Engineering Seamless Journeys Across Physical and Digital Touchpoints

Priya Sinha at UXDX EMEA. Video: https://youtu.be/we16GpvIdvw

Readable transcript: edited from the recording's captions for readability (fillers and false starts removed, punctuation and section headings added). Wording is the speaker's own. Timestamps are positions in the video. Names marked [?] could not be verified against the audio.

An experience problem, not a connectivity problem

[00:00:08] Good morning, Berlin. Quick show of hands: how many of you have had to explain yourself to three different support teams to get one problem solved? Oh, there are some. Keep your hands up if you've ever chased three different support teams to figure out where your order was. It seems to be a good set of people. We must be around 100, 150 people in this room, and collectively surely we must think that in 2026 this should not be this hard.

[00:00:54] Exactly. And that right there is the problem. Not connectivity, not coverage, not even technology. This is an experience problem: an invisible operational drag that slowly destroys customer trust. Because customers today don't compare enterprise telco mobility to other telcos. They compare it to every seamless digital experience that they have: Amazon, Uber, Netflix, Apple maybe.

[00:01:33] And honestly, that's the opportunity, because connectivity is no longer the differentiator. Nobody wakes up excited because their mobile provider provides, well, mobile connectivity. The real battleground is the experience layer: how fast you respond, how invisible the complexity is, and how little effort the customer has to expend.

[00:02:02] A little about me. I've spent 23 years in product and technology, 10 of those years in telecoms, which means I've spent an incredible amount of time watching incredibly smart people trying to solve very, very expensive operational problems. If you work in enterprise systems, you know exactly what I'm talking about: those problems that people were trying to solve with spreadsheets.

[00:02:35] Over time, I became obsessed with one question. Why does enterprise mobility lifecycle management still feel harder than ordering food delivery? Think about it. You can track a pizza in real time. You know when it leaves the restaurant. You know when the driver's turning left. You know when the driver's emotionally unstable and probably speeding. But ask an enterprise customer about the onboarding of their 5,000 devices and nobody knows the answer. That gap between what a customer experiences as a consumer versus as an enterprise client became the thing I wanted to fix.

[00:03:22] I'm Priya Sinha. I head product engineering for managed mobility within Vodafone Business, and today I'm here to share what was broken, what we tried to fix, where AI and automation helped us, and the painful lessons nobody talks about or puts in glossy transformation decks. Because behind every AI-powered customer experience slide, there's usually a terrified and sometimes exhausted operations team and 40 spreadsheets fighting to survive.

What managed mobility really is

[00:03:59] So what is managed mobility, really? It's more than the sum of its parts. It's not just about devices, not just about SIMs and tariff plans, support and service management. It's not just about managed services. It's an operational choreography of everything. It's making sure that when an employee opens a device box on day one, everything works: configured, connected, secure and ready. No tickets, no escalations.

[00:04:36] And that experience doesn't end after onboarding. The lifecycle keeps going. There's the provisioning of devices and tariff plans, there's support and service management, there are the device upgrades, the shiny new objects, roaming, billing, and of course bill shocks, device replacements, contracts and renewals, and end-of-life recycling. Physical plus digital, automation plus the human touch, systems plus experience. That entire invisible mechanism behind the scenes, that's the product. And when customers don't notice it, that's actually when we've succeeded.

[00:05:24] Now let's really talk about the reality of telecoms. This is not a calm industry. It's a pressure cooker of sorts. Connectivity is a commodity. Margins are shrinking. ARPUs, average revenue per user, are declining. Competition is brutal. Customers are getting smarter. Procurement teams arrive armed like they are ready to negotiate a hostage release.

[00:05:53] Then we layer on GDPR, Brexit, data sovereignty, sustainability reporting, ESG, eSIM fragmentation and disruption, non-telcos selling connectivity cheaper and faster, cloud convergence, hyperscaler encroachment, and CIOs asking us to do more with less. That's basically executive language for, "Please perform miracles, but cheaper."

