Openreach's Brutal Honesty About What it Really Took to Transform CX at National Scale with NiCE

Upgrading 25 million homes to full fiber broadband is a logistical feat in itself. Doing it while delivering a consistently good customer experience – across 700 communications providers, each with their own messaging – is another challenge entirely.

That was the reality facing Openreach before it deployed NiCE's proactive AI agents. Engineers were showing up to empty homes, customers were getting conflicting information depending on their provider, and the company's NPS sat below zero.

In this video interview, CX Today's Rhys Fisher sits down with Chris Herbert, Director of Customer Service at Openreach, Eifion Lloyd, Senior Manager of Customer Enablement and Strategy, and Giles Bryan, General Manager of Proactive AI at NiCE.

Together, they walk through how a proof of concept built on daily Excel uploads eventually scaled into a deployment spanning 15 million customer journeys – reducing missed appointments by a third, cutting inbound contact volumes by a similar margin, and pushing Openreach's Trustpilot score from 2.0 to 4.7 across 300,000 reviews.

The conversation covers what it took to get internal and external buy-in, how NiCE's platform handles landline-only customers without a separate configuration, and a generative AI result that didn't make the press release: a 22% reduction in install cancellations from an ask-me-anything chatbot trialed late last year.

On the lessons learned, Herbert is candid:

"We could have stopped when it was a bit hard. But we knew by listening to our customers that this was worth pushing through."

Check out the full interview to find out precisely how a CX vendor delivers an implementation of this scale. 

Transcript

SPEAKER_00 0:09

Hello and welcome to CX Today. I'm Reese Fisher, Associate Editor, and today I'm delighted to be joined by quite a few people actually. I've been I've been spoiled this afternoon. Uh we've got with us today Chris Herbert, the Director of Customer Service at OpenReach, his colleague Ivan Lloyd, the Senior Manager of Customer Enablement and Strategy at OpenREACH, and Giles Bryan, General Manager of Proactive AI at Nice. Thanks for joining me, guys. How are you all doing?

SPEAKER_03 0:38

Very good. Yeah, thanks for the invite. Tremendous.

SPEAKER_00 0:45

Yeah, I'm really looking forward to the chat today. You know, we're going to be talking about the the recent partnership between NICE and OpenReach, kind of with NICE's AI agents being used to help upgrade the UK broad brand deployment. And I hope, you know, maybe really digging into that process and kind of picking up on some of the key lessons from that. So I guess I wanted to start with you, Chris. Before NICE came in, what was the customer experience like during a broad brand upgrade? And when did it maybe become apparent that the way you were currently operating perhaps wasn't going to quite cut it?

SPEAKER_03 1:20

Yeah, so uh just firstly, OpenReach um builds and maintains the UK's uh broadband and voice network. Uh, we serve millions of homes and businesses, essentially the backbone of the UK, uh, power in homes, schools, businesses, hospitals. We are a wholesale provider to 700 communication providers. So I'm gonna call those CPs going forward, communication providers. And engineers work in end customers' homes, right? But those customers are technically the comms providers, customers, and historically, CPs owned their own customer communications, right? Notifications, uh, appointment confirmations, updates, not open reach. So, with over 700 partners, I think uh at the time customers were receiving inconsistent messaging, different formats, different tones, uh, depending on the on the provider. And I think more importantly, that inconsistency created uh an engagement gap that OpenReach couldn't really directly fix with the tools available at the time. So for me, operational impacts were really clear. Engineers sometimes arrived at empty homes due to perhaps incorrect appointment confirmation. Uh, customers frequently received conflicting information, and that really drove frustration. Uh, end customer sentiment was really hard to uh evidence. Um, the survey data didn't really reconcile easily with our own internal metrics, and effectively customer experience indicators were really poor, right? So trust pilot was 1.6 bad, our NPS scores were less than zero, uh, and competitive pressure was rising. So that you know, this made that change in approach really, really urgent.

SPEAKER_00 3:04

Yeah, yeah, absolutely. Thanks for that, Chris. Thanks for really nice kind of overview of where the lay of the land. I guess building on from what you said there, Giles, I wanted to come to you. Obviously, like Chris outlined, OpenReach is quite it's in an unusual position in a lot of ways. I think it's kind of this one infrastructure owner, but it coordinates customer engagement on behalf of your skies, your Vodafones, and you know, plenty of others. From a nice perspective, what made this deployment different perhaps from what you would normally see? And how did that shape what you you actually built in the end?

