later era, Levitt hopes for more ethical and more customer-centric choices, and the recognition that without happy customers there is no business, no staff and no money. Extending the timeframe, she predicts the near-term fight — 2025 rather than 2043 — will be over what a machine-learning machine should be allowed to do and what should stay with a person. The question of which parts of Levitt's job a bot can take over is already live, and she observes that a bot can only rehash what is already out there. Ask one how to become more customer-centric and you get a list of plausible-sounding suggestions that offer no practical direction — care about your customers, build empathy, get brainstorming, be innovative. None of it is actionable, and her point is that artificial intelligence is not yet a thinking machine with strategic judgement.
New projects of her own to look for include a site at customercentricity.com, a domain she owns, and details of her books and workshops at DeltaCX.media. The book itself has a digital version available for as little as a dollar, an attempt to keep it accessible to readers in places or circumstances where the full price is not affordable.