[00:06:29] The old telecom model was to sell connectivity at scale. That's over, because connectivity alone has almost zero emotional value. Nobody renews because your SIM card was inspirational. Customers renew because of the richness of the managed service offering that you provide: onboarding was painless, their order arrived on time, support was instant, pricing was transparent, and their teams didn't have to spend six months battling operational friction. The winners in telecom will not be the ones with the fanciest and loudest 5G campaigns. They'll be the ones who remove the most friction.

Where AI and automation fit

[00:07:16] So naturally everybody asks, "Where do AI and automation actually fit in all this?" Let me be very clear. This is not magic. This is not even strategy. This is not a substitute for broken operations. If your processes are chaos, AI will simply help you scale it faster. What it is good at is repetitive workflow handling, reducing delays, surfacing anomalies, simplifying decisions and making complexity feel invisible.

[00:07:51] And in enterprise mobility, complexity is everywhere. Thousands of devices, hundreds of thousands of end users with different tariff plans and roaming, multiple countries, multiple cost centers, supplier ecosystems, different regulations, disconnected systems. Humans just cannot operate that efficiently at scale anymore.

[00:08:18] So we looked at the lifecycle and asked, "What value can automation really create for the business?" The results: 50% reduction in onboarding time, 50% faster quotations during the sales phase, three times the inventory and operational visibility during the in-life phase, and 30% reduction in support call volumes during the in-life phase. Those metrics impact customer retention, operational cost, employee productivity and renewal revenue. That's the difference between digital transformation theater and actual business impact.

[00:09:06] One of the biggest mistakes companies are making is that they try to automate everything at once. That usually ends with exhausted teams, expensive platforms and one lonely dashboard that nobody logs into. So instead, we followed the friction. We asked, "Where do the customer or our internal teams face the most pain?" And the answer was surprisingly consistent.

[00:09:36] Four moments. Building and managing quotations in real time in response to the customer sales cycle. Onboarding the customer, always a challenge. In-life commercial complexity and its management. And finally, daily support on small matters. That's it. Fix those four moments well, and the entire relationship changes. Ignore them, and no amount of automation and AI branding can help you survive. Because customers don't experience your org chart. They experience delays. They experience confusion. They experience silence. And unfortunately, those experiences become your brand.

Quotations: start where friction is highest

[00:10:25] Let's start with quotations. This one was painful. We discovered sales teams were having to manually stitch together OEM pricing changes for devices and accessories, which happen quite regularly, managed service bundles, tariffs, device purchase options (CapEx, OpEx, device leasing options), and approvals across disconnected spreadsheets. At one point, I genuinely believed Excel was our core architecture.

[00:11:03] And here's the scary part. By the time the quotes were ready and presented to the customer, the pricing was already outdated. Which meant rework, escalations, delays, margin leakage and damaged trust. Here's where we learned our first major lesson: start where friction is highest, not where technology is coolest. Not where the AI demo looks impressive, but where customers or our salespeople are bleeding time, money and confidence. Sorry about that. Because every delayed quote creates emotional drag inside a buying cycle.

[00:11:55] So we built a unified digital pricing tool platform: a single source of truth, real-time synchronization, automated approvals, integrated product catalog management. But importantly, we didn't bypass our business support systems landscape. That's another hard-earned lesson. Integrate across your BSS landscape; do not bypass it. Because every time companies build shiny experiences disconnected from core systems, that's just future technical debt disguised as innovation. And eventually, somebody in operations has to manually reconcile everything at 2:00 a.m., usually while questioning their life choices.

[00:12:44] So instead of creating another silo, we connected the ecosystem end to end. Now here's the important part. This wasn't just an efficiency project. It became a competitive advantage, because when enterprise customers are evaluating suppliers, speed creates confidence. The results? Quote turnaround time reduced from hours to minutes, significantly higher pricing accuracy, faster conversion from quote to contract, and internally, for the first time, salespeople trusted the tool to do the job in minutes. Which felt like a spiritual breakthrough.