SPEAKER_01 3:35

Yeah, so uh good questions, Reese and good point. We're we're used to operating at scale, but OpenMeach is definitely one of those organizations that's at the top end, if you think about it, you know, there's 25 million homes passed, um, you know, tens of millions of customers. Uh, it's a very it's a proper enterprise scale business that the whole of the UK effectively relies on. Um so that's the first area of complexity. The second area is uh it's effectively a B2B to C model. Um so uh, you know, Sky and Vodafone and EE and TalkTalk all depend upon OpenReach to deliver these services. So, as well as having that massive enterprise scale, engaging with millions of customers a month, you've also got the complexity of the fact that uh essentially you're engaging on behalf of up to 700 CPs. And that means that the data integrations you need uh become necessarily very, very complex, very sophisticated, because not only have you got multiple data feeds from all these different organizations, you've also got all the different actions and system updates that need to flow back into those organizations. It's basically like doing 700 deployments in one. Um, plus, it's all at enterprise scale. Uh, and and to be clear, this isn't about you know saving some call center costs or a bit of automation. This is a major transformation right at the heart of the OpenReach, and that's the British Telecom, the BT group business. So that made it um, you know, it's ideal for our technology, but also something that had to be taken very seriously and how we uh as a team we rolled out.

SPEAKER_00 5:14

Yeah, yeah, absolutely. Yeah, when you really outlined the scale of it, it's it's it's almost shocking. It's uh it's an incredible feature, I'm sure. I guess I want to touch on maybe some of the some of the specifics of of the of the process. They want to come to you with this first uh avion. You know, quite often I think reactive customer service is is perhaps the default for organizations. You know, they wait for the call, then they deal with it. But you know, switching to a model where you're reaching out before customers even have a problem is is quite a big shift in that regard. How did that conversation kind of go internally and was there any pushback on that?

SPEAKER_02 5:54

Yeah, there was there was actually what's quite interesting is that the the pushback was quite similar for our external CP customers as well as internally. So shifting to proactive communication led by OpenReach on behalf of the CPs, we needed their buy-in and we had to work very hard, rightly so I must add, to earn their trust. And we had to work just as hard internally. So common objections that we heard included that's the CP's job to contact the end customer, or you know, we can't contact end customers, we don't have the data right. So our approach very much relied on building evidence to overcome resistance. Um, we we we started small, we innovated in increments, but we moved very quickly. We we started with a proof of concept, we scaled that to a proof of value, we then scaled use case by use case. Now the early delivery was very manual. Um we used Excel's, uh, there was a little bit of a lag in the sending of the messages, um, and occasionally things went wrong, occasionally the platform wasn't stable. But the goal of the proof of concept was to prove it, not for it to be an elegant um kind of way of working. But I think one of the key inflection points for us came when we launched our repair use case for customers who have a broadband or a voice fault. Um, this is a journey at scale. We do tens of thousands of engineer visits every week to fix faults in the network because of our scale. And what we found is appointment reminders plus simple self-help tools such as switch your router off and on, check that the cable is plugged in, check, you know, basic, basic, very, very basic troubleshooting checks, that actually resolved around 4% of faults even without sending an engineer. When you scale 4% up for business of our scale, it doesn't sound a lot, but that's a big number in the thousands per week. And that reduction in avoidable engineer visits freed up capacity to redeploy engineers hours into full fibre build. So getting customers onto the full fibre network that we've been rolling out. All of that evidence enabled confidence building internally and externally with RCPs, and that's what's helped the work we do with NICE go from strength to strength over the years. Um, it's also enabled us to get to get the investment to go ahead over the last five years. So I think in general, we've we've moved both internal and external audiences from scepticism to strong credibility, and we've actually now got teams within open reach coming to us with additional use cases that they want to use the NICE and customer contact platform for.

SPEAKER_03 8:25

No, I no, I think uh Avian has very elegantly answered the question. Uh there was there was a lot of pushback at the in the beginning, a lot of pushback, uh and through persistence, perseverance, and ultimately letting the proof of concept prove, yeah, to Avian's point, wasn't, you know, wasn't particularly elegant in the first few weeks, months, but it proved that the value for our customers and our business and our CPs was there. And from then it started to snowball right through to we've got quite a lot of demand internally now for different user cases that we are often talking to Jiles about how quickly the team can stand those user cases up.

SPEAKER_00 9:10

Fantastic. You know, that's what uh elegant aside, that's what it's all about, isn't it? It's delivering for the customer at the end of the day. That's great to hear. I guess sticking with kind of some of the specifics, another one, and I want to hear from from both sides here on this, really, but uh I'll start with you, Giles. You know, one of the things that I think jumped out to me in this deployment is is the landline only customers, which is I think a group that's perhaps forgotten about quite easily in this uh digital first era we're written. You know, um how do you actually build that kind of inclusivity into a system that when it's operating at this scale, do you think?