Onboarding: 75 steps and measuring customer outcomes

[00:13:28] Now the customer signs. Fantastic. This is where the onboarding experience begins, and that's where many enterprise experiences collapse. Because onboarding as an enterprise customer in telcos is where operational complexity goes to war. We thought onboarding involved maybe 20 steps. We were wrong. It was 75. 75 steps across multiple suppliers, different teams and stakeholders, from operations to supply chain to billing, and different platforms: BSS platforms, ERP systems, warehouse systems, the customer approval chains. Honestly, mapping it all felt less like process re-engineering and more like uncovering an ancient civilization.

[00:14:25] We had to ask the question: why 75 steps? The onboarding teams had to get everything checked, and then double-checked, and then triple-checked by the customer, the supplier, the operations team and the commercial team. This revelation was a humbling experience. It meant everyone was terrified of taking any action. This is actually the delicate nature of enterprise customer relationship management. And we realized that this is not an AI or automation problem or a performance problem. This is about behavior setting. This is about having to set expectations with the customer.

[00:15:11] The impact was that the user experience, both for the customer and for the internal teams, was terrible. Repeated forms, repeated data entry, no visibility, no ownership, and weeks and weeks of waiting. That became our second major lesson: measure customer outcomes, not just process closures and metrics. Internally, teams were celebrating closure of their own KPIs. Meanwhile, customers were still waiting and saying, "Great, but where's my order?" Customers do not feel your internal KPIs. They feel waiting, confusion, silence and repetition.

[00:15:54] So we adopted a core principle: never ask the customer for the same information twice. If data was captured at pre-sales or during contract signature, it flows forward automatically. That sounds obvious, and yet in enterprise it's revolutionary. We automated post-contract demand forms for onboarding, product SKU creation, supplier workflows, onboarding approvals, ERP systems, and live process progress dashboards.

[00:16:30] The impact? 50% reduction in onboarding effort, less manual effort, 30% faster onboarding, improved NPS, and dramatically better operational performance and transparency. And perhaps most importantly, a happy customer. Customer experience improved after signing the contract.

In-life complexity and signal-response workflows

[00:16:55] Here's something we rarely talk about in telcos: the hard part, which is after the contract is signed. Because telcos and enterprise mobility are constantly changing. OEM pricing changes. OEMs make policy decisions that suddenly impact device circulation across the world. eSIM-only devices suddenly appear. Customers expand geographies. New device categories appear. Old device categories disappear. Employees relocate. Devices reach end of life and compliance evolves.

[00:17:35] Every single one of these triggers a manual workflow, a manual intervention. Disconnected workflows, spreadsheet updates, tons of emails back and forth between the customer, the ops team and the supplier. And what does it cause? Operational delays. In other words, humans acting as middleware, which is expensive, fragile and impossible to scale.

[00:18:04] So we had to look into signal-response workflows. An OEM changes pricing, which happens quite regularly: systems synchronize at the click of a button. A customer expands into different geographies, contracts change, they want new countries covered: product catalogs and procurement approval workflows update. Devices hit end of life: the circular economy workflow begins. Secure data wipe, refurbishment assessment, regrading of a device, either to responsibly use it for a second shelf life, contributing to the circular economy of devices, or to responsibly recycle it.

[00:18:49] Enterprise customers increasingly ask, "What happens to these 20,000 devices after year three, after we've finished the contract and finished using them for three years? Where do they go?" That is where we have to be worried about sustainability. Sustainability has moved from being a marketing gimmick to a real, earnest C-level ESG compliance conversation.

Compliance built in, and scaling what works

[00:19:20] Speaking of compliance and security, customers expect us to offer them secure products, of course, from day one. We design compliance directly into the workflow of product development from day one, not afterward. Because another lesson we learned was: build compliance in, not on. Nobody wants a transformation project where security and compliance arrive later on as late dinner party guests. So GDPR controls, approvals, auditability, all embedded into the product dev cycle from the beginning, not bolted on later. Because not every automation has to be pronounced, but it has to be embedded.

[00:20:08] There's another lesson we learned: scale what works, kill what doesn't. Because not every automation even has to survive. Some workflows delivered massive value; some just didn't. That discipline matters, because organizations right now are flooding themselves with pilots. Pilots are good, but pilots are not progress. Operational improvement, that's progress. Did it reduce customer effort? Did it reduce operational cost? Did it reduce the onboarding time? Did it improve visibility? Each of these has a meaning in terms of the customer experience. If it does, scale it. If it doesn't, move on.

The AI assistant

[00:20:58] Now let's talk about the AI assistant. This is where customer expectations were drastically changing, because modern enterprise users expect a consumer-grade experience. They don't want long ticket chains, call transfers between multiple support teams, or somebody saying, "Please contact another department." Which, by the way, is one of the most emotionally damaging statements you can make in enterprise support.

[00:21:28] So we embedded a GenAI assistant directly in the portal: an assistant that is able to understand the customer's tariff plans, the devices, the catalog, the contract and the account context. It can converse with the valued customer, provide them with context, and where needed, escalate to the right support team, with full context and conversation history attached, no repeating information. Because nothing destroys trust faster than explaining the same problem five times. AI should not make experiences look robotic. It should make them look more human.

[00:22:11] We also introduced AI-generated multilingual portals, and that helped with the deeply international customer base. Honestly, people make better decisions in their native language. The outcome: reduced call volumes, higher NPS, faster resolution, onboarding and handling, and lower operational pressure on agents. Very importantly, the AI assistant did not replace the agent. It removed repetitive work so that the humans could focus on the complex cases.

[00:22:44] That distinction matters, because nobody wants AI handling emotionally sensitive enterprise escalations unsupervised, especially legal, or finance, or roaming bills. God forbid roaming bills and bill shocks.

The uncomfortable lessons

[00:23:08] Besides the lessons I spoke of earlier, I wanted to reiterate some key lessons here, the uncomfortable lessons. Lesson number one: you start with the data. Before we automate anything, the data has to be clean. We actually realized that a large part of our data, our records, was not correct or accurate across systems. Had we automated first, we would have just scaled a problem many times over. What do they say? Rubbish data in, rubbish data out. So we spent a considerable amount of time cleaning it, and nobody celebrated it, but I think it had the highest return on investment of the work that we did.

[00:23:56] Lesson number two: you cannot automate what you do not understand. We discovered 75 onboarding steps, remember? Nobody had fully mapped them. Some existed purely because, well, they'd been around for a long time. That is corporate language for: nobody remembers why they exist, but everybody is afraid to delete them.

[00:24:19] Lesson number three: keep the human in the critical decision-making loop. AI drafts, humans approve. AI identifies anomalies, humans investigate. AI accelerates workflows, humans own accountability. Enterprise customers actually prefer that, because trust is more important than automation.

[00:24:43] Another universal lesson: change management is harder than technology. The systems were not the biggest challenge. The biggest challenge was helping teams recognize the changes, understand them and accept that it would actually work. That requires transparency, communication, empowering teams, and patience. Lots and lots of patience.

What comes next

[00:25:13] So where do we go next? Today we're moving from reactive automation to predictive intelligence. Today's systems respond to events. Tomorrow, systems anticipate them. We're exploring predictive churn, AI-assisted price optimization, quotation management, proactive lifecycle management with autonomous email reminders for delayed return of devices, and eventually agentic workflows. Systems not just capable of answering, but taking auditable approvals before action: triggering orders, launching workflows, approvals, optimizing renewals.

[00:25:56] The technology is arriving fast, but here's the truth. Technology is no longer the bottleneck. Data maturity is. Process maturity is. Change management is. Organizational trust is. The companies that win won't necessarily have the fanciest models. They'll be the ones with the cleanest foundations and simplified workflows.