SPEAKER_01 9:46

This is a really interesting um difference about the way we approach this. Most organizations, when they're thinking of engaging with customers, they think about the communication channels. You know, they've got a an SMS group, they've got a voice calling organization, they've got a digital group, they've got an in-app group. And as soon as you do that, the inclusivity and the simplicity break down because you've basically got four separate specifications, four separate owners, uh, four separate configurations. Uh, the way that Proactive AI has been designed within NICE is that it's just about the customer journey. The customer journey is what you configure, and the channels of communication are a subset of that. So how you communicate that journey to any particular individual depends upon what that individual has available and what they feel comfortable with. For some, it'll be SMS or RCS or WhatsApp or voice calls. It just depends on what works. And that means that inclusivity is kind of built into the platform from the core. So actually, there's nothing specific we have to do to make it inclusive, and it's just about that's that's how the platform operates. Having said that, uh, as I'm sure Avian or Chris will speak about, there are some particular focuses that we had in the uh in the open reach rollout to make sure that some um some parts of the organization and customer base were dealt with appropriately.

SPEAKER_03 11:08

Yeah, Reese, so can I can I add to that? Is that okay? Um well let me let me switch share a few words, Avian, and then perhaps you you jump in as well, because you know I know you are very passionate about this space, uh, Avian. So I think it is a multi-channel approach, right? So whether it's SMS, whether it's email, uh automated voice, and we are now re-established in the sort of trialing the element of AI voice, which allows us to start to look at that landline only only base, right? And I think AI voice, it's I think it's seen as potentially um game-changing for that greater reach. Uh, and from what we are developing together, it's looking really immersive, it's sounding really uh reassuring and authentic. And I think it's going to reach those customers that we've been struggling to get to. However, I do want to be clear, you know, human support, I think, still does remain you know important for customers who might need it. Uh, and that of course includes those perhaps slightly more vulnerable uh segments or or journeys perhaps that are trending a little bit more negatively. So I look, I think a lot of our work handles you know 90%, I'd say, avian of routine demand. Now, what that does for us as a business is it frees up teams to focus on you know highest need cases where they can work their magic, where humans can still work their magic. So I think from an OpenReach perspective, I think you'll see us do this more and more, you know, harness the technology and the capability that we've worked really hard with Giles and the team on. Uh, but really the goal for us is to improve the customer experience, uh, which you'll see from us is going to be an even bigger focus area going forward. Uh, Avian, anything else you want to add to that?

SPEAKER_02 12:53

Um, you've explained the points well, Chris. I would just add that this means we we don't have to set up multiple contact centres to speak to customers. So the the proactive AI agents, as you said, Chris, are able to handle most of the routine demand. The levels of containment are very high, they handle the um stuff that is kind of low effort and relatively transactional. The the integration between channels is very good as well. So a customer can can start off on email and then they can respond on automated voice. And the journey integrates and stitches together all of the different responses and all of the different channels. Um, and for me, this is really exciting because we can reach every customer through a combination of channels, but then we can then prioritize our human agents on those customers who really need the support if there's a complex journey or if they're more vulnerable. And what that means is better customer experience alongside a more efficient, targeted operating model and no customers left behind there. So that's a real win-win for me. Yeah.

SPEAKER_00 13:51

Yeah, yeah, I can imagine. Thanks for that, guys. I think it's so often we hear about these kind of we hear about the overview of what these tools do. I always enjoy these opportunities to speak to people like yourselves who are kind of working at the cold face and can really provide these details about what they actually are doing for customers day in and day out. So yeah, really enjoyed that stuff. I guess Chris, I just wanted to come back to you on some of the stats from the upgrade, which I thought, you know, there's some really impressive stuff in there. I think a third fewer missed appointments, a third fewer inbound contacts, and your trust pilot score going up to 4.7. Were there any of these perhaps that surprised you? And is there a number that didn't maybe make the press release that you're almost quietly quite proud of as well?