[00:26:23] So I leave you with this. The best enterprise mobility experience is one customers never have to think about. When the quote arrives instantly, onboarding feels effortless, devices simply work, support resolves things, there's seamless transition between teams, and renewals happen smoothly. That's when technology disappears. And when technology disappears, experience finally becomes visible.

[00:26:56] Steve Jobs said, "Design is not just what it looks like and feels like. Design is how it works." That idea applies just as much to enterprise mobility as to consumer apps. Because behind every elegant customer experience, there's usually a brutally complex operational machinery in the background. Our job is to make that complexity invisible: for productivity, for trust, for business impact, and ultimately for customer loyalty.

[00:27:31] Thank you, Berlin. I'm Priya. I'll take your questions now, and I promise you I will not ask 75 questions just to answer a question from you. Thank you.

Q&A

[00:27:45] Host: Thank you very much, Priya. This was amazing. What stuck with me was when you said never ask a customer for the same data twice, and how much complexity you had to handle to actually not do one thing twice or more. And also the great outcomes that you highlighted after that. Let's move on. First question: if automation is so advanced, why does customer service fail more often to route me correctly today? Or someone has a personal problem here. I don't know if you want to take this one.

[00:28:23] Priya: My whole story was about the fact that automation, though it exists, is about the pace of adoption of that automation, and understanding what process we're trying to automate, which can seem like an operational drag and be emotionally exhausting for the customer, of course. Yes, there's still a problem. Customers still face that challenge, and that's because in a lot of organizations, and I'll include mine as well, these things are delivered in pockets across the organization. It's not consistent, and the teams have not really understood the real problem statement and applied automation to it. So yes, it still exists in pockets, for sure.

[00:29:09] Host: Totally, especially in large-scale enterprise, where there's so much fragmentation across departments. Are we really at the stage where AI assistants as a first contact are actually leaving a positive feeling for customers?

[00:29:24] Priya: It's a good question. Telco is a highly regulated industry, and it's an enterprise contract, so we have very little wiggle room, so to speak, in terms of how we can use AI. We've got to be careful about all the different regulations in place, especially in the enterprise industry. So yes, AI assistants are helping customers resolve simple queries, in particular in the context of: why is usage so high? Why do I have a bill shock? Why am I paying hundreds of thousands of euros or pounds for five users who happened to travel on a work trip across five continents?

[00:30:10] These are the kinds of questions that AI assistants can absolutely help with, because they're all related to the back-end tariff plans, the roaming plans, the contracts that they've signed with us. So yes, but AI is still being used, may I use the words, cautiously optimistically. We're being very cautiously optimistic about where we can use AI in the next leg of transformation.

[00:30:34] Host: And how that can create trust or break the trust. You said customers don't buy your org chart. I love that, actually. What's your advice on how to get the org to buy into that, given inevitable silos and their likely OKRs or goals that are not aligned across the org chart?

[00:30:56] Priya: So the question is about how we actually influence large organizations to buy into this whole process of OKRs. I think at the end of the day, what really matters in business is revenue, customer experience, customer relationship and people productivity. These are pretty much the four key pillars for most large organizations. And I'll sound cheeky, but I'll say: show them the money. Show them the renewal revenue. Show them the customer experience or the customer relationship going up. Any large organization's executive sponsors would absolutely pick up on that.

[00:31:41] So for us to actually solve this problem, to get organizations to buy into OKRs, it's important to show them the value of that change. It's important to show them how renewal revenue is linked to that benefit, how customer experience and NPS scores are linked to that benefit. That's the first thing you would do to make organizations buy into it.

[00:32:07] Host: Thank you very much, Priya. We're actually running out of time. Thank you so much for your contribution today. It was a really strong reminder that customer experience is not only about what happens on the screen, but also about what happens behind the scenes in large-scale organizations. Thank you so much.

[00:32:22] Priya: Thank you.