SPEAKER_03 14:37

Right, so I think that you've you've touched on the numbers here that we're really proud of, right? So we've reduced efforts for our our customers, our CPs, and for our end customers. Uh and we've made that that more efficient. But ultimately, the end game is it's a greater customer experience. A trust by that score is something we are also very proud of. Uh, and that's across you know, hundreds of thousands of reviews, by the way. So it's not just a few, it's hundreds of thousands of reviews. Now, the business case uh has been reinforced through three, and Giles will know this well, uh, OpenReach Investment Board approvals to fund expansion. And I want to be really clear, as a very large uh multi-billion pound business that we are, getting through these approvals is no easy thing for clear reasons, right? Um, the program has exec level sponsorship and it delivers a positive return on investment, right? Uh, significant benefits that I think we've shared before. But for me, it's not just about the financials. That's good, of course it is, but the experience has been game-changing and I think at times industry leading. And you know what? We are going to continue to push the boundaries of this partnership, this program that continues to drive um brilliant customer experience that is essential uh for our for our CPs and customers. Uh, as we've sort of touched on, the benefits span, um, you know, avoided wasted uh uh visits from our engineers, uh, lower cancellations, improved retention and growth, and ultimately that ease of doing business. And if we could capture that, I think ease of doing business with us is is something that we are we are really focusing into because we know that we're not always easy to do business with, but we are absolutely improving that significantly. Now, clearly, you know, efficient efficiency is good, but what does it mean? It allows us to reinforce engineering effort into building the fiber of the future, into connecting customers onto the brilliant full fiber network and really you know driving that that forward. So obviously, trust pilot improvement, as I said, is a standout outcome and it's an external validation. You know, we can't we can't control those. I think it's over 300,000 reviews now we've got actually. Yeah, it reflects the impact of a proactive customer journey and the quality delivered by our engineers. So um, I I I think um, yeah, I I think A, anything else you'd add, actually, Avon, to to that before I waffle on too much?

SPEAKER_02 17:10

Yeah, I think um a point that we haven't we haven't really shared very much yet because it's all quite new. So we're we're starting to progress now from the more deterministic flows. So basically, if A happens, say say X, and we're moving out towards more generative AI flows using large large language models, and um we anticipate that the benefits are going to extend even further, right? We're we're looking to see what NICE can do in terms of analysing customer photos and customer videos. So, you know, ahead of a complex installation, can we get the customer to send us a video showing the surface type and the distance between part of the open reach equipment and their property? But we we A B tested some of this generative AI approach on a on an Ask Me Anything chatbot earlier before Christmas, and that reduced install cancellation by 22%. Now that is a very significant number, and that is a great intervention. And this has all been done with the AI having very strong guardrails, being very compliant, whilst also maintaining fluency in responding to customers and being really flexible. So, you know, I think that start, the 22%, that is really exciting, and I'm just very excited to see what happens when we scale the capability.

unknown 18:18

Yeah.

SPEAKER_00 18:20

Yeah, yeah. No, sounds like really exciting stuff. And I think yeah, that trust pilot school especially is the one that stuck out to me, like you said, that you you know, you can't fake that. That's really impressive stuff. And I appreciate hearing you talk about Chris being quite honest about perhaps some of the areas where you did need to improve a little bit, and that's you know, how necessary you do that. We don't uh we don't often hear that side from uh from companies, so yeah, I appreciate that. And I guess on that note, and again, this is an area I'd like to hear from kind of all of you on this, but maybe I'll start with you, Giles. You know, no rollout this size is going to go completely smoothly end-to-end, you know, it's impossible. What were perhaps some of the hardest moments, you know, be it technical or organizational, and is there anything that you would maybe do differently if you were starting this again?

SPEAKER_01 19:06

So um we're familiar to working at scale. So we've got you know huge, um huge organizations we work with globally, uh big telcos such as Verizon in the in the US. So we're used to the scale. What was particularly different about OpenReach was uh all the multiple CPs and um communication providers. And and uh I think probably we would have, if we could have done, we would have rather avoided having the spreadsheets and what we call the tactical uploads. Uh we'd have much rather gone to API-based, you know, proper data integration uploads from the start. And I think probably that's something I would uh I would insist on in the future, because that certainly made it very difficult because without those kind of structured data um data um uh layers, it's very hard to be sure the data you're getting is accurate. And that certainly made things a little challenging for us all, um to say the least. That's something I would uh I would probably not want to repeat. Um the team did an amazing job, to be clear, but that would be something I would uh I would insist on, perhaps.

SPEAKER_03 20:12

Yeah, Reese, am I okay to add to that? Um yeah, I I think that the the early tactical journeys, as you say, Charles, they were very, very reliant on daily Excel uploads and and had had limitations, uh, shall we put it uh that way? Uh and I think there were very difficult moments that that we would but we worked together as a as a partnership because we knew the outcomes that we wanted to achieve were really clear. Um, I think there's also a couple of other points. You know, we we have extensive governance um, you know, in place here. So, you know, how do we how do we push the boundaries of taking a little bit more risk? Because we have historically been fairly risk-averse, but in in a safe way, right? How do we make sure that we have the right uh guardrails in place, particularly now as we're starting to work move into the Gen AI piece? How do we get the right stakeholder sign-off? How do we have the right governance? You know, now that can slow progress, but I think it's the right thing to do. And I think we've learned and adjusted so that that progress is quicker, but still going through the right you know, sign-off. Um, and I think, you know, would we would absolutely repeat, I think, Giles, Avian, the same launch and sort of scale profile that we did, but I think we'd push towards that strategic solution much, much sooner to get us off of the Excel based processes and more onto the strategic, safer, more controlled processes than we did. I I think for me, and again, Avian, I don't know whether you want to add to this, I I think that was probably the core learning. We should have pushed towards strategic sooner because we knew it was working, we had that gut feel, we saw the results. We saw the sentiment from our customers, but we still pressed on with let's get more use cases, let's do more on Excel because we we wanted to see the customer experience benefit improve even further.

SPEAKER_02 22:11

Would you agree, Avian? I agree, yes. I think I think we uh pushed the tactical Excel-based data infrastructure to its absolute limits. Um I think we were we were possibly guilty in the business of thinking Excel and robotics will will do it or carry on going, it's good enough. And I think for customer experience, you have to think much, much bigger than that.

SPEAKER_00 22:35

Yeah. Yeah, and again, repeat what I said last time, but yeah, thank you very much for your honesty there, guys. I think these are always really, really useful conversations for people, and that's what this is all about. And I guess on that note, just the final question here, um, and again, you know, I want to hear from both sides, but I'll start with you, Giles. You there'll there'll be plenty of people watching this perhaps who are weighing up, maybe a similar move, you know, but perhaps they are put off by that sheer scale and the risk of trying to implement something like this. Based on what you've been through, what is the most important thing to consider, do you think, before they get started?

SPEAKER_01 23:11

So the thing that the Open Reach team understood right from the start is that by the time a customer calls in, the chance to deliver a truly great customer experience has already been missed. Right? So proactive is is literally transformation, it's a game changer. And once people understand that, and then they say, okay, we are going to become a proactive organization, uh, that's really the biggest hurdle you have to get through, is that you know, you have to get through all the call center metrics and people who call in, get from that to realize that any good business, if it's going to be successful in the future, needs to become a proactive business business that serves its customers. So that's the thing I would say is the most important leap people have to make.

SPEAKER_02 24:00

Uh I'll follow if that's okay, and then I'll ask Chris as well for his views. Um, for me, it's about investing in customer experience. So, as we discussed just a few moments ago, you can you can start small, test, learn, you can absolutely use a tactical data infrastructure that enables you to build expertise and then scale. But for me, the key the key lesson is actually we knew it was the right thing to do, we knew that the benefits would be there. So have have have confidence with customer experience. You you sometimes have to invest first and then the benefits will come. So my my takeaway is to be bold and continue to push forward the customer experience agenda. Chris.

SPEAKER_03 24:38

Yeah, I mean it's it's similar to to me. I think it's having that confidence, thinking big, having that vision, you know. Um we could have stopped when it was a bit hard. Yeah, we could have um could have gone on to something else, but actually we knew in our we knew by listening to our customers, both our CPs and our end customers, because we understand them deeply, that this was something that meant, you know, that was worth pushing through. And again, you know, having that brave ability to take a leap of faith, you know, work with someone like Giles and the nice team to build something small that proves there is big, deep benefit, and then holding on to that innovation that you have in your head and your mind and that vision of where you want to be, which is you know, the most effortless organization to do business with, is something that we want Oprah Reach to be. We are making great progress in that space. There's more to do, but guess what? We're gonna do it, right? So I think it's a mixture of what Giles and Avian have said, added on to what I just said there, Reese. Hopefully that makes sense.

SPEAKER_00 25:49

Yeah, yeah, absolutely. Um, and I think that's that's probably a great thing, a great place rather to wrap things up. Thanks so much, guys. I really I've genuinely really enjoyed that chat. Like I said throughout, I think these discussions where we can really dig deep into what these implementations actually look like in practice, unpick and unpack, and really the lessons you learn from that. I think those are the insights that are really, really going to be beneficial to our audience. So, yeah, thank you all for your time today.

SPEAKER_01 26:16

Well, thanks for yours too. Thank you. Thank you all. Bye, bye.

SPEAKER_00 26:20

Excellent. Um, I did just want to quickly also thank our viewers for tuning in. If you enjoyed this discussion, and I'm sure you did, uh, please remember to like and subscribe to the channel, and also remember to head on to cxtoday.com for more stories like this. Until next time. Thanks for watching.